War is the one thing we humans don't leave home without. If anything, we've shown our strengths lie in accelerating the cycle between technology discovery and application in the weaponry domain. Weapons of wonder are being developed now, capitalizing on microwave, electromagnetic radiation, and acoustic technologies. They've been used in Somalia, and possibly at the siege at Mt. Carmel. This weaponry can assault the mind and affect human behavior as never before seen. Much as been written about these weapons, these technologies, and their potential. Much has been written about the intelligence communities' interest in remote viewing, psychic abilities, etc. Has all of this changed the very nature of war? Have our technical abilities outgrown our humanity? What will the next war look like?
4. jonesatlaw - 6/6/2000 5:13:31 PM
I am sure that we will see new hardware in the next war. I am also sure that the next war will be unlike some previous wars and like others. Desert Storm was WWII North Africa writ large. Vietnam was unlike any other experience for the US. War in the short term looks like it will have more to do with "forgotten wars" of the 19th and 20th centuries than most of the history we are familiar with.
5. AytchMan - 6/6/2000 5:30:35 PM
jonesatlaw--
I'm not sure it's relevant to the discussion because there are other examples that will support your point. However: except for the presence of tanks, North Africa was almost nothing like Desert Storm. From the theater geography on down to the actual terrain, communications, electronic warfare, tactics, supply, numbers of troops involved, length of the campaign and so forth. All different. Respectfully submitted.
6. AytchMan - 6/6/2000 5:36:21 PM
Slight backtrack. Since you said "writ large", a couple of my objections don't apply. Nevertheless, I think a far better example would be the Six-Day War.
7. AytchMan - 6/6/2000 7:16:14 PM
>>Has all of this changed the very nature of war?
Depends on what you mean. Technology has always affected tactics and the cycle will continue. But the fundamental objective of war will remain the destruction of the enemy's will and capacity to fight.
>>Have our technical abilities outgrown our humanity?
This question makes my head hurt. But just to move the ball forward a yard or so, let's assume it means "Will some weapons be too immoral to use?" To which I reply, "Not at least once". But I think it's just barely, just remotely possible that some will not be used twice in anger (I consider the WW2 A-bombs to be a single event since they were three days apart and had never been used before). But every weapons technology in history has whizzed past some foreigner's head somewhere. Why break a streak?
Before I mangle the third question, does anybody here think that war will become obsolete and disappear anytime soon? Say in the next 50 years?
8. CalGal - 6/6/2000 7:22:49 PM
I think it almost has disappeared as a viable option for advanced countries. At least war in the sense that we used to think of it.
9. Cellar Door - 6/6/2000 7:29:54 PM
We don't need war anymore. Now we have polls.
10. SheRex - 6/6/2000 7:35:22 PM
re - war as a viable option, war in the sense we used to think about it.
That's because the media have become willing accomplices to the propoganda machines.
The US invasion into Panama was covered as a heroic event for the US. The film crew (who won an academy award - Best Feature Documentary, 1992) recorded the U.S. General who commanded the invasion saying something quite the opposite.
From - The Panama Deception review
"The film challenges the 'official' rationale for the war and uncovers striking visual evidence of tremendous devastation and gross human rights violations not reported by the mainstream media. Eyewitness accounts and never-before-seen footage contradict Pentagon and media reports that deaths and damage were minimal. THE PANAMA DECEPTION raises fundamental questions about our government's control and manipulaton of the media, as well as the media's complicity. A range of human rights and legal experts provide commentary, along with invasion commander General Maxwell Thurman."
11. AytchMan - 6/6/2000 7:36:35 PM
CG--
Would you attribute that to nuclear weapons or the current geopolitical environment? Or something else?
12. SheRex - 6/6/2000 7:40:44 PM
More on Panama Deception
Washington Post review
13. CalGal - 6/6/2000 8:02:26 PM
Aytch,
A combination of nuclear weaponry and Vietnam Syndrome, I think. I also think that She and Cellar are right about the media influence--but then, I think that they are responding to the public demands, not driving the tone themselves.
But in other words, we can't fight toe to toe with any genuine competitor, due to nuclear fears. We won't fight without nuclear weapons with any lesser competitor, and the public will not tolerate a ground war and risk losing loved ones. So as long as we can go in and bomb lesser countries that don't pay us heed, we'll be able to pretend we have war. But it's certainly not what it used to be.
14. AytchMan - 6/6/2000 8:41:26 PM
SheRex--
>>re - war as a viable option, war in the sense we used to think about it.
>>That's because the media have become willing accomplices to the propoganda machines.
I don't understand. CG argued that war has practically disappeared as an option for advanced countries. But if the media have become willing accomplices, that would make it more acceptable and likely, not less. Yes/no?
Or, if you mean the "how we think about it" part, again, willing media accomplices would presumably make it more acceptable. And yet we seem willing to accept ever fewer casualties. Every operation today must be practically free of casualties.
What am I missing?
15. AytchMan - 6/6/2000 9:15:09 PM
SheRex--
Different subject. Take a look at post 722 in the WW2 thread plus the 20 or so posts on either side related to the morality of using nuclear weapons. Is this an appropriate/interesting subject for discussion here? As it relates to the future?
16. iiibbb - 6/6/2000 10:50:26 PM
At first glance, this post may seem to be a little facetious, however I think it's has a sound hypothetical basis.
I'm a fan of Star Trek. This is a popular issue for Star Trek, since they love to comment on the morals of the future... or more to the point, the more things change, the more they stay the same.
One particular episode from the Next Generation that's always gotten my goat was the episode where the Enterprise was to go on a secret mission to retrieve a secretly developed cloaking device. The conflict came from the fact that Piccard was in the dark, while Commander Riker had served on the ship that was testing the device and couldn't tell the Captain anything.
Anyhow, it turns out that this cloaking device was a violation of the Romulan treaty which pretty much said that the Federation agreed to never develop a cloaking device if the Romulans never agreed to cross the neutral zone (I hope that the _brilliant_ federation diplomat who dreamt this deal up was hung for treason).
An interesting side effect of the federation's 'cloaking' was that you phased out of real space... so not only couldn't you be seen, but you actually weren't there. So weapons were ineffective as well.
Anyhow... if I'd been the federation, I'd have called it shields, and the fact that you couldn't see me would be secondary to the fact you couldn't shoot me.
17. iiibbb - 6/6/2000 10:50:40 PM
But I digress...
...my main point is why would the federation (or any nation) agree to not pursue a technology simply to avoid war for war's sake. War will sometimes find you whether you want it or not. The Romulans aren't the _only_ threat in the universe. A device like that would have been very helpful in fighting the Borg, as well as the Foundation ships that basically kick federation ass.
A sane person does not seek war, but they don't fail to prepare for it either.
I'm not saying there aren't some pretty unethical weapons out there...such as germ warfare... however, you still have to research the technology...if only to develop ways to combat against such technology if they're used against you.
We should know from WW2, how technologically unprepared we were compared to the Germans and Japanese. Americans wanted to avoid war, so we didn't prepare, and when we were dragged into it we suffered initially. We're not always going to be able to rely on opening a can of "whup-ass" whenever a war comes a callin'. From a technological standpoint, we were lucky many times over in WW2 that our lack of diligence early on didn't cost us in the end.
18. iiibbb - 6/6/2000 11:12:21 PM
AytchMan
Message # 753 in thread 63
I certainly don't necessarily advocate the use of technology at the earliest opportunity, but in cases like WW2 where the issues are relatively black and white, but you can't really expect us second-guess the future. Once the genie is out of the bottle… you can't go back; this is as true for nuclear weapons as it is for other technologies. No matter how horrible the weapon, someone out there would make it if they had the means and thought it would give them the upper hand.
My previous two posts here already addresses some of my reasoning on why technologies still must be explored. The scary part is when we have anyone who believes that a technology would justify initiating a conflict... imagine if the US were to try something preemptive against China, seeing as most people concede that they will be a super power within a few decades.
Now, perhaps a little bit of context. I am of the basic belief that the US basically will do the right thing when it comes to technologies like this. I may be wrong, but that's my belief. Countries like China kind of worry me.
19. iiibbb - 6/6/2000 11:21:10 PM
... I don't know, you can hum-and-haw about the ethics of the development and use of future technology... but it's really hard to judge. There is always the context of a given conflict, and there is also an element of luck involved
Unless you hold the view we can exist without war, then you can't really prejudge a technology unless it poses some real threat for mass destruction.
20. AytchMan - 6/6/2000 11:24:40 PM
iiibbb 16-17 --
>>A sane person does not seek war, but they don't fail to prepare for it either
Your post makes a sound point although I'd substitute something like "sane and honorable". Throughout history, many sane leaders have actively sought war. Though ugly and regrettable, war does not equate to insanity as some would argue.
The implications of your point reverberate widely. Most notably, perhaps, in the idea that technological arms races are, to some extent, unavoidable and even worthwhile (!). For example, at the height of the Cold War, a reduction in nuclear stockpiles from 18,000 apiece to maybe 500 or 1,000 would have been noble and admirable but a reduction to zero suicidal. Because if the other side managed to secretly stash just an extra 10 or 20, well, one can take it from there...
And, to stay squarely on topic, this situation may recur in the future.
21. AytchMan - 6/6/2000 11:34:11 PM
jonesatlaw--
My msg 5 is poorly worded. I mean my comments may not be relevant, not yours.
22. iiibbb - 6/6/2000 11:48:58 PM
As I think about my views however, totally reflect the view of a super power... we're king of the hill, and someone out there always wants to be where we are (China). Or some view themselves as regional powers (Iraq).
However, what to do about lesser countries who engage in future arms races? India and Pakistan for instance... it seems to me they have more to lose in an arms race than to gain. It seems to me countries like that are better off avoiding war for war's sake...
It's too bad the US can't seem to get it's act together when getting involved in these grey conflicts. We do great when the issues are black and white... but our track record is relatively poor when we try to throw our weight around in places like Kosovo.
23. AytchMan - 6/7/2000 12:17:21 AM
>>India and Pakistan for instance... it seems to me they have more to lose in an arms race than to gain. It seems to me countries like that are better off avoiding war for war's sake...
I think that's arguable. Certainly an arms race is expensive and deflects them from pursuing other goals. But if one believes there's a non-trivial chance that the other guy just might drop one, then the survival of one's country is a pretty big gain.
As cynical as I am, I personally find it hard to believe that any country is, at present, anywhere near prepared to launch a first strike nuclear attack. I must admit, however, that the Indians and the Paki's worry me.
24. CalGal - 6/7/2000 12:24:46 AM
And then it's just the heighth of arrogance for us to tell other countries that they'd be better off not engaging in arms races. I mean, what do we say? Really, people, it's best left to your betters to concern themselves with such activities. You're not up to them.
25. AytchMan - 6/7/2000 12:50:40 AM
Addressing the thread header:
>>What will the next war look like?
I think it depends on the type of war being fought. Let's narrow it down a bit. For the foreseeable future (the next 30 years or so), I think there'll be three types of conflicts.
First, there'll be lots of Class D minor-league "police actions" like Somalia and Grenada.
Second, there'll be the standard field-issue "brushfires" such as the Kosovo air war and the Tournament of Chechnyans (sp?). While the American people won't like either of the first two types very much (with rare exceptions), the conflicts will prove very valuable to the US military (or whichever big power is driving the Zamboni). Both from the standpoint of keeping the machine lean and mean as well as testing and perfecting the technology this thread seeks to examine.
The third kind of conflict will be very rare, perhaps non-existent. This is Serious War and would include two or more major combatants. These will range from Falklands-type quarterfinals (Brits and Argentina) up to and beyond the upcoming India-Pakistan All-Asia Invitational. Major mobilizations, serious casualties, big stakes. Ugly.
Perhaps these somewhat arbitrary definitions can help focus the discussion. I've suggested them because I think there'll be significant differences in the way technology is applied.
26. iiibbb - 6/7/2000 9:22:35 AM
As a super power, our pressure to develop weapons is artificially high because we want/have to stay on top. Super-powers are not allowed to display weaknesses. That's why everyone is freaking out about our vunerability to terrorist or rogue nation attacks. No one thinks they can beat us, so they resign themselves to embarissing us. No one really knows all the ways to take a super power down, since USSR has been our only example...
CG - I don't think that our problem is that we stick our noses where they don't belong. There is some truth to the notion that, as a super power, one of our jobs is to provide certain protections to those nations who can't. Our problem is that we are completely inconsistant on when and where we apply pressure, and how much force we use to apply that pressure. Our credibility in those matters is shot as a result.
I think our adopted role of going after agressor nations is fine, but you are right, most of the time we should just let these countries figure it out for themselves, because we can't be everywhere. Kosovo is the ultimate example of a no-win situation for us. Whichever ethnic group is in control, attempts to purge the other. Without us taking over completely (which is not an option afforded us) there is no way for us to prevent what is happening.
27. angel-five - 6/7/2000 10:48:53 AM
Forgive me, but I think it's kind of remarkably short sighted to assume that the recent trend in warfare is anything but a recent trend, and it's slightly fatuous to brush warfare off as some kind of social pressure valve. War might seem like just another variety of bread and circuses when viewed through the jaded eyes of an American sophisticate, and there's no denying the effect a 'good' war will have on a population. Warfare is nevertheless generally conducted for other reasons.
History is replete with episodes where technological or tactical innovation dramatically changed the nature of warfare and the politics surrounding it. The phalanx, then the legion. Heavy armored cavalry, then the advent of the longbow. Fortifications, and then siege engines, then gunpowder and even more massive fortifications. Fast frigates. Troop mobility. Automatic weapons. Ironclad ships. Explosives and poison gas. Trench warfare followed by blitzkrieg and airmobility. The point is that at the time of each of these innovations, they were thought to have changed warfare immutably, and in some cases even served to end warfare altogether.
Today we laugh at the notion that heavily armored knights will crush any opposition before them, and we smirk at the idea that the advent of rapid-fire weapons or poison gas has made warfare too horrifying to contemplate. After all, we came up with ways to deal with knights and ways to limit the effectiveness of machine-guns and mustard gas. Yet since nuclear weapons happen to be the current boogeyman of large-scale warfare, and we haven't yet come up with an adequate military defense against them, we see people falling into the trap of assuming that nuclear arms (or chemical and biological weapons) have irrevocably changed the way in which wars must be planned and waged. A quick review of history shows the fallacy in such thoughts.
28. angel-five - 6/7/2000 10:54:38 AM
Our problem is that we are completely inconsistant
on when and where we apply pressure, and how much force we use
to apply that pressure.
Not at all, unless you frame your notions of warfare around such ideas as 'The United States goes to war around the world to prevent injustice and save those who cannot help themselves'. This is a rather myopic Western view, fed by the press. It's a mistake to assume that a majority of people in the world believe it, so the point on credibility is, well, pointless.
I may be opening a rather large can of worms here, but I'll ask anyway. Whatever are you talking about when you refer to psychics in the military and so on in your first posts?
29. jonesatlaw - 6/7/2000 11:35:57 AM
I think that the trend in warfare from the US perspective changed with WWII. The US picked up a great deal of technique concerning deception, intelligence gathering, and the "black arts" of war from the British.
Vietnam was a failure in this regard because the US was outdone by the North Vietnamese. There were pockets of success with alternative styles of warfare designed to win over Vietnamese support, but all in all we still thought too much in the way of large unit engagements.
With the increasing use of superpower forces for intervention into local wars, there should be a shift to smaller unit based tactics and weapons, along with logistical support.
30. angel-five - 6/7/2000 12:13:34 PM
The last part of 28 should have been posed to Sherex.
Jones:
I don't think there's any one point where the trend changes, just as I think there's no one reason why the Vietnam conflict turned out so poorly for the US. One can point to the advent of airmobility via helicopter, the development of accurate tactical missiles, the implementation of accurate space or near-space reconaissance, the 'discovery' of naval airpower, or any other number of innovations, and find a subsequent dramatic change in the way the US approaches and carries out military operations.
You can also look at the effect a strong telecommunications infrastructure has had on the public's reaction to warfare.
Right now we are indeed seeing a trend in the military which favors smaller engagements and increased intelligence. I think it's important to keep in mind that this trend is as much an artifact of prior military strategy as it is the result of technological advances and social changes. The US military's primary mission for a very long time was to prepare to engage a large foe of superior numbers (the USSR and the Warsaw Pact) in a defensive battle over preselected terrain -- Western Europe and the North Atlantic. Any shift from this style of engagement toward smaller engagements in diverse settings will seem exaggerated as a result as the military catches up to the new mission.
31. SheRex - 6/7/2000 12:22:21 PM
#14 - AytchMan
"CG argued that war has practically disappeared as an option for advanced countries. But if the media have become willing accomplices, that would make it more acceptable and likely, not less. Yes/no? "
More acceptable and likely - Yes. How many conflicts have we participated in just in the past 2 decades alone? Panama, Gulf War and the continuing conflict, Nicaragua, Caribbean states (Grenada), Somalia, Bosnia, Serbia/Kosovo .... We've carefully looked the other way on Chechnya, and we may end up looking the other way when it comes down to Taiwan.
I've seen some articles on the web by medical professionals describing their concerns over the population depletion in Russia. It is being estimated that the toll of drugs, crime and Aids/HIV will reduce the population of the former Soviet Union by nearly 50,000,000 by 2050 or even earlier. China could resolve alot of their population and resource problems by just easing on west a bit into the former SU. Would we choose to be involved or look the other way?
Individually, yes, I agree that we seem less willing to accept casualties ON OUR SIDE and in individual conflicts, but overall, I believe we're involved in more conflicts.
re your question on appropriateness of topic: fire away, anything's game in this whole domain.
32. SheRex - 6/7/2000 12:24:42 PM
angel-five:
Exploring non-standard intelligence gathering methods:
Military Remote Viewing Book List
Psychic intelligence is information obtained by paranormal means
Some Aspects of Anti-Personnel Electromagnetic Weapons --Synopsis prepared for the ICRC (International Red Cross) Symposium The Medical Profession and the Effects of Weapons
The Mind Has No Firewall by Timothy L. Thomas, From Parameters - US Army War College Quarterly, Spring 1998, pp. 84-92
US News & World Report cover story on wonder weapons 7-7-97
33. SheRex - 6/7/2000 12:25:01 PM
Low-Intensity Conflict And Modern Technology, ed. LtCol David J. Dean, USAF, part 1
Low-Intensity Conflict And Modern Technology, ed. LtCol David J. Dean, USAF, part 2
Low-Intensity Conflict And Modern Technology, ed. LtCol David J. Dean, USAF, part 3
Space and Electronic Warfare Lexicon - JCS
Africa2000 -Laser weapons use in Somalia
Nonlethal Weapons - A Global Issue
Ed Dames, psychic warrior
Non-Lethal Death, by David Guyatt
I've seen a reference to Jack Anderson having written a book on Psychic Warfare, but I haven't found the book yet.
34. jonesatlaw - 6/7/2000 12:32:30 PM
I wasn't very clear in my last post, I think that the US was crippled in Vietnam by political problems. Militarily the US was largely successful in its engagements with the Viet Cong and the NVA. However, the war was lost because of problems at the point where military and political objectives intersect and overlap. More than hardware, this what will shape future conflicts. Hardware responds to the needs of this political/military intersection. As more and more police type actions, and regional conflicts dominate US concern, more weapons and support systems will be designed around these needs. The carrier battle group was orginally designed around naval superiority growing out of Mahan's vision. It is now surviving as a means of projecting power across the globe in Mahan's "littoral" areas. It is also a staging point for incursions farther inland. I look to see these forces integrated with pre-built staging areas and support systems similar to those created in Saudi Arabia around the Kuwati war.
35. CalGal - 6/7/2000 12:40:59 PM
Did anyone read the recent New Yorker piece on Hirsch's book? Or the book itself? It discusses the questionable performance of McCaffrey's division in the Gulf War.
36. iiibbb - 6/7/2000 1:00:03 PM
What seems to be a consistant struggle in the military are those who want diversity of technology (i.e. specialized weapons such as the A-10 tank killer which excel in their role of close air support), and those who want versitility in technology (i.e. creating modular attack systems which don't really excel at the task, but can be made to fill a mission, such as adapting the F-16 to fill the role of the A-10). The later is obviously more appealing to accountants, and sound good... but in practice wouldn't you rather have the best?
I am also disturbed by innovations in weapons which rely so heavily on technology that humans lose contact with their environment, or lose control of the technology.
This new integrated system they're developing with HUD's for footsoldires, night vision, GPS, and a backpack computer/communications unit. It sounds great, and it allows fewer men to hold much more territory. However, if an enemy can manage to jam those systems and cause them to fail, our men could potentially be left high and dry.
Smart guns are another worry. Sure you can make transponders that allow only a specific soldier to fire his weapon, but the enemy may be able to overwhelm/jam that signal so the weapons are rendered useless. Or suppose a soldier loses his weapon and picks another of a dead soldier nearby, but can't use his weapon.
37. angel-five - 6/7/2000 1:00:26 PM
Can of worms, indeed. I'll peruse these links to see if there's anything new in them. All I've ever read or seen of 'remote viewing' was exceptionally questionable -- any project in which you feed remote coordinates to a psychic and they tell you what's there, and it turns out that the coordinates you give them don't matter at all, is more than a little dodgy.
38. Indiana Jones - 6/7/2000 1:02:42 PM
SheRex: Those articles that you think are particularly useful you can use your maintenance menu to add as links to the "butterbar" on the side so they're alway present.
39. angel-five - 6/7/2000 1:10:36 PM
Sonics and EM weapons have been around for a long time. We've been doing research on them since the early days of the cold war. I see no fundamental difference between them and other conventional weapons; they are designed to either kill their target outright or else wound or incapacitate their target, which is actually more debilitating to an enemy force in terms of reducing immediate manpower. No one stops to help a dead man; a wounded or incapacitated man requires aid from his comrades.
Of course, it's interesting to think about nonlethal weapons systems which can be used against rioters, etc., but we already possess potent nonlethal weapons. To me, the excitement about 'new' weapons is more directly related to their novelty than their utility.
40. DocBrown - 6/7/2000 2:28:49 PM
I've got to disagree with this, angel-five:
Yet since nuclear weapons happen to be the current boogeyman of
large-scale warfare, and we haven't yet come up with an adequate
military defense against them, we see people falling into the trap of assuming that nuclear arms (or chemical and biological weapons) have irrevocably changed the way in which wars must be planned and
waged. A quick review of history shows the fallacy in such thoughts.
The whole concept is weak. Technology has always given us a greater ability to destroy than to protect. The artillery of World War I could still inflict considerable damage on my house or my city, no force fields or deflector shields have been invented to protect me.
I am quite certain that we will not see anything capable of protecting us from nuclear weapons in our lifetimes. We can protect against various delivery systems, but not the warheads themselves. By the time we do invent some sort of protection (if it is even possible) we will surely have weapons that are many times more destructive.
For this reason, I believe that the way we wage war has permanently changed, thought today's weapons of mass destruction are only the first chapter in the new book of war. The old concept of wars for conquest is no longer feasible.
41. DocBrown - 6/7/2000 2:29:03 PM
I think we are down to two or three types of wars:
1) Wars of attrition. Here I include India / Pakistan and The Cold War. Instead of invading the enemy you win by outlasting them.
2) Wars of Control. These include Civil Wars. In such cases both sides consider themselves patriots and will not use weapons of mass destruction on each other.
3) Police Actions. These are really just expansions on the Civil War concept.
Right now, no country is in any position to attempt a war of conquest. SheRex suggested that China might invade the former Soviet Union. I doubt they would do this, due to the presence of weapons of mass destruction. The only way it might work would be if the former SU were in such governmental turmoil that it could not decide on a response. But I classify that as a civil war.
42. AytchMan - 6/7/2000 2:47:20 PM
Throughout our modern age (at least), there has been a steady drumbeat of moral outrage directed against most major new weapons technology: submarine warfare, poison gas, nuclear weapons; even the use of machine guns was judged as inhuman for a time. Generally, this outrage has been applied to weapons that engender a shift in either battlefield tactics, the balance of power or both. In other words, no one seems to get too upset over an improved grenade that merely shatters bodies a bit more effectively.
Which of the following weapons would provoke the most moral outrage, if any could be developed in 2005:
1. An electromagnetic pulse weapon that incapacitates everyone within a fifty-mile radius. The victims are rendered unconscious for 48 hours. Otherwise unharmed, they are, of course, subject to whatever may befall them while unconscious.
2. An EMP weapon that kills all living things within a one-mile radius.
3. The cloning of a single super-capable soldier for clandestine battlefield use only (a spy).
4. Psychic rifles that disorient individual soldiers for 15 minutes. The victims suffer powerful hallucinations and hysteria. Some become permanently insane.
5. Psychic gear that remotely controls individual enemy personnel indefinitely. The victim is otherwise unharmed.
Feel free to amend the list. Bonus points for explaining what your choice says about us as a society.
43. AytchMan - 6/7/2000 3:28:10 PM
Doc--
>>The whole concept is weak
I don't think so. Far be it from me to get into the middle of what could become an interesting exchange but here's one thought...
While I'm certainly not predicting it, I can envision the possibility that some branch of science will discover either an antidote for radiation or some sort of blocking technology to short circuit the process of critical mass at a distance. If this occurs, the trump card of nuclear weapons could perhaps be invalidated. Thus, Angel's premise would obtain as we reverted to more conventional methods of planning and fighting wars.
44. CalGal - 6/7/2000 3:31:46 PM
I would be delighted if technology that wiped out nuclear weapons' effectiveness were developed. I wonder if any work is being done on that? Probably not.
I think war works best if everyone involved has to get their hands dirty. I think the current American tendency to go in, bomb people, kill more of their own with friendly fire than the enemy can ever manage to take out, hold carefully scripted press conferences and play their own videotapes, and then come back home clasping their hands in victory is more than a bit distasteful.
45. DocBrown - 6/7/2000 4:03:21 PM
AytchMan, what is the physical basis for your idea? How would it interrupt the fusion reaction in an H bomb?
This technology does not exist. Do you believe that anyone is funding research into this nonexistant technology right now?
Besides, there are other types of weapons of mass destruction, and defending against them could be equally difficult.
46. DocBrown - 6/7/2000 4:08:17 PM
Those who expect technology to bring warfare back to its "glory" days on the battlefield are a bit like the WWII Admirals who dreamed that they could fight with battleships instead of airplanes.
47. AytchMan - 6/7/2000 4:18:23 PM
Doc--
I have no knowledge that any of this is in the works or even possible. I'm not making a technical argument, rather a technological one.
To me, the key word in Angel's post is 'irrevocably'. I readily agree that nuclear weapons have brought about a major shift in military affairs. I just think it likely that the march of technology will one day overcome present-day weapons and render the shift 'revocable'. Yes, new weapons will take their place but I think that supports Angel's argument rather than opposes it. Other strategies will evolve and the cycle will continue: offense vs. defense, concentration vs. dispersal and so on.
48. SheRex - 6/7/2000 4:30:14 PM
Laser shoots down rocket for first time -U.S. Army 6-7-00
49. angel-five - 6/7/2000 8:36:42 PM
Do some background reading on that laser system. The Pentagon tests on it demonstrate that it can't distinguish warheads from mirved decoys effectively: the only tests in which the weapons system accurately neutralized the actual warhead were tests where there was only one decoy and the decoy was nothing like the warhead itself.
The Pentagon analysts say that they didn't do a more exacting test because of time constraints.
Laser technology is real; this weapon system isn't, and will take a great deal of money to make it otherwise.
50. angel-five - 6/7/2000 9:13:44 PM
Doc:
You know, I might have expected that scientist-literalist boxed answer from you. It quite resembles what someone proposing a bulletproof fabric or a palmtop computer might have heard in the turn of the last century.
There's this concerted feeling in the modern social consciousness that we finally managed to open Pandora's box that day in New Mexico when Oppenheimer's team successfully detonated the first nuclear warhead. That's blindness -- as every technological age before us has done, we have been struck by terrible awe and fear by the capability of technology to kill, by the undreamt-of extent to which our weapons can obliterate the living. We've been sufficiently overawed, and lived with the fear of nuclear attack for so long, that we by and large don't imagine anything different in the future.
It happened with the advent of automatic weapons, with high explosives, with chemical weapons, with heavy bombing, with tactical nuclear weapons, with strategic nuclear weapons, and no doubt it will also happen with the weapons which military researchers are now working on (matter-antimatter railguns, EM disruption weaponry, orbital firing platforms, and so on).
The maker of the gatling gun, like so many other weapons researchers in their time, was so horrified by the destructive power in his creation that he hoped it would shock humanity into an end of warfare. He certainly didn't foresee the advent of the technologies which rendered his creation obsolete or the tactics which made it much easier to deal with a fixed machine gun. I humbly submit that something similar is bound to befall all those people who currently can't imagine a) anything worse than thermonuclear war or b) a potential future in which technology and tactics have rendered nuclear weapons obsolete.
51. angel-five - 6/7/2000 9:14:03 PM
If you can't imagine anything along those lines, well, you're hardly alone. It typically takes an Agincourt or a blitzkrieg campaign to convince most people that the old ways are obsolete.
Your digression on ways you imagine nuclear missiles might be countered, and problems with those methods, suggest that you're merely hung up on the specific means nuclear weapons might become obsolete, and not the process by which it's happened with every other major weapons innovation in the history of warfare.
The artillery of world war one might indeed do significant damage to your neighborhood. Any army fielding it would be annihilated, though, because modern ordnance is much more accurate and destructive over a much longer distance, and that's why no serious military still fields those weapons. Massed box formations of heavy bombers can level a city. No one fears them anymore because antiaircraft technology makes such boxed flights suicidal. A massed launch of ICBMs can render an entire continent morbid... but people have been working on anti-ballistic missile technology for decades, and the work continues.
52. angel-five - 6/7/2000 9:43:23 PM
Some might, indeed, point toward the way in which technology has made it easier to kill more and more efficiently on a larger scale as proof that it is indeed Pandora's Box are tinkering with. Weapons systems have become more deadly and more specialized, indeed; the thing to keep in mind is that they have become more deadly and specialized in a setting where other technologies are of similar make and scope.Witness the improvements in sword and armor technology and medieval military tactics, producing at their apex the heavy cavalryman. The heavy cavalryman (a knight) wore heavy armor invulnerable to most blows, a shield to block a foe's attack, a heavy lance which on the charge could punch through almost any defense, swords designed for specific types of armor and fighting, a heavy hammer or mace to bash a heavily armored opponent, and so on. He stood at the top of the military pyramid, the ne plus ultra. Tactics and strategy centered on his presence or absence from the field of battle. The (for lack of a better word) industry of armament shaped itself around heavy armor and the lack thereof. Elaborate training and extravagant outlays of cash were necessary to field heavy cavalry, a necessity which depended upon and in turn shaped the socioeconomic realities of war and government at the time.
And it all seemed so irrevocably set in place, until a dysenteric, ragtag army of peasants armed with English longbows and picks showed the world that things had changed. The scale has changed, but the pattern hasn't.
53. AytchMan - 6/8/2000 12:16:43 AM
Does anyone know of any near-future weapons technology that they believe should not be developed (for reasons other than cost)?
54. sakonige - 6/8/2000 1:28:22 AM
yow
55. Indiana Jones - 6/8/2000 8:53:32 AM
AytchMan: I'm not sure about the concept of moral outrage over a weapon...knowing how to do something but rather choosing to do it (i.e., using the weapon) would be the moral question. With that caveat, here's what I think about the weapons described:
1. Not any more terrible than existing weapons.
2. Worse than 1, but comparable to what we have.
3. I find this one offensive, assuming the clone has no free will in the matter. Creating essentially human slaves should not be permitted for any reason.
4. No worse than current technology.
5. Also immoral. I guess as in #3 above I'm bothered more by enslaving someone's mind rather than destroying his body.
56. Indiana Jones - 6/8/2000 8:55:24 AM
Addendum on #5: Might be okay in a war, but developing the technology is just too risky because I don't think its use would be limited to warfare.
57. DocBrown - 6/8/2000 10:58:24 AM
Angel-five, I have trouble understanding your position. You said (after your Gatling gun reference):
I humbly submit that something similar is bound to befall all those people who currently can't imagine a) anything worse than thermonuclear war or b) a potential future in which technology and tactics have rendered nuclear weapons obsolete.
Are you accusing me of falling into this category?
I absolutely can imagine wars worse than thermonuclear war and a future in which nuclear weapons are strategically obsolete. Even today, biological and chemical warfare could both be worse, depending on how they are waged.
Inventing a weapon more destructive than Hydrogen Bombs would render them obsolete. However, it would not reduce their ability to destroy the entire surface of the Earth in a matter of seconds. This makes any attempt to invade a nuclear-armed nation by force a silly fantasy, even if the invader has much bigger bombs.
Neither the H bomb nor the A bomb marks the ultimate moment of this change. The presence of any weapon of mass destruction is sufficient to produce a permanent change in the usefulness of warfare to conquer other nations.
58. AytchMan - 6/8/2000 12:54:49 PM
Indy--
>>knowing how to do something but rather choosing to do it (i.e., using the weapon) would be the moral question.
I agree, I could have worded it a bit better. But, having said that, my guess is that there are many people who think that merely developing a weapon is immoral because it could be used.
Interesting choices on the weapons. Anybody else?
59. DocBrown - 6/8/2000 2:33:46 PM
AytchMan, I find your list of science fiction weapons a bit boring. All of them are related to human combatants or targets. The obvious counter to them is a computer-controlled retaliation. The Doomsday Machine described in Doctor Strangelove, easily built with today's technology, would be utterly superior to all of them.
You could make a much more useful weapon technology simply by developing superior information gathering or processing techniques.
60. AytchMan - 6/8/2000 2:53:20 PM
Doc--
>>a bit boring
Rats, I've been found out.
Humor me and respond to the post (please). I'd be very interested in your take on it. There's a reason.
61. SheRex - 6/8/2000 2:57:26 PM
Joint services documents on strategic planning for future conflicts (available on web - totally unclassified) project that future conflicts are more likely to be small skirmishes where nuclear weapons are just not feasible (too many friendlies around). Hence, the drive towards developing high-tech anti-personal weaponry (EM, microwave, acoustics).
62. CalGal - 6/8/2000 2:59:16 PM
Aytch,
1 and 2 are just different methods of killing people. 4 and 5 make me uncomfortable, but I suspect that eventually they, too, would be considered just different means of achieving an objective.
3 is the morally problematic one, to me. It is problematic on genetic tinkering grounds, not because it's a nasty way of fighting a war.
63. SheRex - 6/8/2000 4:24:10 PM
#5 Psychic gear that remotely controls individual enemy personnel indefinitely. The victim is otherwise unharmed.
This would elicit the most moral outrage.
Because everyone would think that someone would use it on them causing them to withdraw all their money out of the bank, or perform perverse sex acts.
64. AytchMan - 6/8/2000 4:48:00 PM
Several issues are falling out of the responses but the one that I was trying to elicit is the distinction between large numbers of deaths and smaller amounts of suffering, broadly defined. To oversimplify it, I guess I shouldn't be so surprised that people readily accept 10,000 deaths (in war) but balk at controlling one person or making a few suffer. No judgments here, I just find this fascinating.
Along the same lines, many people oppose nuclear power because 40,000 people might someday be killed in an accident yet accept 40,000 deaths every year from automobiles. But that is a separate issue for another thread.
65. CalGal - 6/8/2000 4:51:24 PM
Aytch,
But the "balking" comes from a different source, in my case. I believe that cloning and genetic manipulation of people--to the point of mandating their destiny--is problematic, no matter what the reason it's used.
66. SheRex - 6/8/2000 6:34:08 PM
CalGal -
Just some clarification on genetic manipulation. If women discover during their pregnancy that their child has a genetic disorder that will cause them to be different from the other children, and this is fixable through gene therapy (genetic manipulation), or if they have genetic tests run on themselves and discover that there is the potential to give their child a debilitating condition, and they can reduce or eliminate the chances by genetic manipulation, are they justified in genetic manipulations?
Is it just the application of this technology in the warfare weaponry domain that concerns you?
67. CalGal - 6/8/2000 6:48:30 PM
She--if you notice, I said, "to the point of mandating their destiny". Obviously, the issues on genetic manipulation and cloning are many and varied. But in this case, we are talking about creating a human being tailored for the specific art of making war. That is well outside the bounds at this time, and it would unquestionably bother me no matter what purpose the person was designed for.
68. DocBrown - 6/9/2000 9:10:52 AM
SheRex said:
> #5 Psychic gear that remotely controls
> individual enemy personnel indefinitely.
> The victim is otherwise unharmed.
> This would elicit the most moral outrage.
Quite the contrary! This would be the prerfect weapon if it were possible. The posessor of this weapon would be able to define morality however he liked. It is impossible to disagree with someone who has the power to control anyone who disagrees with him.
Who wouldn't love this weapon? A Libertarian would use it to make everyone a Libertarian. A Conservative would use it to make everyone embrace conservative values. A Tree-hugger would use it to make everyone into vegetarians.
Once a victim is under control, he/she would be incapable of using any defense to break the control. He she would not even want to break the control.
Great concept for a fantasy story. Too bad it is impossible in the real world. If I had it I would turn you all into classic car enthusiasts . . . for your own good, of course.
69. SheRex - 6/9/2000 12:47:32 PM
DocBrown: "Too bad it is impossible in the real world."
Not so. The technology of microwave hearing places 'the voice of God' inside someone's head, and has been in development for quite a few years. How many people act without question on something God says?
All sorts of electromagnectic weaponry is in development right now to accomplish this very thing.
And, to take an even more radical stand - what the heck do you think TV is all about? No, you don't buy anthing that's advertised there, do you?
70. DocBrown - 6/9/2000 2:56:01 PM
The technology of microwave hearing places 'the voice of God' inside someone's head, and has been in development for quite a few years. How many people act without question on something God says?
Too many, I will grant you that. But if this works it is still just a form of communications, not "controlling individual enemy personel indefinitely."
The good news is that America already has this weapon. It is called Hollywood. The bad news is that Americans love to turn this weapon on themselves.
71. JayAckroyd - 6/10/2000 4:58:46 PM
I guess you're really serious about this psychic stuff. I hit message 42 and you're still making references.
There are no psychic powers to exploit. It just illustrates how willing we are to pour money down a military pork hole.
It's like the wacked notion that it is possible to construct a missile defense system. The Israelis have demonstrated the only method that can possibly work--hit them on the ground.
Of course, there are things we could do to reduce proliferation--mostly by reducing US and Soviet warhead inventories. It is the height of insanity (or a profound corruption, on the order of the defense and intelligence depts collaboration to overstate the Soviet threat)to deploy missiles to threaten missiles, when the other side is willing to reduce deployments if you will.
And that the reduction in deployment may be blocked by this pie in the sky nonsense of a missile defense is really shocking.
72. Uzmakk - 6/11/2000 7:27:36 PM
#70 Doc Brown:
cc(chuckle chuckle)
73. angel-five - 6/12/2000 12:10:43 PM
Not so. The technology of microwave hearing places 'the voice of
God' inside someone's head, and has been in development for quite a
few years. How many people act without question on something God
says?
I'm tempted to leave this to Doc, but...
This is nowhere near feasible, and won't be for a long time. Do you have any idea at all just how horrifically complex it would be to induce nerves to fire in a pattern replicating speech at all, let alone a pattern recognizable as the voice of a god? You would have to possess a complete and exhaustive neural map of the intended target and have that target surrounded in all directions by equipment capable of tightbeaming a signal on a rapidly moving target micrometers in size, retaining accuracy to within a few angstrom. I probably don't need to add that if you already have the opportunity to obtain the neural map in the first place, you have absolutely no need to bother with the rest in the first place.
And in any case, the tech to do that is a generation away at best. I don't know what you've been reading, but if it proposes induced hearing as a soon-feasible weapon then I'd humbly suggest it be taken with a boulder of salt.
Aside to Doc: No, I wasn't suggesting that you believed nothing could be worse than nuclear weapons. Don't be so quick with that glass slipper, Cindy.
74. DocBrown - 6/12/2000 3:41:24 PM
I was not sure if you were making that suggestion or not, Angel. I found your messages 50-51-52 to be confusing. The text was clear, but your argument with me was not. It is still not clear to me.
I envision a bright and promising future for artificial intelligence on the battlefield. Not only are computers more expendible than humans, they are also braver and have the potential to be smarter.
75. angel-five - 6/12/2000 9:21:05 PM
What's hard about it? You suggested that nuclear weapons would never be made technologically obsolete by the advent of new technology (read: a better defense, not a better bomb). I painted your suggestion (and similar ones) in myopic shades. It is short sighted to assume that defensive systems will never render nuclear weapons obsolete.
Jay:
You seem confident that antiballistic missile technology is a pipe dream. Are you limiting that observation to the present and the near-future?
76. angel-five - 6/12/2000 11:03:11 PM
Returning to Aytchman (I think it was Aytchman) and the list of things that are potentially unacceptable:
None of these weapons systems listed are any more offensive if used wisely than other more conventional weapons systems, and some are less.
They're unsavory? Warfare is unsavory to begin with; you have to be ignorant and/or shockingly divorced from reality in order to grow squeamish over the practice of mind control as opposed to the practice of carpet-bombing.
The problem with 'hyperwar' is merely the problem with our society as a whole -- the sanitized convenience of our comfortable existence allows us to lose touch with the unsettling side of reality. We are, in our safe vantage point, allowed the luxury of our common soothing illusions and the result is that killing an enemy soldier is less offensive to us than controlling his mind.
The killing's done by someone else; we don't have to shoot a man, and then watch him die. We don't have to imagine the course that his life might have otherwise taken, or listen to his wife cry at night, or explain to his children why we shot him until he bled to death. Above all else, we're in no danger at all. So for us, by and large, the issue of killing versus mind control boils down instead to the fact that mind control might possibly be used against us, whereas those big bombs get dropped on other people.
Needless to say, that's pretty fucking shallow.
To me at least, the real argument here has very little to do with specific weapons systems and whether or not they are disturbing to contemplate, let alone use. Rather, the real argument is whether or not we should embrace feasible, efficient and useful military technological advancements as they become available.
The problem with all weapons systems is exactly the same -- it isn't the consequence of their use, but their abuse, that most affects the world which we'd pass on to our progeny.
77. angel-five - 6/12/2000 11:03:25 PM
It's been argued in this community, with some merit, that the bombing of Hiroshima ultimately saved millions of lives. Yet the price of the Manhattan Project also includes the costs and fears surrounding the nuclear arms race and countless despicable acts perpetrated in the name of superpower realpolitik as well as the death toll in Hiroshima and Nagasaki. We must also include in the tally a modern world in which the fate of billions of people depend upon the actions of whoever the least stable and altruistic nuclear power happens to be, be it NORAD or HizbAllah. In order to be responsible, we have to construct this sort of ethical spreadsheet for any weapons system we'd develop.
Here's the rub, though: it really doesn't matter how the spreadsheet turns out. If the technology is feasible, efficient, and useful, we have to develop it anyway. Someone else assuredly will, otherwise. No matter how horrifying it is, no matter how terribly it might be misused, we have to develop and understand it so that we can be prepared for it in the hands of an enemy that will use it less scrupulously than we might. Trusting that others will follow your lead if you refuse to develop a technology, because you are concerned with the implications of a particular weapon, is to invite someone less scrupulous to master the technology and then master you. History is replete with examples of nations and people who couldn't keep up with technology or refused to embrace it once they had it. They don't get to write history books.
Yes, it wholly, utterly sucks that our choices seem to be either to develop a horrible weapon ourselves or else be at the mercy of whoever else develops it. It's awful. But it's also reality, and there's no better moral alternative currently available than to embrace the new technology with full caution to its misuses and maintain appropriate accountability for whatever government or military entity controls it.
78. rubberducky7 - 6/13/2000 9:42:10 AM
Re: Message # 58, AytchMan.
"... my guess is that there are many people who think that merely developing a weapon is immoral because it could be used."
of course there are such people. this is the country of moral outrage over a simple INS deportation matter and Jerry Springer waxing philosophical on nation tv about the "lessons" learned from leather biker nuns and the anorexic Mormons who love them.
79. JayAckroyd - 6/13/2000 10:52:41 AM
A5-
I don't think the missile defense problem is solvable in our lifetime.
There are two problems:
1) Decoys will always be cheaper than devices that will destroy the missiles.
2) You can't really test the things.
We've been at this for nearly 20 years, spent at least $60 dollars (there must have been black missile defense programs). For that amount of money we could have bought the bulk of the soviet warheads, and reduced force levels to the size of France.
It's possible to imagine technology, like brilliant pebbles or laser firing space platforms. But when you face the problem of hitting a five foot cylinder falling free in space surrounded by a hundred other five foot cylinders ten minutes from the time of your discovery that you have acquired a problem, your talking about very sophisticated, very widely deployed technology.
You can still beat the system by launching before it's complete. In fact, you create an incentive to do so.
If all else fails, use a truck bomb.
The weaknesses of this approach as compared to negotiating reductions in nuclear force and alert status with states like China or Russia, while using with conventional deterrence to stop proliferation seem stunningly obvious to me.
80. mandolin - 6/13/2000 1:01:13 PM
Another problem is software. During the Ronnie Raygun years, the Pentagon estimated it would 40 million lines of code to run a missile defens system. Even given the much reduced scope of current plans, there will still be many millions of lines of code -- the hard parts like pattern-recognition don't scale.
When I was at (cough-non-disclosure-cough) we mad a telco switch with about 8 million lines of code. After more than ten years of production, with hundreds of these things in the field, our goal one year was to reduce the number of known bugs to 3/thousand lines of code - that's 24,000 known software errors as a goal.
And as Jay points out, you can't test this software. It has to be right the first time. The physicists and such were all over SDI -- the computer guys were laughing. And even if God intervened to make it work, you could sail a mighty big bomb up the Chesapeake Bay in a tugboat.
81. AytchMan - 6/13/2000 1:46:33 PM
>>And as Jay points out, you can't test this software. It has to be right the first time
I don't understand this. Why can't one test the software? Just fire some dummy missiles over the desert and see if the system can shoot them down.
Otherwise, your argument applies to nuclear weapons themselves. Yet we developed those and there aren't many people who think they won't work. Indeed, that's the problem -- everybody agrees that they will work.
82. mandolin - 6/13/2000 2:30:19 PM
If it was that simple, that switch wouldn't have had so many errors. Just make some calls and see what happens, right?
Nuclear missiles are trivial in complexity compared to a system designed to intercept and destroy them, and the physics is well-understood, but we don't have a good handle on the design of complex software systems. Look at recent posts in the Technical Thread about getting a floppy drive to work -- and Win95 is absolutely trivial compared to a missile defense system (though I hear Win2000 has more than 20 million lines of code).
One design problem is that the inputs to the program are highly unpredictable: how many warheads per missile, how many dummies released at what point, what EM countermeasures are deployed by the enemy and by us, how many missiles from where at what speed, and dozens of other factors, many of which won't be thought of until the system faces them. You can't test for conditions you haven't imagined. If the software isn't properly written for all of them -- and it won't be -- it will fail.
Another is side-effects. Often some working feature, in some conditions, breaks something else, and the error does not become apparent till after many thousands of hours of use. How do we ensure new features don't break old ones? Simulations are not good enough. Regression testing is a major effort, and subject to the same failures of imagination as design.
I could go on about problems in design and test, but I'm probably boring people already. Then there are typos, operator errors, bugs in the compilers and assemblers, and plain old stupidity.
A telco switch has to shrug off error conditions, it being far better to drop a call than shut down a switch and lose 60,000 calls. A missile defense system has to do the same -- it has to keep going even if some missiles get through.
And it will never stop a bomb in a truck or a boat or a 747.
83. AytchMan - 6/13/2000 3:14:44 PM
>>If it was that simple, that switch wouldn't have had so many errors. Just make some calls and see what happens, right?
Not at all. I understand the complexity of the software development problem. I'm simply taking issue with the statement that one can't test the software. A lengthy, expensive process? Absolutely. But, if the extremely difficult technical issues can be resolved, I see no reason why a credible test program cannot be developed. I think putting a man on the moon in the '60s was of similar complexity. But NASA developed a rigorous, horrendously expensive test program. And it worked.
The big difference in the two efforts is that NASA couldn't literally test the last step, the actual flight to the moon. With missile defense, we can test the last step by firing dummy missiles as we expect they will be used.
Finally, rather than getting into a lengthy debate over specific software issues, let's just stipulate that not every contigency can be exhaustively covered. Perhaps this is the basic argument?
If not, I'm just saying that the software can be tested, not that any specific anti-missile technology will work.
84. mandolin - 6/13/2000 3:36:51 PM
Getting to the moon, while a huge technical project, was relatively simple conceptually, and the computing power necessary probably sits on your desk or your lap.
We knew what we had to do to get to the moon, but we have no reliable method for the design or management of very large software projects, and this is by far the largest ever conceived. It is not only large, but involves something we do effortlessly but which has proven to be nearly intractable for computers: recognizing objects from various vantage points and in various conditions. We don't know how to do this.
And if we did, it is still virtually guaranteed that any set of tests we develop will not match what the world delivers to the system. And even if we could, what is the target of such a system? A country with the resources of China or Russia can build missiles and decoys so much more cheaply than we can build the defense that they could overwhelm it easily -- and, as someone said above, it's existence provides an incentive to do just that. A country like North Korea or Iraq, or a terrorist group wouldn't even need or want to try -- why build an expensive, chancy, and easy to track (in the sense of seeing where it came from)missile when there are lots of boats? Imagine what even a Hiroshima class device would do in New York Harbor.
It's money spent for nothing except greater insecurity.
85. mandolin - 6/13/2000 3:39:49 PM
Still, you're right that, if the technical problems can be overcome (a very large if), then we can test the contingencies we plan for.
86. AytchMan - 6/13/2000 4:16:15 PM
mandolin--
No need (on my account) to mount a general attack on missile defense systems. I just asked a question about testing out the software.
>>Getting to the moon, while a huge technical project, was relatively simple conceptually, and the computing power necessary probably sits on your desk or your lap.
As one who lived through the era (and participated peripherally in some NASA work), I find this a surprising sentiment. Particularly from a software standpoint. I don't think this was considered a relatively simple concept at all at the time, expecially in light of the computer power available.
It seems at least possible to me that, 40 years from now, we might be able to substitute 'missile defense' for 'getting to the moon' in your paragraph and achieve a similar effect. Only time will tell.
87. DocBrown - 6/13/2000 4:23:39 PM
mandolin wrote:
It is not only large, but involves something we do effortlessly but which has proven to be nearly intractable for computers: recognizing objects from various vantage points and in various conditions. We don't know how to do this.
Sure we do. We know that the human brain accomplishes this with a type of neural network. We can simulate small nural nets in a computer. It is possible to build a machine that is millions of times larger and more complex than current computers.
The problem with neural nets is that they behave like humans. They can make mistakes. It may be true that the neural net is the best pattern recognition system in the Universe. If so, the best neural net in the Universe might be only a little better than a human at recognizing a complex event like a missile attack.
A missile defense system that makes errors is worse than no missile defense system at all. Even if the error rate were less than .01%, that system would still shoot down dozens of innocent aircraft per year.
88. DocBrown - 6/13/2000 4:23:51 PM
Angel said:
It is short sighted to assume that defensive systems will never render nuclear weapons obsolete.
Okay, maybe 1,000 years from now someone will have a force field that can protect you from a nuclear bomb in your desk drawer. But I doubt it.
This is a different argument. When you were arguing that offensive technologies were constantly becoming obsolete because better offensive technologies were being produced I agreed with you.
I will also agree that defensive technologies have allowed us to better and better deal with weapons' delivery systems. We can intercept AGMs and distract SAMs and AAMs. However, the only protection we have against the warheads themselves is armor, a technology that has improved only slightly in thousands of years.
A defense that could neutralize or absorb a nuclear blast could not be made of normal materials as we know them. Atoms and molecules just won't stand up to that. Can you describe the technology that might accomplish this, even vaguely?
89. mandolin - 6/13/2000 4:38:32 PM
AytchMan -- I eny you the chance to work, even peripherally, on that great achievement. And yes, given the computing platforms available then, there was considerable challenge. But we knew how to apporach the porblems algorithmicaly -- there's no guarantee that the recognition problem can be solved in that way.
DocBrowns' neural nets hold some promise, but the brain's nets are not just any neural net -- they're optimized over millions of years by natural selection to be very good at recognizing objects in the environment and attaching appropriate value to them though chemical and nervous connection with the rest of the body -- and, as he says, they make mistakes.
Genetic algorithms might provide an answer of a sort -- but Danny Hillis grew some sorting routines that, while they work better than any human-written sort, are incomprehensible even to him. He doesn't know how they work. How do you maintain software like that?
90. AytchMan - 6/13/2000 4:59:10 PM
mandolin--
Sorry for the lack of clarity -- I lived through the era and later worked on a NASA contract. I didn't work on any of the lunar programs.
Not being a mathemagician, I can't argue about what is or isn't algorithmically possible. Are you arguing that we should forego a missile defense system because we don't know how to do it now?
91. mandolin - 6/13/2000 5:07:47 PM
I think the more important point is that it costs a lot of money and does not enhance security. Lobbing big stupid, missiles with lots of warheads and decoys is so much cheaper than building systems that could stop them that antagaonists with the resources will be tempted to try just that, provoking an arms race --or worse, tempted to use their weapons before the system is operable. Smaller foes don't need to try since there are other, cheaper, more reliable ways to deliver one or two nuclear weapons -- or anthrax, for that matter.
92. AytchMan - 6/13/2000 5:24:42 PM
>>Lobbing big stupid, missiles with lots of warheads and decoys is so much cheaper than building systems that could stop them...
If that's true, then you win. The question, of course, is whether it's true. If near-future research turns up a relatively cheap, effective defense, should we deploy such a system?
93. mandolin - 6/13/2000 6:01:25 PM
maybe -- then the temptation to use or lose becomes fairly large for aggressor nations. I think then it's a diplomatic bargaining chip.
94. AytchMan - 6/13/2000 6:28:34 PM
mandolin
I'm not sure I'm following. Are you saying that, if a system becomes feasible, we shouldn't deploy it because the bad guys might attack us before we complete deployment?
95. AytchMan - 6/13/2000 7:50:39 PM
In rereading most of the thread, I've found the different takes on the concept of technology itself to be very interesting. To overstate the differences a bit, some are respectfully wary of technology and wish to take a cautious approach to minimize the attendant problems. Others place great faith in technology and feel confident that, sooner or later, it will solve most current problems.
First, is this a fair assessment? Second, does this correlate at all with political leanings (in general, not you guys)? If so, are there implications for future weapons development?
96. JayAckroyd - 6/13/2000 9:41:47 PM
Note that any testing program has to distinguish decoys from decoys, because you're not gonna launch any actual warheads. All a foreign power has to do is use decoys identical to the ones that were struck in the test.
Note further that the trip to the moon was done in gradual steps, testing all along the way. First the launch, then the orbit around earth, then lots of orbits around earth, the unmanned trips to the moon, then orbits around the moon, then a landing. I guess you could argue that's what DOD is trying to do, but they still haven't gotten to the orbit around the earth stage--reliably hitting anything with anything nearby. It took 11 years to get from an unmanned launch to the moon. We're approaching 20 years, and we're at the Alan Sheperd step.
Moreover the moon is easy. You're calculating a ballistic for a great big object that has a known path. You can spend months working out and testing the algorithm and parameter values to solve this known problem. A warhead destroyer has to solve the same problem for a tiny object surrounded by junk in a hitherto unknown position. In ten minutes.
There's a reason the devices keep failing the tests. It's too hard to do.
97. AytchMan - 6/14/2000 2:55:22 AM
jay and mandolin--
I realize I'm discussing this with two people at once but the arguments are kinda swirling around. Can we restate the main points against developing a missile defense system?
Please place in order of importance (most compelling arguments):
Cheap decoys
Inability to adequately test
Enemy attack before completion
Cost
Lack of effectiveness
Non-missile countermeasures (like truck bombs)
98. mandolin - 6/14/2000 8:05:59 AM
AytchMan -- I'd say they're all compelling, but I'd restate "enemy attack before completion" to read "risk of rekindled nuclear arms race."
It's just a bad idea.
99. JayAckroyd - 6/14/2000 9:25:24 AM
Cheap decoys
It depends on whether you're talking about the old SDI system or the new "we can stop one missile" system.
Old SDI
Lack of effectiveness
Cheap Decoys
Inability to adequately test
Enemy attack of system before completion
Enemy nuke attack before completion
Non-missile countermeasures (like truck bombs)
New idea
Lack of effectiveness
Cheap Decoys
Non-missile countermeasures (like truck bombs)
Inability to adequately test
Enemy nuke attack before completion
Enemy attack of system before completion
But all these arguments are important. And any one of them seems compelling. For example, I don't think attack before completion is a huge risk, for the same reasons I don't think it makes any sense to build the weapons. No leader is going to invite retailiation. If it really is the case that we have to fear attacks by rogue states, then building a defensive system will accelerate their efforts to get it done and launch before the thing is done.
100. AytchMan - 6/14/2000 5:31:34 PM
jay & mandolin--
I've found your comments on the missile defense problem very interesting. Time permitting, I'd like to get into the issues with y'all. Just for the record, I have no agenda to push -- I am, I suppose, an agnostic on the subject. That is, I can see no reason to pursue such a defense if the costs are staggering and the system won't work effectively. On the other hand, I'm one of the technology-faithful -- I tend to believe that we can solve most of the problems over time. Finally, I'm much more interested in learning the reasoning behind your views than in persuading you to change them.
So, maybe the meta-technological argument is the place to start. You all seem to think that the technological problems (pattern recognition, more complicated physics, inability to test software) are insurmountable. Is this based on first-hand info, specific analyses that you've read or some other source? Or perhaps it's just our different views of the way the world works and we'll have to disagree from the start. Comments?
101. mandolin - 6/15/2000 8:48:54 AM
AytchMan -- I don't think the tech problems are insurmountable, just very hard and, especially, much harder than the problems posed by such a defense for an attacker.
I have personal experience with writing and maintaining complex software systems, though none even within an order of magnitude of the complexity required for missile defense. They will do things (not necessarily failures) you don't expect; they will have errors; they will fail in unanticipated ways. Determined attackers -- even 15 year olds on hacked-together Linux systems -- will find ways to challenge the system and defeat it. Then the fix adds another layer of complexity and potential instability to a system already beyond the comprehension of any single human being.
The identification problem -- telling decoys from real warheads -- is one of the things we thought would be easy at the beginning of AI research, but it has turned out to be extremely difficult, while other things we thought would be hard, like playing chess, have turned out to be relatively easy. There is a vast literature, both on the computer science side and, lately, on the cognitive science side, on just how difficult it is. We are only now beginning to get a handle on how the brain and body work together to solve the problem, and the answer does not appear to be algorithmic, meaning that there is not a definable procedure guaranteed to reach a solution in a finite time. If that turns out to be true, traditional programming methods can't solve the problem. We may be able to use trainable nets and genetic programming methods, but, as DocBrown said, they are likely to be vulnerable to the same kinds of errors we are. They'll just be wrong a lot faster.
But more important than the technical difficulty, I think such a system is destabilizing, providing an incentive to a new arms race.
102. mandolin - 6/15/2000 8:59:53 AM
I said that trainable nets and genetic programming methods are likely to be vulnerable to the same kinds of errors we are, but, actually, it's worse than that. Minds are embodied -- they don't just use a particular hardware (or wetware!) as a platform, they are constrained in what they can think and how they think by their physical instantiation. There will be inherent errors and blindnesses in a genetically programmed system which will be different from our own and which we may be incapable of understanding.
103. JayAckroyd - 6/15/2000 1:32:01 PM
I hadn't thought about the way mandolin has expressed it, but I do think similar ideas are at the heart of the my conviction that this is more like fusion reactors than it is like fuel cells.
You really do need the equivalent of an AI to make this work. People couldn't calculate the ballistics or do target evaluation quickly enough. Computing power is orders of magnitude greater than it was when I first heard about AI in a college course where we were time sharing a PDP/1170 with 64 MB of RAM. People were very excited about it.
It's 20 years later, and we've beaten the hell out of the hardware, and have made no progress in AI. Promising avenues, like neural nets, have been explored and found wanting. We still can't reliably simulate the mind of a cockroach, never mind an AI capable of picking out the right five foot cylinder in free fall at a speed not too far from escape velocity.
That's a really hard problem. A good simulation would be a very challenging video game. So it's not just a matter of testing the code; the code has to work in environments you didn't test for.
So I guess one fundamental objection is I don't think we have (or will have) computing power to solve this problem, because it is not algorithmic in nature. It requires pattern recognition and judgment, both skills computers are extraordinarily bad at AND it requires very rapid calculations which people are bad at.
104. JayAckroyd - 6/15/2000 1:35:12 PM
Another fundamental objection is that the failure rate has to be zero. If you happen to have read Feynmann's recounting of his shuttle investigation, you'll recall a section where he forces NASA engineers (not managers) to admit that the official odds for failure are wrong by two orders of magnitude.
There are too many points of failure, with non-zero probabilities of failure, in a defense system to claim failure rates sufficiently close to zero to justify deployment.
105. JayAckroyd - 6/15/2000 1:42:03 PM
The arguments daisy chain. Suppose I'm wrong. Suppose that the tests are passed with flying colors. Someone figures out a specialized pattern matching system based on biological components, and it actually works. It's doesn't have many points of failure, so the probability argument is undermined. (I can't come up with a pseudo-plausible story on how to overcome decoys that look just like the test warheads that were hit, but there are plenty of people cleverer than me.)
Suppose all that, arguendo (as MS said in their scathing attack on Judge Jackson). Then the use 'em or lose 'em scenario grows dramatically in probabilty. You can't hide the launch facility. China, fearing for its survival, announces that if you build the facility, they will destroy it by conventional means during construction. North Korea (in a coordinated announcement) says if the conventional attacks fail, it will go nuclear against this military target. It regards the construction of the facility as an act of war.
Do you still build the facility? Is the world a safer place because of it?
106. Indiana Jones - 6/15/2000 1:47:38 PM
An old fairy tale: A long time ago in Germany a groom was visiting his bride's family, when the father sent the bride down to draw some more beer from a keg in the cellar. While she was filling the pitcher, she noticed an ax was stuck in the ceiling of the cellar. She though to herself, "What if Hans and I had a son and sent him down here to draw beer and that ax fell on his head and killed him?" Thinking of this, she sat down and began to cry.
Upstairs, everyone wondered what had become of her so the mother went down to see. She found her daughter sitting and crying with the beer overflowing on the ground. The daughter explained the cause of her grief, and the two sat down to cry.
Then the father came down and the same thing happened.
Finally, the prospective son-in-law came down. When the family explained why they were crying he reached up and pulled down ax (and lapped up all the wasted beer BTW).
The End.
107. Indiana Jones - 6/15/2000 1:51:27 PM
The point of course is, Are we looking at the simplest solution to the problem? (Sometimes I'm not even sure which problem the ABM system is supposed to address.)
108. Indiana Jones - 6/15/2000 2:11:39 PM
One problem that goes along with what Jay was saying about both being relatively untestable and needing to be fullproof: Suppose after the construction you are the American president and a rogue state blackmails you with the nuclear destruction of NY unless you concede on some point. You have a battlefield untested system that all the experts assure you will work.
Do you say, "Fire away, mothers!"?
109. Wombat - 6/15/2000 2:13:52 PM
At risk of getting a bit off topic, if the US is that concerned about the "rogue" state nuclear threat, wouldn't using most of the the vast sum of money that they are sinking into a speculative ABM system be better spent coopting or overthrowing the regimes in some of the so-called rogue states? There also seems to be the assumption that they will be around forever, which strikes me as highly unlikely.
Iran is certainly more open to an increasingly "normal" relationship with the United States, as is Libya (although it would be easier once Qaddafi departs). North Korea is a wild card, although setting a sum aside to help South Korea reincorporate the North if/when it implodes would be useful. Who else is there? Afghanistan? Iraq?
110. mandolin - 6/15/2000 2:16:05 PM
There is a rabies vaccine for people, and all personnel at the vet school here have to take it. And if one of them gets bit, they take the rabies treatment. No one is willing to test the vaccine's effectiveness.
111. mandolin - 6/15/2000 2:19:48 PM
Wombat -- there are terrorists and criminals as well as rogue states. But any of them, including the rogue states, would find it easier and cheaper to smuggle a bomb aboard some Liberian freighter and sail into San Francisco Bay -- or up the Chesapeake to the Pentagon.
Intelligence and enforced non-proliferation are our best hopes.
112. AytchMan - 6/15/2000 11:46:13 PM
Jay 105--
>>Suppose all that, arguendo (as MS said in their scathing attack on Judge Jackson). Then the use 'em or lose 'em scenario grows dramatically in probabilty.
To start with this issue at the very end of the development chain first: I'm having a lot of trouble addressing it because it simply runs counter to my sense of history. I'll give it a shot but we may have to chalk it up to differing world views and leave it at that.
First, doesn't it amount to simple blackmail? If so, the historical lesson to me is clear -- knuckling under leads to further threats and ultimatums.
Second, I can't imagine a scenario in which China would risk a full-scale conventional war with the US over such an issue. They might not lose but they could not under any circumstances win. I also cannot see how they could destroy a system across 6,000 miles of ocean with conventional forces (truck bombs are unrelated to this particular issue).
cont
113. AytchMan - 6/16/2000 12:09:03 AM
Third, if NK threatens nuclear war, there's two possibilities: they mean it or they don't. The relevant option of course is that they mean it. If so, we must either back down or take some Serious Damage and then vaporize them into the Cosmic Void. But this is a false case because they can do that now independent of a missile defense system (MDS). What if they threaten nuclear war unless Jay takes a job as Jerry Springer's personal assistant? What would you do, risk a nuclear war?
But even looking at a more general scenario where some bad guy somewhere objects and rattles his sabers, it just seems to me that history teaches a different lesson: appeasement, tribute and ransom lead to both of Churchill's bugbears, Shame and War. As I stated at the outset, perhaps we have differing world views. Even so, I think the whole subject is really an issue of how we conduct foreign policy and unrelated to MDS. Yes/no?
Finally, are there comparable instances in 20th-century world history in which a major power developed a new weapon, the bad guys directly threatened war over the weapon, the nation withdrew or negotiated away its potential advantage and there was peace?
114. AytchMan - 6/16/2000 12:25:33 AM
Indy 108--
Although your example is not specific (thus admitting of all sorts of qualifiers), I must say just that as a final position. Obviously, it depends on what they're demanding exactly. Also, I'd be willing to talk to them and negotiate in a very minor way. But on basic issues that I considered necessary to US security, I see no other alternative to a very diplomatically worded, "...and the horse you rode in on, pal."
But, as stated in the last post, the more I coj on this, the more I think it's an issue of foreign policy and unrelated to the MDS. Yes/no?
115. AytchMan - 6/16/2000 12:51:29 AM
j & m--
On to the truck bombs.
They (and related weapons) are a real threat and must certainly be countered. But I respectfully submit that this argument is also unrelated to an anti-missile system (let's call it AMS from now on). If this is an important argument against AMS, as you've stated, isn't it also an argument against military bases in general? And embassies? All of these facilities are equally vulnerable. And I think you'll have to agree that if everything else magically fell into place, we wouldn't want to cancel a cheap, effective AMS because somebody might try to park a Peterbilt a quarter-mile away. True?
I think we have to limit the evaluation of an AMS to the technological arguments, cost, testing, and decoys because these things are directly related to the system. And, quite frankly, you guys may well have enough there to carry the day.
116. AytchMan - 6/16/2000 1:10:17 AM
j and m--
Indy's 107 has nudged me to consider that we may not all be on the same page as far as what kind of system we're talking about. My impression is that we're discussing an anti-missile system (AMS) to destroy relatively small numbers of incoming missiles directed against the US by second-order bad guys (NK, Iran, Iraq, South Philly). The actual technology is unspecified but will be determined by near-future research (up to 30 years out). Comments?
117. JayAckroyd - 6/16/2000 8:59:48 AM
Even so, I think the whole subject is really an issue of how we conduct foreign policy and unrelated to MDS. Yes/no?
I think the decision to commit funds to this research is intertwined with foreign policy alternatives to reducing proliferation and limiting deployments. To the degree that the question is "Should we try to build a system that would stop a single missile?", we need to assess how likely that is in the current foreign policy regime, and then, if we consider the threat to be real, modify our foreign policy regime to reduce the likelihood. The proponents for missile defense have to believe 1) the threat is real and 2) the defense system will reduce the threat.
If the threat is real (I don't think it is, for the reasons you outline in your blackmail scenarios), then the foreign policy regime that includes the building of a missile defense increases the threat during construction.
A foreign policy regime which included a policy to take out with conventional weapons any offensive missile emplacements during construction would not increase the threat. A foreign policy regime with a focus on reduced nuclear forces in all countries could also include sanctions, enforced perhaps by NATO and Russia in concert, against nations that try to join the nuclear club, up to and including destruction of facilities.
This regime is cheaper, and reduces rather than raises the threat nuclear weapons use.
118. JayAckroyd - 6/16/2000 9:08:05 AM
And I think you'll have to agree that if everything else magically fell into place, we wouldn't want to cancel a cheap, effective AMS because somebody might try to park a Peterbilt a quarter-mile away. True?
You're missing the truck bomb point. If it is true that NK may want to nuke A US city, and we need to defend against such a threat, then a successful missile defense does nothing but remove a delivery mechanism. There are alternative delivery mechanisms, with the truck bomb the most obvious. Moreover, these alternative mechanisms are much more readily available than missiles are the groups most willing to use such force--terrorists with no clear affiliation to a particular state. Fortunately those terrorists are few, and delivering a nuke in a truck requires a resource and expertise level that is likely beyond them.
The best way to reduce the threat from those groups is stop funding, training and providing them with material in pursuit of tactical foreign policy objectives, as in Afghanistan.
119. JayAckroyd - 6/16/2000 9:09:31 AM
"these alternative mechanisms are much more readily available than missiles are the groups most willing to use such force"
should read:
"these alternatives to missiles are much more readily available to the groups most willing to use such force"
120. JayAckroyd - 6/16/2000 9:33:01 AM
On 112 and 113, I think you may also be missing the point wrt use 'em or lose 'em.
Again, you have to start with the assumption that you're wrong when you say this:
Second, I can't imagine a scenario in which China would risk a full-scale conventional war with the US over such an issue.
You have to assume that absent the construction of the system China, Korea or some other state is preparing to nuke the US with a single missile. If that state is China, then they've already made an internal commitment to nuclear war with the US under certain scenarios. A logical time to start that nuclear war would be just before the missile defense is completed.
You have to make that assumption because otherwise you don't build the system. You don't need it. To the the degree there is a real threat, it is deterred by something even stronger than mutually assured destruction--losing a nuclear war.
That's why I separated out the nuke the plant vs nuke a city alternatives in you request for ordering the arguments by how compelling I think they are. Let me repeat the argument once more in a slightly different way. To justify the system you must believe:
1. Washington DC is at risk from nuclear attack by a nation with only a few missiles. (You can't stop hundreds of missiles, so we're only defending ourselves against such states with this system.)
2. You can build a system that will stop such an attack.
3. The nation will wait until you finish the system before launching such an attack.
4. There are no alternative methods for launching the attack other than missiles, or that all such methods are already defended against.
121. JayAckroyd - 6/16/2000 9:33:21 AM
cont
When I said earlier that I find all the arguments I'd presented compelling, it's because you need all four of these assumptions to be true or you don't build the system. IMO, they are all false.
1) There's no real threat because retaliation deters.
2) You can't build it because it's too hard to do.
3) If you do try to build it, you're creating an incentive to launch.
4) There are other ways to deliver small numbers of nuclear warheads, such as trains, planes and automobiles.
122. Indiana Jones - 6/16/2000 9:38:42 AM
AytchMan: I would like to have a missile defense system capable of stopping a massive bombardment by a nuclear power equivalent of Russia and on down. I'm just not sure doing one that has a reasonable degree of effectiveness is worth the resource allocation required.
As an aside, I would assume that a defense system first of all would have to keep its method of operation secret to avoid counter measures, so I hope whatever they're really working on is more subtle than what we know about. For example, maybe they have some way of detecting decoys from the real thing by a hyper-sensitive radiation detector.
For 55 years we have been able to keep anything like this from being necessary. Surely surveillance and other security technologies are better now than in 1945.
If the problem is a massive bombardment, does anyone believe the threat will be worse than that posed by the Soviet Union in its heyday? If the problem is the single madman scenario, aren't there more effective ways to exploit our resources to solving it (intelligence, diplomacy)?
I think you cut back on the amount of this stuff floating around. No nation is a threat to our national survival through conventional means, so we have little to lose by reducing the importance of nukes in general. Even a nuke-free world wouldn't end American hegemony. Maybe we ought to concede some of our own arsenal to get others to reduce theirs and bribe the remnants with the money we'd be spending on missile defense.
I do think we continue the research in a limited fashion--enough to stay ahead of everyone else--and because you may discover technologies applicable in other areas. (Some have talked about being able to stop a meteor from hitting the earth, for example.) But without a Soviet Union-style threat, I'd have a hard time making it a budgetary priority.
123. JayAckroyd - 6/16/2000 10:16:32 AM
For example, maybe they have some way of detecting decoys from the real thing by a hyper-sensitive radiation detector.
That won't work. You can't use real warheads in your tests, so you can't use target acquistion methods that require the warhead to be real. As I said before, all the tests will do is teach the enemy what the decoy has to look like to fool the system.
You can't use velocity--the warheads and the decoys are ballistic objects, falling under gravity, so they will be going at the same speed, and with the same acceleration.
The only solution is to destroy all the decoys. In Teller's SDI days, that was the assumption. A dispersed set of independent systems would shoot everything down, with laser beams or little explosions or something. These sf ideas and Reagan's relentless advocacy of the concept did two things: 1) convinced a majority of Americans (and some key Russians) that it was possible 2) created a bureaucracy and collection of contractors.
So now it is politically expedient to take money from said contractors to win the election to give 100 or 1000 times that amount back to the contractors, and to maintain the bureaucracy.
This is a bad thing when it's an unneeded highway construction project. It borders on criminal when the pork project may create nuclear instability.
124. Indiana Jones - 6/16/2000 10:39:10 AM
Jay: Couldn't you use non-weapons grade uranium?
Anyway, my point was that an essential element of a defense would have to be not letting the other side know how it works...which is a pretty big reason for not "sharing" the technology.
Supposing for the sake of argument we spend a trillion dollars to develop a system that uses factor A to detect decoys from the real thing. We then share the technology. No one is now going to use that information to defeat the system (and likely for a lot less than a trillion dollars)?
Equalizing some of the uncertainty of having an untested system requires not letting anyone know how it works. That way, the element of bluff is at least as much on your side as the aggressor. Maybe the system isn't fullproof, but at least the other side doesn't know it can defeat your defenses.
125. JayAckroyd - 6/16/2000 12:39:28 PM
I'll concede that it isn't out of the realm of possibility that you could construct a test decoy that simulated a warhead. I do think it would be very difficult to simulate a warhead.
I agree that such simulation would have to be kept very secret. I suspect it would be very hard to keep very secret. The Russians got the A-Bomb and the H-Bomb design materials within a few years of US development.
That's because these were large projects, and it wasn't hard to find someone inside. (I'm forgetting the names of the people now, but this is discussed in Moynihan's Privacy.) An effective decoy simulator would also be a large project, and very difficult to keep secret.
Not least of the reasons is that our security kinda sucks. I did some work for a company that makes satellite components. The way secret procurement works is the work is divided up among contractors so they don't know what the ultimate product is. However, this is often done for not secret projects, to keep from compromising the secret projects. However, said a guy at the company, they always knew when they were working on black projects because the checks came from the department of agriculture.
When you look at this last ham-handed potential disasater in Los Alamos, you've gotta wonder if you really want to bet on something that has to be kept very secret.
126. JayAckroyd - 6/16/2000 1:09:30 PM
Speaking of US security, and the future.
Be scared.
Be very very scared.
What's really scary is the apparent cluelessness of the entire government in deal with technical change in the security area. A good book on this subject is Whitfield Duffie's Privacy on the Line
127. concerned - 6/16/2000 1:24:00 PM
I need to comment on post 79 because it is rife with misconceptions and unjustifiable assumptions.
I don't think the missile defense problem is solvable in our lifetime.
There are two problems:
1) Decoys will always be cheaper than devices that will destroy the missiles.
2) You can't really test the things.
Wrong on both counts. Decoys will *not* 'always be cheaper' than anti missile defenses. Any effective decoy will have to behave precisely like an actual missile, provide an impenetrable shield or osbcure said missile *for its entire transcontinental flight*. This is far from trivial. Additionally, the more effort and materials invested on 'decoys', the less there is for the payload (and/or the larger the overall device on launch which makes it easier to detect and neutralize before deploying any 'decoys'.
Tests can be performed to any confidence level desired to verify the capabilities of anti-missile systems. It's far too simplistic (to put it mildly) to just assume that the only way to debug and develop an effective anti-missile system is to experience an actual nuclear attack.
128. concerned - 6/16/2000 1:25:06 PM
We've been at this for nearly 20 years, spent at least $60 dollars (there must have been black missile defense programs). For that amount of money we could have bought the bulk of the soviet warheads, and reduced force levels to the size of France.
Our space program of the '50's and '60's had few successes for the first eight years, and such an infinitely less worthy and relatively more costly program received little of the type of criticism that is being indulged in here. Is a little consistency too much for me to hope for here?
Additionally, ICBMs will probably never travel appreciably faster or be much smaller for a given payload than they are now, whereas the potential and scale of possible anti-missile defenses has barely been developed, let alone considered by most people.
Consider also, that the SDI program has been greatly research oriented with small emphasis on implementation to date, so it strikes me as a more than a bit disiengenuous to make a mad rush to merely negative judgment as to the potential efficacy of all such programs, given the current state of things.
129. concerned - 6/16/2000 1:25:35 PM
You can still beat the system by launching before it's complete. In fact, you create an incentive to do so.
This is idle speculation that is never backed up with examples, and with good reason. It cannot be justified thusly, and probably cannot be at all at this point in time.
If all else fails, use a truck bomb.
Conventional security measures properly applied will usually neutralize the 'truck bomb' threat. IAC the level of destruction of any such device detonated at ground level is at least an order of magnitude less than a nuclear air burst from an ICBM (not to mention MIRVing) for a given size device, and this is factored into the equation by the serious and knowledgable.
The weaknesses of this approach as compared to negotiating reductions in nuclear force and alert status with states like China or Russia, while using with conventional deterrence to stop proliferation seem stunningly obvious to me.
China and Russia are not the primary initial threats that an SDI type program will deal with. I also feel it morally unaccepable, as well as very repugnant to contemplate exterminating tens of millions of foreigners in a nuclear exchange if any other avenue is possible at all.
130. concerned - 6/16/2000 1:29:50 PM
Sorry about any typos. I'm a bit under the weather and rather tired right now.
131. JayAckroyd - 6/16/2000 1:37:42 PM
This is idle speculation that is never backed up with examples, and with good reason. It cannot be justified thusly, and probably cannot be at all at this point in time.
I don't understand that response. Have you not read Strategy of Conflict? These kinds of arguments are absolutely standard.
The argument is simple. A country has a dozen missiles and poses a credible threat to the US. The US responds by building a missile defense system. If the country was really a credible threat, you'd expect it to launch before completion.
The closest historical example is the Israeli bombing of a (Libyan?) conventional missile site when it neared completion.
Conventional security measures properly applied will usually neutralize the 'truck bomb' threat. IAC the level of destruction of any such device detonated at ground level is at least an order of magnitude less than a nuclear air burst from an ICBM (not to mention MIRVing) for a given size device, and this is factored into the equation by the serious and knowledgable.
First, the defense systems don't hit missiles. They hit warheads, so your MIRV comment is off base. MIRVs are an argument for those who say the problem is too complex to solve. Sure, truck bomb nukes are smaller than an air burst. How does that factor into the equation? (Your statement seems to imply that there are acceptable levels of casualties. I'm sure that's not what you meant, given your last comment.)
I also feel it morally unaccepable, as well as very repugnant to contemplate exterminating tens of millions of foreigners in a nuclear exchange if any other avenue is possible at all.
I agree. Reduce the forces. Shut off the hair triggers. Get serious about slowing proliferation. Create foreign policy initiatives with "dangerous" states that opens up trade relations. But don't piss away $60 billion more dollars on a pipe dream.
132. JayAckroyd - 6/16/2000 1:38:57 PM
I meant to add a "destabilizing" pipe dream.
133. JayAckroyd - 6/16/2000 1:45:32 PM
Tests can be performed to any confidence level desired to verify the capabilities of anti-missile systems. It's far too simplistic (to put it mildly) to just assume that the only way to debug and develop an effective anti-missile system is to experience an actual nuclear attack.
Please explain how these will work. I think the simplistic assumption is that you can test. And the recent failures and dumbing down of the tests support this position.
Wrong on both counts. Decoys will *not* 'always be cheaper' than anti missile defenses. Any effective decoy will have to behave precisely like an actual missile, provide an impenetrable shield or osbcure said missile *for its entire transcontinental flight*.
So you launch 20 10 MIRVed missiles with 10 nukes in the mix. The rest are dud warheads. The defense system has to pick out 10 nukes from 200 identical objects in real time running. To make it interesting, launch a 21st missile that deploys chaff.
You say you can test this scenario? Cheaper than it can be deployed? In secret?
Hah.
134. AytchMan - 6/16/2000 1:49:07 PM
jay--
I definitely misread the truck bomb point. For whatever reason, I assumed you meant directed against the AMS. Now I understand.
You guys are making some very good arguments. While I would still contest the technical merits, I think the most telling point is cost. The more I think about it, the more it seems that we will need to spend tons of money ($100 billion or more doesn't seem out of the question) even if everything works just peachy. On that basis, I think the money could be better spent elsewhere. Does anybody else think that AMS could be brought in for significantly less money? In any event, I think some level of research should continue as insurance if nothing else.
By the way, is the Blue Ribbon for First Motie To Ever Change An Opinion still available?
135. JayAckroyd - 6/16/2000 1:52:12 PM
Consider also, that the SDI program has been greatly research oriented with small emphasis on implementation to date, so it strikes me as a more than a bit disiengenuous to make a mad rush to merely negative judgment as to the potential efficacy of all such programs, given the current state of things.
We're in the hole for $60B plus whatever black projects were commissioned. We've been at it since the early 80s. There hasn't been a successful test, or I guess I should say, a test that met the pretest criteria. The goal has gone from an inpenetrable shield against an evil empire's 6000 warheads to stopping a couple of missiles launched from a third world country. That goal is nonsensical--it presumes a suicidal third world country.
I suppose you think Cuban embargo is poised on the edge of successfully pushing Castro out of office. Here's a heads up: it wasn't supposed to be by keeping him in power until he died of old age.
(That last comment is meant as irony. But I do think the defense initiative compete with the Cuban embargo and the B-1 bomber as clearly failed initiatives that were not terminated when they should have been.)
136. concerned - 6/16/2000 1:52:18 PM
Your statement seems to imply that there are acceptable levels of casualties.
Making any such assumption is not justified based on my statement that truck bombs are far less effective than equivalent devices exploded at altitudes calculated for maximum destruction. However, I'll state that the whole concept of MAD depends on the idea that tens or hundreds of millions of nuclear casualties are 'acceptable', as does anybody who adheres to this idea.
The Clowntoon administration, as a side note, has has the net effect of accelerating nuclear proliferation with its policies. The genii is now certainly out of the bottle, if it wasn't before.
Better to do something useful about the current and developing state of affairs than to maintain the pretense of stuffing it back in.
137. JayAckroyd - 6/16/2000 1:52:33 PM
Sorry
138. JayAckroyd - 6/16/2000 1:52:44 PM
139. JayAckroyd - 6/16/2000 1:52:55 PM
I'm really sorry
140. concerned - 6/16/2000 1:55:06 PM
You say you can test this scenario? Cheaper than it can be deployed? In secret?
Hah.
You're laughing at your own straw men. I never imposed such an onerous set of restrictions and conditions on the verifiability of anti-missile defensive systems.
141. concerned - 6/16/2000 1:57:44 PM
The first two paragraphs in my previous were meant to be italicized. It looks like somebody put their toys away before I got to doing it for them:)
142. JayAckroyd - 6/16/2000 1:58:37 PM
The (Clinton)administration, as a side note, has has the net effect of accelerating nuclear proliferation with its policies.
That's true.
The genii is now certainly out of the bottle, if it wasn't before.
Better to do something useful about the current and developing state of affairs than to maintain the pretense of stuffing it back in.
That's not true. The Russians would cooperate in a reduction of force. They might even cooperate on a treaty to take out any new missile installations with conventional forces. India and Pakistan are very scary, but those missiles aren't targeted at us.
Speaking of targeting, Obsolete and destabilizing targeting .
143. JayAckroyd - 6/16/2000 2:02:10 PM
You're laughing at your own straw men.
Again, my apologies for the toys.
You say tests can be done. I've offered reasons why I don't they can be done. Your response is to say that those reasons don't apply, but offer no other rebuttal. If you could point to some successes, or some progress you'd be on somewhat stronger ground, but in fact the progress has been entirely negative.
This is not about defense. It's about providing Gore cover for a "democrats soft on defense attack" and keeping money flowing
between Senators/Congressmen and defense contractors.
144. JayAckroyd - 6/16/2000 2:06:55 PM
On that basis, I think the money could be better spent elsewhere. Does anybody else think that AMS could be brought in for significantly less money? In any event, I think some level of research should continue as insurance if nothing else.
By the way, is the Blue Ribbon for First Motie To Ever Change An Opinion still available?
The argument for the defense system would of course be much stronger if it were gonna be a lot cheaper. At this point, we've seen plenty of evidence that this is way harder than we thought, and certainly much more expensive. The alternatives--paying for currently deployed systems to be taken down, reducing our force levels, lower the state of alert--are all either cheaper or a reduction from current costs.
Dusty has said he's changed his mind about transparency issues on the net because a thread here. I've certainly refined opinions here. I think my view of the new economy has changed because of discussions here.
You don't see me in Politics much because I don't see much evidence of a Blue Ribbon be awarded anytime soon in that forum.
145. Wombat - 6/16/2000 2:09:40 PM
MAD assumes that 10s of millions of casualties are not acceptable: no country will risk that many casualties by initiating a nuclear exchange. MAD worked extremely well in the Cold War, and still does today. I cannot speak for the Indians and Pakistanis, who may be capable of anything, but it will most like be against each other.
146. JayAckroyd - 6/16/2000 2:09:47 PM
I would support basic research into technologies that would underlie an effective missile defense system--mostly into trying to define the problem in a usefully algorithmic way or trying to work out a specialized pseudoAI. That wouldn't be direct DOD spending. University grants and so forth. First make some progress on the underlying technology before trying to build the devices.
147. JayAckroyd - 6/16/2000 2:12:01 PM
MAD assumes that 10s of millions of casualties are not acceptable
Good catch wombat. The current program rests on the belief that turning North Korea into radioactive glass is not a deterrent.
148. JayAckroyd - 6/16/2000 2:25:38 PM
. I never imposed such an onerous set of restrictions and conditions on the verifiability of anti-missile defensive systems.
Missed this.
You never imposed any set of restrictions. Isn't it time? Isn't $60B+ and nearly twenty years enough time to determine some of these parameters? If this were a program doing research to reduce criminal recidivism through counseling and work release, would you hold the program to same standards of failure?
149. JayAckroyd - 6/16/2000 2:29:05 PM
concerned one more comment on I never imposed such an onerous set of restrictions and conditions on the verifiability of anti-missile defensive systems.
This seems to imply you wouldn't impose realistic tests on the system. Why not? Seems to me that it would be kind of important to know whether or not it would work.
150. mandolin - 6/16/2000 2:53:00 PM
I haven't abandoned this discussion -- just been thinking Jay's been doing an admirable job.
AytchMan raises an interesting point about the dismal history of appeasement -- and I agree with him, but think it actually counts against AMS. A rogue state or terrorist group won't be stopped from trying by some probablity, no matter how high, that theor attack would fail. But if it knows that any attack will be catastrophic for the attacker -- it if knows that even trying to construct the weapon will be catastrophic -- it may well think twice.
The retaliation need not be nuclear. Carpet-bombing the government buildings and every military installation in the country will do the job nicely against a rogue state, without the collateral radiation damage to neighbors. Against terrorists or other criminals, a response similar to but more robust than our response to the bombing of our African embassies can also be effective. Note that Afghanistan, after a decent face-saving interval, became uncomfortable harboring bin Laden.
Strictly enforced non-proliferation, reduction of overall weapon counts, and good intelligence are more feasible, effective, and cheaper than AMS.
151. Wombat - 6/16/2000 3:01:53 PM
Mandolin:
I agree that retaliation need not be nuclear, but the use of a nuke against the US would lead to a much more severe response than what you describe. It would mean the destruction of the regime in place by conquest and occupation. Even "rogue" states realize this.
152. mandolin - 6/16/2000 3:07:08 PM
I hope they do -- but occupation has its own headaches for us. It would be better to make d=sure they're dead and no one left has an army and then get out before we become targets for revenge-minded teenagers with AK-47s and Stingers.
153. concerned - 6/16/2000 3:32:30 PM
Strictly enforced non-proliferation, reduction of overall weapon counts, and good intelligence are more feasible, effective, and cheaper than AMS.
Strictly enforced non-proliferation?!?!!?? Muahahahah! Delusional Lefty on the loose! We're nowhere with the nonproliferation gambit and losing ground fast.
154. mandolin - 6/16/2000 3:34:35 PM
Strong gust of wind, eh?
155. arkymalarky - 6/16/2000 3:39:35 PM
Psst! Mandolin! Mention Clinton and see what happens!
156. concerned - 6/16/2000 3:51:09 PM
Re. 149 -
Jay, with all due respect, I don't believe you (or I) can sit in our virtual armchairs and dictate to the SDI systems developers what constitutes an allowable level of functional verification of their systems.
What has been patently obvious, though, is that the more vociferous opponents of any form of SDI and civil defense in general are on record as having vehemently and arbitrarily used this ruse in the hopes that it would obstruct rational discussion of the matter.
157. concerned - 6/16/2000 3:59:57 PM
Re, 152 -
I hope that post was some form of dark humor....
158. JayAckroyd - 6/16/2000 4:17:28 PM
156
"Trust them" is your response? Do you trust government officials on other issues?
And why are you citing other irrational people? We're having a discussion here. I've ready plenty of rational analyses by scientists who say that an effective missile defense is beyond our current abilities. Your response is that the government officials in charge of this, and the contractors with their snouts in the trough must know what they're doing?
The objections I'm raising don't seem unreasonable. Does this mean they don't have a response to these objections, that you know of?
159. JayAckroyd - 6/16/2000 4:24:33 PM
We're nowhere with the nonproliferation gambit and losing ground fast.
The US has undermined all non-prolif agreements by refusing to cut their stocks. This is especially appalling in light of the Russian need to cut their stocks. We could be under 1000 warheads in a couple of years, which would greatly reduce the risk of prolif and take the away the nonnuke countries position on the moral high ground.
160. JayAckroyd - 6/16/2000 4:27:11 PM
And, btw, when did any lefty propose pre-emptive conventional strikes on nuclear facilities in countries trying to proliferate?
161. concerned - 6/16/2000 4:34:45 PM
Re. 158 -
An effective missile defense may possibly not be implementable on a short time scale, given our current capabilities, I agree. I'm talking about, say, less than three years. But, this is subject to change as technology advances.
I credit our SDI developers with enough perspicacity to be able to anticipate and plan for feasible decoy technologies. For examples, a rogue nation will not likely invest in 100 decoy missiles for every one with a live warhead, because that would be prohibitively expensive. With a staged missile defense approach, from launch to re-entry, it is the decoys which become increasingly infeasible, since they will have to convincingly emulate or mask their 'assigned' live missiles through every stage of their trajectories.
I, of course, am not advocating simply 'trusting' government officials (given my opinion of this administration, how could I?), but I think a non-offensive system such as SDI has a huge potential payoff in saved human lives and wrt ecological concerns, and is thus very worthy of serious investigation.
162. JayAckroyd - 6/16/2000 7:23:35 PM
The ideahas huge potentials.
So does fusion for electrical power.
But neither is close to being real.
163. mandolin - 6/17/2000 12:36:21 PM
I think fusion power is closer, and AMS inherently destabilizing and thus a bad idea, even if practicable.
164. angel-five - 6/17/2000 10:06:25 PM
We should credit military developers with enough perspicacity to assume that they'd have accounted for obvious flaws in their products?
Whyever should we do that??? I mean, this isn't a tabula rasa. A review of the recent history of military acquisition shows -- repeatedly -- that military suppliers will sell faulty products to the government and that the government will buy them. We have B-2 bombers that cost over 2 billion each and they can't fly in the rain. We have B-1s and F-111s that were purchased at well over the initial price and then required billions more in expenditures and years more in tinkering to render them battleworthy. We have Apache helicopters that are so prone to breakdown that typically a unit of Apaches can only field a quarter to a third of their unit strength at one time. We have Bradley APCs that you can shoot through with a 12 gauge round. That's just for starters.
These are among some of the American military's most vaunted, highest profile assets.
So, no, I don't think we should just give these people the benefit of the doubt -- not when they get paid for convincing the government to buy a product whether or not it ultimately works as advertised.
165. angel-five - 6/17/2000 10:06:43 PM
Defense industries lay out large sums of money to develop new product lines, and allocate large portions of their own resources in the act. These are projects which often run a dozen years before they produce something useful. If, say, Lockheed sets out to develop the next generation of high-speed reconnaisance aircraft, by the time (if ever) their planners have a prototype to show the USAF they've spent tens of millions if not hundreds of millions in R&D. IF they get the conditional go-ahead from the government to start production -- provided that certain problems are corrected with the product -- they will have spent a great deal more money by the time they have a finished product ready to deliver -- whether or not the problems are corrected.
Executive jobs, not to mention the jobs of blue collar workers who will be laid off if the company doesn't get enough contracts to employ them, are on the line.
There is consequently a great deal of pressure on the defense contractor to sell the product whether or not it is useful or workable, and on the project staff to hide the inadequacies of their product when presenting it to the government. That's elementary, as elementary as the simple greed which also factors in defense sales. In an ideal world, such pressures wouldn't lead to the development and sale of flawed or useless systems, and in an ideal world the government acquisitions people would make sure they were buying a good product. Our world doesn't seem to work that way.
166. angel-five - 6/17/2000 10:29:12 PM
I definitely believe that some form of workable antiballistic missile defense will be developed if not fielded in our lifetime, for the simple reason that such technology will afford the possessor a great advantage over the others, and the current problem is one of scale and not of technique. Jay's arguments are logical, and there's a lot to them, but they only look at one angle -- a ground based missile defense system that has to pick and choose between MIRVed warheads and decoys. As opposed to a system which interdicts the missile before it releases its MIRVs, or a system which targets both warheads and decoys alike, or a jamming system that crashes the missile system or redirects the missile, or something else more esoteric than I am currently imagining.
I'm also not entirely convinced that the problem with pattern recognition is unsolvable within our lifetime, or even within the next twenty years (in my case :) ). It's not a matter of faith in science, the surety that new technology will solve all our old technological problems if we try hard enough. I have very little faith in that. It's simply that if you examine history, you'll find countless instances where an unforeseen technological discovery or economic change has rendered the previously impossible possible. Look at the pioneers in computing who scoffed at the notion of home computers, or the people who provided exhaustive lists of reasons why man would never reach the moon.
167. angel-five - 6/17/2000 10:50:29 PM
Oh, and while missile decoys may be cheaper than antimissile technology, a missile can only carry a number of them which is limited by a) its size and b) the willingness of an aggressor nation to sacrifice warheads for decoys. The number of targets which an antimissile system can identify and engage is under no such limitation.
168. JayAckroyd - 6/18/2000 7:41:36 AM
A-5,
The heart of my belief that this is not possible in our lifetimes is based on my conviction that it can't be done without AI, and that AI is not possible in our lifetimes. (Thanks, mandolin, for getting me to realize that was implicit in my position.) I don't think that intelligence can be acheived by a device that can be represented as a sequence of zeros and ones on a strip of paper. All of our computers are Von Neumann machines with that characteristic.
Now I agree that there is always the possibility of some wild-ass breakthrough, these days most likely involving biological components. But the time to consider building these defensive systems is after the breakthrough.
169. JayAckroyd - 6/18/2000 7:58:28 AM
Jay's arguments are logical...but they only look at one angle -- a ground based missile defense system that has to pick and choose between MIRVed warheads and decoys. As opposed to a system which interdicts the missile before it releases its MIRVs, or a system which targets both warheads and decoys alike, or a jamming system that crashes the missile system or redirects the missile, or something else more esoteric than I am currently imagining.
I focused on that particular system because that's the one under development. Keep in mind that it's the one under development after exploring alternatives like the ones you're raising. As I've said, our history on this particular project is unidirectional. In the beginning we were going to stop all the missiles with clever technology, like laser beams on a space-based platform, or tiny intelligent suicidal satellites. Steadily, as we've been unable to develop such systems, the system requirements definition has degraded to something that is closer to our capabilities--a ground-launched defense against a single missile.
That's a silly mission. No matter what the number is that can be stopped with, say, 90% probability, one, two, three, seven missiles, whatever, all you're doing is telling the rogue state how many decoys and warheads they need. You can't conduct the tests in secret. So even if there is a successful test, all that amounts to is a requirements definition for the response. (And, please remember, this argument assumes that the threat from the rogue states is real, which isn't the case, and that it is possible to build a system with 90% reliability to stop one warhead with a few decoys, which tests indicate is also not the case.)
170. JayAckroyd - 6/18/2000 7:58:35 AM
The need for the continuation of the program for political purposes has overwhelmed the mission. And now it's reached a point where the program is not merely an expensive pile of pork but is actually counter-productive, strategically and in foreign policy terms.
It's appalling that Clinton prefers to give Gore cover on this issue instead of doing what I am sure he knows to be the right thing. But as A-5 points out, once these programs acheive a certain momentum, they are very hard to kill, whether the devices work or not.
171. DocBrown - 6/20/2000 12:30:45 PM
The strategic and foriegn policy liabilities to which JayAckroyd alludes are the most dangerous parts of the anti missile system. The cost in money and brainpower is of less concern to me.
If America spends a zillion dollars on a missile defense system just to flex its muscles in front of the world it will not bother me. This would be a lot like the Apollo program that put a man on the moon. The value of the project itself is dubious, but the value of the technology we must invent to make it work (e.g. AI) will be inestimable.
Apollo also taught us some important lessons in goal setting, project management, and logistics. These lessons do not come cheap.
172. mandolin - 6/21/2000 11:13:23 AM
GPS slips by while debate on useless missile defense continues: good article in Slate.
173. mandolin - 6/21/2000 11:23:22 AM
some great quotes from earlier Slate piece on the issue:
"Muammar Qaddafi, for example, may seem erratic, but look what happened when Ronald Reagan gave him a sanity test. American jets bombed Qaddafi's house as punishment for sponsoring terrorism. The question was: Would Qaddafi a) retaliate, b) not retaliate but maintain a conspicuous association with terrorism, or c) start keeping a lower profile? He chose c) and thus passed the test."
"Kim Jong Il is about to embark on a dialogue with South Korea that may well lead to enough Western aid to feed the masses and thus save his political future. And one reason he'll get so much aid is that he fired those missiles and scared George Will half to death. That isn't crazy. It isn't even 'mysterious.'"
"Is it possible that some rogue dictator would dismiss the threat of American retaliation not because of an insouciant disregard for continued existence but because he doubted American resolve? ... (Saddam Hussein may be laboring under a misconception or two, but American reluctance to drop bombs on Iraq isn't one of them.) Still, in the event that the credibility of American retaliation is a problem, I have a solution. ... If Congress conspicuously passed a law authorizing the assassination of leaders whose states had launched a nuclear attack, that might help clarify their thinking. You think I'm joking? Hey, it would cost $60 billion less than missile defense and be much more effective."
174. iiibbb - 6/21/2000 12:14:19 PM
http://slate.msn.com/Earthling/00-06-20/Earthling.asp
175. DocBrown - 6/21/2000 12:46:47 PM
Last Friday I ran a Road Rallye for my DeLorean club. Many participants used GPS to navigate and they were glad for the enhanced accuracy. The funny thing was that the car that got the most disoriented was a GPS car.
Our tax dollars paid for GPS, and I am glad that we can now use it to its fullest ability. It was like getting an early Christmas present.
The idea that terrorists will now use GPS to guide missiles is silly paranoia. Does anyone really believe that the older, less accurate GPS data ever deterred any terrorist? Did a terrorist say, "Darn it, we would like to use GPS to guide a missile into New York, but the data is only good enough to get us within a block of our target! Too bad for us, we might as well just pack up our Scud launcher and go home!"
176. mandolin - 6/21/2000 12:57:44 PM
Doc -- it's not scuds.
It's model planes -- it's trucks, it's cars. And it's not Saddam, or even necessarily foreign terrorists, but somebody like the Army of God, or the next Timothy McVeigh.
You missed the point of the article, which was not that accurate GPS is necessarily bad (he emphasized that repeatedly). The point isthat missiles from rogue states or terrorists are a bogus issue, while GPS can really be used on the cheap by all kinds of people for purposes that were never debated.
177. DocBrown - 6/21/2000 1:09:25 PM
So what? Roads and sewers can also be used on the cheap for nefarious purposes, too.
Unless the terrorist is using an unmanned, long range aircraft such as a missile, GPS offers very little advantage. Users of model planes, cars, and trucks can achieve the same results with a good, old fashioned road map. Shall we debate the security issues surrounding Rand McNally?
178. mandolin - 6/21/2000 1:31:59 PM
When hobbyists can use GPS to navigate a pilotless plane across the Atlantic, there is considerably more capability attached to GPS than you suggest. It would, in fact, make the sewers easier to use inplanting a bomb (not necessarily nuclear).
Look, neither I nor Noah (the Slate piece's author) think GPS should be restricted for security reasons. But it is a much more real threat than missilies from "states of concern" or from terrorists. And the decision was made with no policy debate. Anti-missile sysytems get promoted and debated because they're big money, not because they have anything real to do with increasing security.
179. DocBrown - 6/21/2000 2:22:01 PM
With your last sentence I agree. However, I'd rather spend the money developing new technology than old model weapons. New technology eventually pays off for me.
180. JayAckroyd - 6/21/2000 7:46:03 PM
I think you folks are talking past each other. High resolution GPS is out of the toothpaste tube. I think you both agree on that.
That may be a bad thing (I found Wright's piece unpersuasive on the side of it being a bad thing.). IAC, it's over. Hirez GPS is how it's gonna be.
181. mandolin - 6/21/2000 10:19:03 PM
The worst thing is that that piece in the California museum won;t be wandering so much anymore.
(it's a mobile piece that tried to keep itself at a certain location as specified by GPS. Used to wander all over. Now it's just quiver, mostly.)
182. AytchMan - 6/22/2000 12:45:05 PM
Since AI was briefly mentioned:
Why is it that, in a field with so much potential (not too far short of genetics, medicine and computers in general), artificial intelligence (AI) has been such a bust? The worst of it is that, as Jay opines, it doesn't look like there will be any significant progress anytime soon, short of some magical breakthrough in, I'd guess, mathematics.
Ironically, AI is the only field I know of where each accomplishment seems to change the definition. That is, every time somebody solves a problem with AI, presto, it's no longer AI. Consider the history of chess-playing computers. Back in the 50's and 60's when the problem came under attack, it was the Holy Grail of AI. "When we solve this problem, we'll have true AI". Well, of course, Deep Blue has solved the problem. Not only don't we believe we've cracked the code, we no longer even define it as AI! It's just a bunch of very powerful algorithms that trade off various choices. What's going on here?
183. DocBrown - 6/22/2000 1:06:07 PM
AI is a fuzzy term, alright.
Generally, I list any computer implementation of any algorithm to make a decision as AI. So Deep Blue and Optical Character Recognition are AI. A computerized automatic teller machine is not.
BTW a few years ago my final Master's paper was on the failure of AI to life up to its promises. My conclusion was basically three parts:
1) Predictions Wild or Mild? The grand predictions for AI mostly came from the 50s and 60s. When I explored them, I found that few of the serious predictions were very far off. Computerized chess players, are a reality, but the Turing test may still be a few years away.
2) Bill Gates. When they made those predictions in the 50s and 60s, AI was THE hot computer field. Starting in the 80s and continuing through the 90s, the industry devoted huge amounts of money and manpower to personal computer applications and GUI stuff to make them a tool for human intelligence. The peoiple who made the predictions did not anticipate this diversion.
3) The Secret Success. In many ways AI has been a success. You may see it every day without realizing it. Your Word Processor uses AI to check spelling and grammar, while your camcorder uses AI for its digital image stabilization. Some accomplishments of AI are underestimated because humans overestimate themselves. Some of us still believe in causeless Free Will.
184. AytchMan - 6/22/2000 1:27:25 PM
Doc--
That's an awfully broad definition of AI, isn't it? Applying it, almost all software gets classified as AI.
While I can't put a formal definition on AI, I've always included a "learning process" in it. In other words, expert systems are not AI because they're completely scripted and should never deliver unexpected results. Is this no longer part of the definition?
185. mandolin - 6/22/2000 1:52:26 PM
There's a conventional distinction between "Strong AI" and "Weak AI." Chess programs and expert systems in general, including dictation software, are examples of weak AI -- specialized systems intended to solve a particular class of problems using algorithmic methods. Learning is often, but not necessarily, a part of such systems.
Strong AI claims that the mind is algorithmic, that is, a Turing machine (sometimes it gets gussied up woth things like "fuzzy logic" or the neuronal models). If the mind is algorithmic, it follows that the hardware doesn't matter (except for speed) as long as we get the software right, and we ought to be able to build a computer that is intelligent in the way we are.
This is almost certainly wrong, IMNHO, since the mind is clearly not algorithmic, nor is it just a matter of getting the weights right in connections between neural chips. The mind is an activity of the brain and the rest of the body, with complex chemical and neural feedback systems, all evolved under the pressure of natural selection, which allows us to make behavioral choices very quickly and with very little information. Wrong decision -- out of the gene pool. It has biases built-in and others that develop in response to the environment (including other people), expressed in the emotional tone of various scenarios, which have nothing to do with logic.
The mind is not a general purpose computer -- it is a survival machine shaped by our history.
186. JayAckroyd - 6/22/2000 1:56:07 PM
When I say AI, I mean a machine that will pass the Turing test with a moderately sophisticated user. We're nowhere near that, although the study of AI has generated an enormous amount of insight into how the brain works. I strongly recommend Consciousness Explained by Daniel Dennett.
The reason that Deep Blue is not AI is that brute force is at the heart of the machine's success. When I took that first (and last) computer class, in the late 70s, the focus even at that early stage was writing software that would "learn." One assignment was to write a tic-tac-toe program that would retain results and get better over time.
In fact, what happened in the 70s and 80s was the definition of AI went from HAL to expert systems. Expert systems are, of course, not intelligent in any ordinary sense. My brother the pathologist was telling me how well the medical Dragon (IBM's voice recognition system) was working for him as a dictation tool. He said that it accumulates phrases and sentences very efficiently. But that's not intelligence. That's storage and retrieval of a massive amount of data.
Again, brute force.
In The Emperor's New Mind, Roger Penrose points out that no matter how good your weather-simulating computer is, it never rains in there.
187. AytchMan - 6/22/2000 2:02:59 PM
Ask and ye shall receive. Thanks, kids.
188. JayAckroyd - 6/22/2000 2:03:47 PM
I agree with mandolin that what stopped the AI guys was a bad assumption--that the brain is like a computer.
It's not. Neurotransmitters and electrical potential shifts are continuous, analog processes. A Von Neumann machine is a paper strip with holes. So I claim:
1) You can't run a good brain simulation with a paper strip with holes, no matter how fast you push the paper strip through the reader.
2) You need more than two states to simulate continuous processes. You can't simulate a neuron with a two-state transistor, because it has continuous states
3) Brute force can only take you so far. Living systems are fundmentally different from strips of paper with holes.
We can't simulate the ganglia of a cockroach in a satisfactory mannner. We're not close to anything like AI. And Dennis Robinson owes me an ounce of gold. He was coworker involved with the MIT Athena project, and I bet him in 1981 that there would be no machine that would pass a Turing test by the year 2000.
189. AytchMan - 6/22/2000 2:31:05 PM
Jay et al--
>>The reason that Deep Blue is not AI...
That's my point. Was my assumption about the changing definition of AI over time wrong? I've read just enough about AI to be dangerous but not useful.
What seems to stick in my mind is the point I made above -- every time we get something that looks like the current definition of AI, we change the definition.
Amazingly enough, this process (if true) makes a weird kind of sense to me. Broadly defined, real he-man AI involves some kind of learning process. As soon as we program the lines of code for the newest "learning process", it becomes just that -- lines of code. There's no real learning (even within the obvious electronic limitations of the machine), it's just a bunch of instructions to the dumb computer.
That's why I think the development of "real" AI awaits the Eureka moment of some now-obscure mathematician. Comments?
190. JayAckroyd - 6/22/2000 2:38:34 PM
There has been a long running debate about what AI is and if it's possible. There's one guy who's name I'm forgetting who rejects Turing tests (Help me out here--he's the one who uses the man in a box processing Chinese characters as his metaphor.) The Penrose book I referred to earlier is a salvo against AI. Dennett believes it to be possible. The Turing Omnibus is a collection of relevant essays edited by AI proponents Doug Hoffstadter and Dennett.
The short answer I gave to the Deep Blue question obviously wasn't clear. Chess looked like an AI problem in the 60s because brute force wouldn't cut it. 60s and 70s era computers had to use some kind of learning because the decision tree was too deep. But we now have sufficient brute force to convert it to an algorithmic solution, not an intelligent one.
Along the way, we learned a lot about how people play chess. In the 50s, chess masters were assumed to be people who saw deeply into the decision tree--4 or 5 moves deep. That turns out not to be so. Chess players see the board in zones, and analyze only a few moves for their effect on the balance of power in those zones. "Control the center" is the human mantra. Deep Blues mantra is pick the tree branch with the maximum value.
191. JayAckroyd - 6/22/2000 2:43:29 PM
I don't think mathematics holds the answer. You need a whole new method of computing, one that isn't bound by an OR gate. Take DNA, for example. There are four fundamental states, four amino acids, to be manipulated. See, fundamental to intelligence is modeling. We (and other mammals, like my dog) are modeling all the time. The absence of the blind spot when you close one eye proves that when you look around, you're seeing heavily processed information. When you walk into a dark room and mistake a coat on a coat rack for an intruder, and immediately jump back, before finishing processing the object, you're running a model for what's in front of you.
This is very complicated processing, driven by associations and sequences of imagine response. Dennett's book talks about this very eloquently, and very clearly.
Computers don't do stuff like the above.
192. AytchMan - 6/22/2000 3:06:54 PM
>>I don't think mathematics holds the answer. You need a whole new method of computing, one that isn't bound by an OR gate
You may well be right about mathematics not hosting the initial breakthrough. But I'm inclined to think that, whatever the solution, we'll be able to quantify it somehow. If we can, then a computer will be able to crunch it, however slowly and artlessly. Obviously, the breakthrough could be some sort of bacteria that we don't even understand. Who knows.
>>Computers don't do stuff like the above.
Certainly not now. I guess I'm saying that I think the breakthrough will be of such a nature as to allow us to process it with future computing power. Again, who knows. Chalk it up to a blind belief in the march of technology.
193. Karl Northman - 6/22/2000 9:41:12 PM
Personal guess: we may be able to "process it with future computing power". However, that power will be either running a highly complex emulator, or be running in a hardware architecture that doesn't currently exist.
Haven't read this whole thing, but the most interesting thing recently I've read about AI (I was interested in AI, and reading neuropsychology in the early 70's, when I was a CS major) was the guy at MIT who was playing with "AI insects". I haven't heard anything about that for three-four years, it seems, but I had the sense that he was on to a promising direction.
And there won't be real AI until someone figures out a true analog (algorithmic or not), for the associative aspects of memory.
194. Indiana Jones - 6/22/2000 9:58:14 PM
Some AI links:
Artificial Life
Center for the Neural Basis of Computation
Cognitive Sciences Eprint Archive
195. JayAckroyd - 6/23/2000 8:21:19 AM
Front page story in the NYTimes on the politics of the missile defense system.
Highlights include the scientist on the commission that recommended treating the rogue state missile threat as significant saying the results were "cooked" and the revelation that the reason we have to violate the ABM treaty in the deployment is to protect the Aleutians from attack. The rest of the country can be covered from our current single ABM installation in New Mexico.
"But then computer simulations demonstrated that the system could not cover the westernmost islands in the Aleutians. And while only several thousand people lived on these remote reaches of Alaska, no one in the Clinton administration was about to tell Senator Ted Stevens, the powerful Alaska Republican who is chairman of the Senate Appropriations Subcommittee on Defense, that the world's most elaborate missile defense system would not protect some of his constituents. "
This is not about defense.
196. JayAckroyd - 6/23/2000 8:35:36 AM
Aytch--
Just because you can imagine, vividly, a technological innovation doesn't mean it's attainable. The Star Trek transporter is easy to imagine, but not gonna happen, ever. AI is closer to that than it is to controlled fusion.
One reason is that when people say "AI," they really mean "alive." That's why I referred to the cockroach earlier. The AIs of literature, Clarke's HAL, Heinlein's Mike Holmes, Gerrold's HARLIE, ST-TNG's Data are all alive.
Or take my dog. My dog isn't as intelligent as a human being, but still exhibits complex behavior in response to her environment. If she sees a chicken bone in the street on our trip from home to the office she doesn't leap for it. She walks past it diffidently, waits till I'm also past it, then turns around to pick it up behind my back.
We're right now at the insect simulation stage, as Karl Northman points out. When you start looking at the details of insect behavior--the fruit fly is especially well studied see Time Love and Memory --it's going to turn out to be a very intractable problem, IMO.
197. JayAckroyd - 6/23/2000 9:14:03 AM
195 That "New Mexico" should have been "North Dakota".
198. DocBrown - 6/23/2000 12:41:00 PM
AytchMan said:
Doc-- That's an awfully broad definition of AI, isn't it? Applying it, almost all software gets classified as AI.
The subject of my research was the failure of AI to live up to its promises. When I researched the serious scientific promises made in the 50s and 60s, I found things like checking spelling and grammar, optical character recognition, synthetic speech, chess playing, missile guidance, weather modeling, and natural language processing. The only place where AI is behind schedule is natural language processing. But every year the Turing contest machines get better and better.
If you want to talk about the non-serious promises that pulp science fiction writers made for AI, that is a different matter. Then we should also talk about personal rocket packs and disintegrator ray guns, too.
JayAckroyd, anyone who says that AI means "alive" falls into the non-serious category.
Yet you make a good point. Marvin Minsky is even more to blame than Bill Gates, since he weilded significant power in the AI community for the past couple of decades and he was obsessed with procedural logic. This form of AI is pretty much a dry well.
Of course hindsight is 20-20, especially in research. In the 70s NASA really thought it could operate the Space Shuttle cheaply, even though it had never been done before. A quarter century and billions of dollars later we now know how difficult it is to run a prolonged manned space program. Lessons like the expense of the Space Shuttle or the failure of procedural AI take many years to learn.
199. CalGal - 6/23/2000 12:43:39 PM
Great discussion. Jay, thanks for the article. I don't give a rats ass about the Aleutians, but the westernmost Hawaiian Islands wouldn't be covered, either, and that's just not gonna fly.
Is Consciousness Explained a good read for a know-nothing, or is technical knowledge required?
200. DocBrown - 6/23/2000 12:56:00 PM
JayAckroyd wrote:
When you walk into a dark room and mistake a coat on a coat rack for an intruder, and immediately jump back, before finishing processing the object, you're running a model for what's in front of you.
This is very complicated processing, driven by associations and sequences of imagine response. Dennett's book talks about this very eloquently, and very clearly.
Computers don't do stuff like the above.
Computers do half of it. An Optical Character Recognition algorithm, especailly a neural based one, can make exactly this sort of mistake. As far as the reaction based on anticipation and a model of the immediate future, there is no reason why a computer could not perform this type of operation. Computers can model the future, and react based on those models.
Practical AI needs only to demonstrate a tiny fraction of human behavior. I want an AI that can drive my car and sort my laundry, not one that can appreciate my poetry and seduce my wife.
Getting back to this thread, I would also like an AI to fight my battles. It need not have personal ambitions, hopes, and dreams. In fact, I would consider these to be liabilities.
201. AytchMan - 6/23/2000 1:07:33 PM
>>Just because you can imagine, vividly, a technological innovation doesn't mean it's attainable
Undoubtedly a true statement but I don't think it's a fair one. As a riposte, I would suggest that the absolute certainty that a Star Trek transporter is "not gonna happen, ever" is simply unsupportable. It may or may not occur. History is peppered with countless predictions of impossibility, some overturned in amazingly short order.
I'm not predicting that any specific solution will appear. And I'm sure that many things are unachievable in any reasonable time frame. Furthermore, I will also second Angel-5's opinion that a brute force technological attack on a given problem will not necessarily solve it. But we humanoids have a remarkable talent for discovering and applying seemingly unrelated formulas, methods and techniques.
You marshalled some very persuasive arguments against an AMS. But here (don't get mad), it seems that since you can't envision a solution, you're arguing that it's practically insoluble.
I guess our fundamental difference is that I look back through history and see amazing and accelerating technological progress fueled by unpredictable discoveries. Rose-colored? Sure. Accurate? Definitely.
I think we're arguing at two different levels here. You're presenting some very difficult technical obstacles and saying we're almost certainly not going to get there anytime soon. I'm saying we might.
202. mandolin - 6/23/2000 1:09:12 PM
It need not have personal ambitions, hopes, and dreams. In fact, I would consider these to be liabilities.
Actually the work of Antonio and Hannah Damasio, and of Lakoff and Johnson, suggests that something like ambitions, hopes, and dreams may be necessary in order to effectively choose paths of action.
203. JayAckroyd - 6/23/2000 1:12:30 PM
The Turing test is a very high hurdle. But, for better or for worse, that's what AI folk have set as the hurdle to clear. When Turing thought it up, he was trying to come up with a test that would indisputably show that the machine was intelligent. I think it is a true test. I also think we'll never see it passed.
BTW, while I think the Turing test is a sufficient condition for intelligence, I don't think it has to be a necessary condition. I've not heard of a lower hurdle that I would agree would indicate intelligence, but one may come along.
I found things like checking spelling and grammar, optical character recognition, synthetic speech, chess playing, missile guidance, weather modeling, and natural language processing.
These were issues at that time because they couldn't be solved except with intelligence. Now we use brute force to solve these problems when we do. (I don't know about you, but I'm not all that satisfied with the spell checking and grammar checking algorithms. Sure Word can look up and see if it finds a word in a dictionary, but that's pretty simplistic mechanics. It can't correct a "form" versus "from" error which is always evident in context.)
The machines solving these problems are not exhibiting intelligence. They're executing algorithms that require more processing power than was available to researchers in the 60s.
If you want to say that AI doesn't really involve intelligence, or the appearance of consciousness, then I claim you've shifted the ground. A robot in an auto plant performing a rote task previously performed by a human isn't exhibiting human behavior. Spell checking code in a word processor isn't exhibiting intelligence, by any reasonable definition of the word.
204. JayAckroyd - 6/23/2000 1:14:32 PM
CG-
In Consciousness Explained Dennett builds up his argument in a step by step fashion. Some of the ideas are difficult, but not because they require technical knowledge, but because they are somewhat counter-intuitive to how we are used to thinking about thought.
205. AytchMan - 6/23/2000 1:14:36 PM
To clarify, the subject of my 201 is AI, not AMS.
206. DocBrown - 6/23/2000 1:20:34 PM
AytchMan wrote:
As a riposte, I would suggest that the absolute certainty that a Star Trek transporter is "not gonna happen, ever" is simply unsupportable.
Actually, it is supportable. Read The Physics of Star Trek by Lawrence Krause. As shown on TV, the transporter violates the laws of physics. Warp drive and deflector shields might work, though.
207. JayAckroyd - 6/23/2000 1:25:26 PM
Aytch--
The transporter as depicted makes no sense. Why is there a transporter room? When you're off the ship, they can beam you around without a chamber. What happens when there's a dupe? The first couple of chapters of Nozick's Philosophical Explanations explores the myriad of problems that arise if you had such a device. But disintegrating people, transporting them with no propulsive device and then reintegrating them with no machinery at the other end? Nope.
On missile defense, I've said several things. One of them is that we are not close to achieving the mission that the defense has been dumbed down to. I've said that even if it were effective, there are other delivery mechanisms. I've also said that I think the underlying reason that we are very far away from Teller's sf devices is that we need an AI to make those, and we're a long way away from that. I've further noted that the reason we're a long way away is that there are, imo, inherent limitations involved with computers that can be represented by a strip of paper with holes in it.
I've also said that there is certainly a possibility of a computing breakthrough involving something other than an enormous number of two-state elements. That development awaits a genius of Turing's level. I think DNA offers possibilities--it bootstraps creatures from a single cell, and keeps them running for quite a while after it builds them.
208. DocBrown - 6/23/2000 1:26:21 PM
JayAckroyd, you are stepping onto shaky ground. Intelligence is like Free Will, we discuss it all the time but we don't really know what we mean.
By your definition, intelligence cannot be measured. I could program a computer to answer all the questions on an IQ test, but no one would call that "intelligence."
But if answering test questions is not intelligence, then what is? How can I ever prove to another human being that I am intelligent?
You seem to say that as soon as we understand how the answer is reached, we cannot call it intelligence. Your definition is interesting, but no better than a definition that says AT&T's telephone switching computer is "intelligent."
209. JayAckroyd - 6/23/2000 1:27:53 PM
Oh, AI. I'm on weaker ground there. It's hard to find a scientist who's not got an interest in it to support missile defense. The AI debate has proponents on both sides.
I've said I don't think we're gonna have an AI device on a Turing/Von Neumann computer. I'll stand by that.
210. AytchMan - 6/23/2000 1:29:16 PM
Doc--
Not being a Trekkie, I'm inclined to withdraw my comment. However, just to be cheeky: are you willing to state that the current laws of physics cannot possibly ever be revised under any circumstances? Ever?
211. JayAckroyd - 6/23/2000 1:32:30 PM
208
It's this problem that led Turing to propose his test. He was looking for a test that everybody would agree was clear evidence of intelligence.
I agree there is a certainly quality of "I know it when I see it" to the notion of intelligence.
I haven't said that when we've solved the problem, it automatically isn't intelligence. I've said the method by which the device solves the problem matters in assessing whether or not it was intelligent.
But I'd be happy to read a definition of intelligence you'd like to propose.
212. JayAckroyd - 6/23/2000 1:37:30 PM
210
That's an assumption. The empirical question of the stability of the laws of physics is not answerable, other than by getting up the next day and looking.
But it's been a pretty solid assumption.
There's a lot of fuss lately about the parameters of the standard models of quantum mechanics and cosmology. In Life of the Cosmos Lee Smolin suggests that there may be a large number of universes, with the parameters set at different values. We're in one that happens to have a lot of black holes, and therefore people exist.
(Of course, it takes several hundred pages to make those connections.) The key point is that the values that parameters take on in the laws of physics seem to be very delicately balanced. Small changes to a small number of parameters, and there are no people.
213. AytchMan - 6/23/2000 1:41:31 PM
>>But I'd be happy to read a definition of intelligence you'd like to propose.
Good suggestion. We certainly need some common ground here. While I can't write a definition, I think it should be along the lines of a Turing test (fooling a moderately knowledgeable human) with a modest ability to learn (a smart expert system or even a brilliant computer game strategist).
214. JayAckroyd - 6/23/2000 1:50:12 PM
Here's Merriam_Webster:
a (1) : the ability to learn or understand or to deal with new or trying situations : REASON; also : the skilled use of reason (2) : the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria (as tests)
Useless, really. "knowledge" "skill" "reason" "think abstractly" are all somewhat circular. Doc might claim the cruise missile's ability to adjust its target represents knowledge, skill and reasoning, while I would deny that.
Does a collection of chess positions stored on a hard drive represent "knowledge?" Does that make a floppy drive with a single chess position "knowledgable"?
My point is that intelligence, in the general sense, connotes both information, manipulation of information and understanding what you're doing. It's the last part that's "alive", and it's what's missing from Doc's examples. Those missiles don't "know" where they're going, and it isn't raining in the weather simulator.
215. JayAckroyd - 6/23/2000 1:51:01 PM
Here's MW on AI:
1 : the capability of a machine to imitate intelligent human behavior
2 : a branch of computer science dealing with the simulation of intelligent behavior in computers
216. JayAckroyd - 6/23/2000 1:56:53 PM
On tests of intelligence, there are two kinds of tests in this world. Objective tests, like SATs and tests of judgement, like Supreme Court cases or oral examinations for a doctorate.
I think the question of intelligence is a test of judgment, like a Turing test, and not a test of accomplishment, like an SAT test.
217. AytchMan - 6/23/2000 2:06:39 PM
For the purpose of discussion in this thread, I'd like to eliminate the "alive" part of a definition of AI. Whether it's pattern recognition for an AMS or a more general strategic decision-making package for some other military purpose, I don't think it needs actual or simulated consciousness. An AI program may indeed simulate consciousness inherently (because that's part of the future technological breakthrough) but it's not required in and of itself (I suggest).
218. JayAckroyd - 6/23/2000 2:13:15 PM
I can't agree to that. I think it's inherent to the SDI problem. It's not inherent to the one puny Korean missile problem. Somewhere in between the system has to acquire judgment.
219. mandolin - 6/23/2000 2:14:46 PM
about Message # 217I think that's a bad assumption, Aytchman -- I don't have time to explain it now, but I'll do it this weekend, and also provide lonks to relevant posts in the Ethics thread (formerly Religion and Philosophy)
Dennett is very good, and very thorough, but I think he misses some important points about embodiment.
220. mandolin - 6/23/2000 2:15:26 PM
wonder what a "lonk" is ...
221. JayAckroyd - 6/23/2000 2:16:02 PM
We should take that over to the slow thread, mandolin. I don't know what you mean by embodiment.
222. AytchMan - 6/23/2000 2:37:27 PM
jay and mondolin--
Being a newcomer, I've apparently missed something important in Philosophy or some other thread so I'll await mandolin's lonks. I'm afeared we're about to go sharply metaphysical.
223. JayAckroyd - 6/23/2000 2:47:45 PM
I haven't read the stuff mandolin refers to either.
224. AytchMan - 6/23/2000 3:02:03 PM
jay 212--
That's an assumption. The empirical question of the stability of the laws of physics is not answerable, other than by getting up the next day and looking.
But it's been a pretty solid assumption.
Quite true and it goes to the heart of our discussion. To oversimplify with abandon, you look at the current technological environment, see almost insurmountable difficulties based on current constraints and hold little hope of overcoming them anytime soon. I look at the same environment, see huge difficulties -- but not almost insurmountable because some constraints will be serendipitously broken -- and believe there's a small but significant chance of success.
Again, this is not a defense of AMS or any specific concept; rather, it's a general philosophy of technological development.
By the way, I think that the outlook to which one subscribes has major implications for future weapons development and war in general.
225. JayAckroyd - 6/23/2000 3:12:11 PM
I understand this to be your point of view. I happen to think it paints with too broad a brush. There is something profoundly different among:
1. Fuel cells
2. Fusion power
3. Artificial intelligence
4. Transporter beams
Problems of type 1 are awaiting economical engineering solutions. Problems of type 2 are awaiting a breakthrough idea.
Problems of type 3 are awaiting a paradigm shift.
Problems of type 4 are impossible.
Where there is disagreement on AI is whether it is a type 2 or type 3 problem.
226. AytchMan - 6/23/2000 3:28:28 PM
I understand this to be your point of view. I happen to think it paints with too broad a brush.
An honorable discussion. You have the last word.
As I stated somewhere upthread, I'm more interested in understanding people's views (and how they came to them) than in persuading them otherwise. This stems from too many heavily-charged late-night discussions back in the '70's.
227. AytchMan - 6/23/2000 3:32:04 PM
jay--
Sorry if that sounded abrupt -- it's been an excellent discussion. And I'm still interested in what mandolin lonks us to. I also have a couple of other questions to pose.
228. DocBrown - 6/23/2000 3:48:54 PM
JayAckroyd, how can any definition of Artificial Intelligence by a type 3 problem?
Unless you believe that "intelligence" requires a "soul" or other supernatural entity, then you must conceed that intelligence obeys the laws of nature. Perhaps there are laws which we do not yet understand, but we expect those laws to obey the laws of mathematics.
As long as a process obeys the laws of mathematics, it may be simulated in a computer. The problem may be complex, difficult, or intractable, but demonstrating the simulation would not require a paradigm shift.
I do admit that a paradigm shift would help. Humans need to acknowledge that their brains are cause-and-effect machines.
229. AytchMan - 6/23/2000 4:12:36 PM
In which areas of warfare has AI had the greatest effect/benefit to date? In which areas of warfare does AI have the greatest potential benefit in the near-future (say, 20 years)?
For these questions, AI is herein defined as current and near-future software concepts and practices, not the breakthrough stuff we've just been duking it out over.
230. mandolin - 6/23/2000 4:14:25 PM
Doc -- you don't have to believe in a soul to believe that the "laws" of mathematics are invented, not discovered, and invented by a particular of kind of brain in a particular type of body. They are not features of the universe, but features of the interaction between our particular bodies, with their particular evolutionary history, and the environments we have been exposed to.
Because our reasoning is constrained by that evolutionary history it is not merely subjective or arbitrary, but neither is it objective in the sense required by your last few posts.
If we build an AI system -- and I think we will be able to -- it will not pass the Turing test. It will be completely alien.
231. DocBrown - 6/23/2000 4:21:54 PM
AytchMan - I would say the answer to your queston is espionage and surveillance. A lot of important information can be gleaned from a program that extracted meaning out of telephone & radio conversations, and various photographs or video footage.
232. AytchMan - 6/23/2000 4:28:06 PM
Doc--
Excellent answer. That would be my first choice as well. Another area I was considering was simulations: flight training and combat sims.
233. AytchMan - 6/23/2000 4:31:59 PM
mandolin--
If we build an AI system -- and I think we will be able to -- it will not pass the Turing test. It will be completely alien
Provocative. More details.
234. Cellar Door - 6/23/2000 8:15:08 PM
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
Pinocchio Bore: Guilty as hell
235. JayAckroyd - 6/24/2000 10:04:44 AM
228
When I say that it requires a paradigm shift, I mean that I don't think AI is achievable on Von Neumann/Turing machines.
When you talk about the soul, I think you're being distracting. That's an even harder to define word than intelligence. OTOH, I do think that a keynote of intelligence in everyday parlance is self-awareness and/or consciousness. My dog is conscious and self-aware, but I have no idea whether she has a soul.
So maybe what's going on is that the AI folks want to keep the word "intelligence" but not its attributes. That may be the source of confusion in discussions like this. You certainly make me suspect that to be the case when you cite spellcheckers and chess playing computers as examples of AI.
236. JayAckroyd - 6/24/2000 10:09:08 AM
I do admit that a paradigm shift would help. Humans need to acknowledge that their brains are cause-and-effect machines.
I'm not sure what this means.
If you mean that you believe that the original AI belief that computers could model brains because they are necessarily just devices, I think that three decades of research points to this belief as simplistic and naive. Brain function, the mind, consciousness are issues of intense debate and a great deal of uncertainty. AI research has played a key role in enriching our knowledge of the brain and the mind, but it has led to more questions than answers.
237. iiibbb - 6/24/2000 10:25:02 AM
Message # 183 et al.
You have to wonder though... maybe there's new 'diversion' in store for AI. As DOC said, they did not anticpate the direction of computer technology moving to GUI's.
However, now we are now seeing new technologies in the form of implants and other biological/technological meldings. My uncle recently had one of those cochlear implants put in his ear to deal with his 95% deafness.
Well... maybe at some point we're going to see technologies move in the directions similar to that movie Johny Mnemonics. Try robocop, or many popular Animae themes, where men join machine. I doubt that this kind of technological innovation is going to be available to everyone... so we may wind up with a whole new class of computers that are part man and machine...
Potentially scary, isn't it?
238. iiibbb - 6/24/2000 10:29:05 AM
Message # 210 Message # 212
Didn't somebody just break the speed of light?...
The implications of that could really throw a wrench in relativity.
239. CalGal - 6/24/2000 10:30:50 AM
Have we built a system that could construct an If...then for a situation that the programmer didn't anticipate? And by "didn't anticipate", I mean also that there was no "If it's none of the above, then do this".
For example, does the Turing machine make decisions or associations in ways that can't be anticipated?
I realize that the terminology used is insultingly simple for AI, but I imagine you get the idea.
240. iiibbb - 6/24/2000 10:35:38 AM
Fuzzly logic does that in a remedial sense Cal...
Vacuum cleaners that are able to compensate for various naps in carpets. Cars that 'adapt' themselves to the owners driving style.
Fuzzy logic is an exceedingly interesting concept, but I am only equipped with a superficial knowledge of how it works. I guess it's somewhere between AI and a machine that has been tuned for a specific function.
241. JayAckroyd - 6/24/2000 2:44:54 PM
The speed of light thing was a technical gimmick, not a true instance of exceeding the speed of light.
CG--Please don't say Turing machine when you mean "a machine that passes a Turing test." A Turing machine, as discussed in Cryptonomicon, is a computer that uses binary states to encode information.
And yes, the idea of an AI device is that it could respond correctly to novel circumstances. Turing proposed his conversational test because human conversation will likely present frequent novel circumstances to a computer.
242. JayAckroyd - 6/24/2000 2:45:58 PM
237
I think man/machine combinations would be a paradigm shift.
243. CalGal - 6/24/2000 2:52:04 PM
Jay,
Sorry. Okay, "a machine that could pass a Turing test". And so you are saying that none have been developed like that?
3i3b,
But vacuums can only adapt to the extent that their programmers anticipated the need.
244. iiibbb - 6/24/2000 3:30:20 PM
Not exactly... They define the boundary conditions a machine may opperate, but they don't program for every contengency... or not every specific contengency. Instead they program the optimum opperating parameters, and then provide the machine with rudamentary feedback mechanisms and controls so that the machine in many ways is independently evaluating the conditions and controlling its own performance.
I think the better example are the cars that actually adapt and optimizes the engine performace based on the driving style of the operator.
I realize it's limited... but the programers aren't using an if this do that kind of programing.
There are better examples... I just can't think of them right this second...
245. JayAckroyd - 6/26/2000 9:02:49 AM
243
No. We are not close. You may remember that I posted, quite a while ago, winners in Turing test type contests. The computers' efforts were laughable.
iiibbb--
My work with fuzzy random variables is a riff on the idea of fuzzy logic. A drug company I was developing project management software for wanted to introduce greater rigor to the process of continuing or stopping projects. The system we designed extended the idea of a Bayesian decision tree (essentially a yes-no network diagram, with probabilities assigned to each possibility on each node). The wrinkle that fuzzy random variables introduce is to allow the user to specify the shape of the probability distribution function instead of merely making a point estimate. He or she can express uncertainty by spreading the distribution out. Or the user can put the probability in the tails instead of the center, if the result of an event is gonna be either really good or really bad. It turns out that you can treat these density functions much as you can treat well-formed probability density functions.
Fuzzy logic in control systems makes blurry the point at which an action is taken. You might set a thermostat to 70 degrees, but allow a variance of 3 degrees either way, and build in some kind of feedback mechanism with a decision rule that allows the control system to find a fuel minimizing on/off strategy within that range. For example, it might be more efficient to let the temp rise to 73, cut in the cooling system, bring it down to 67, cut it off and loop. Fuzzy decision systems let boundaries and threshholds become ranges within which predefined decision rules can be updated by data.
246. JayAckroyd - 6/26/2000 9:10:11 AM
But that's not AI.
However, AI systems will include judgment--and the first step in judgment is respond to feedback with a decision. In the instances cited here, the device has no ability to construct a new decision rule. It can only operate within the assigned decision rules. What AI folks have hoped is that by creating a lot of fluid, interacting decision rules in an architecture that permits circuit rerouting (such as neural nets), judgment beyond that which has been preprogrammed in will arise.
In the book by Dennett I cited earlier, it is his contention that the way human consciousness works is that there are constant mini-evolutionary battles for the dominant interpretation of a stimulus or the creation of a thought. (That capsule summary does not do justice to the book or the argument.) I find that a very persuasive story, one that explains why simply increasing complexity of Von Neumann devices hasn't worked to improve intelligence.
In areas where we've used evolutionary techniques, such as the game of Life or programs competing in multi-period Prisoner's Dilemma type games, we've learned quite a bit.
247. DocBrown - 6/26/2000 4:28:57 PM
I the very very far out definition of AI that JayAckroyd uses, my favorite example is the work ofNavlab at Carnegie Mellon.
The computer drives the vehicle. It must constantly make judgements based on real time inputs. It must plan the immediate future, and try to anticipate what other drivers will do.
The judgements, plans, and anticipations are not as good as a human's, but the computer does make them. The day when cars drive themselves is still decades away, but the Navlab team did some great work relevant to this thread. In order to do a lot of useful work (stuff that would be practical on the highway or the battlefield) the AI does need to respond to novel circumstances (like being cut off on the highway). But it does not need the creativity to invent new responses. All it needs to do is use the responses it already knows for the novel circumstances.
If my AI gets cut off on the freeway it only needs to execute an ordinary braking or lane change maneuver. If it gets attacked by a new type of weapon it only needs to execute a normal counter attack.
There is no question that more sophistocated forms of cognition and reasoning, those of a human for example, might be nice. But they are certainly not needed to create some very useful military and consumer AI products.
248. JayAckroyd - 6/26/2000 7:37:14 PM
Doc,
My so-called "very very far out" definition of an AI is passing a Turing test. The Turing test is one that was proposed by an AI advocate and has, IME, been continuously supported by AI advocates. In this discussion, I've frequently said that I don't think Van Neumann/Turing machines can pass a Turing test. That is, I've said that they cannot pass the test that AI folk have proposed. I don't see what is far out about that assertion.
I've also noted that I consider Turing test a sufficient, but not necessary test for AI. I suspect there is a lower hurdle that most people would agree is a test of intelligence. I've urged you to propose a less restrictive test.
"The judgements, plans, and anticipations" you talk about are none of the above. They are algorithmic responses to predefined conditions. They are brute force solutions to problems that can be calculated given sufficient time. My (unchallenged) point is that the method matters in any test short of a Turing test(until you offer an alternative). You don't get AI by running the processor really fast against a predefined algorithm, or a collection of predefined decision rules. You get AI from judgment in response to a novel situation.
Again, I've made these points, and you've chosen not to refute them other than to make ad hominem comments. I'd welcome some more substantive responses.
249. DocBrown - 6/27/2000 12:49:13 PM
JayAckroyd, I did give a substative response in 208. I said: "Your definition is interesting, but no better than a definition that says AT&T's telephone switching computer is "intelligent."
You have arbitrarily set your definition of Artificial Intelligence as something that somehow involves judgement. How nice for you. Do you also define beauty as something that somehow involves roses?
Artificial Intelligence is like beauty, in that there have been many different opinions about both. I can safely embrace opinions that say a 1963 Corvette has beauty or a telephone switching algorithm has Artificial Intelligence. You are free to disagree with either opinion, but insisting that I must prove or even argue in support of my opinions is a waste of time.
Nonetheless, here is my definition of Artificial Intelligence: anything made by man that makes useful decisions.
Under my definition, both Navlab and a centrifugal steam pressure regulator have AI.
250. JayAckroyd - 6/28/2000 9:07:54 AM
A decision implies understanding and judgment. Is it a "decision" for a "q
251. JayAckroyd - 6/28/2000 9:11:09 AM
A decision implies understanding and judgment. Is it a "decision" for a "q" to appear when I strike the key? Does a car make a "decision" when it deploys or refrains from deploying an airbag?
And, again, using a Turing test as an AI standard is not arbitrary. It's hallowed.
252. DocBrown - 6/28/2000 1:43:56 PM
Sure, a Turing test is an interesting AI standard. But I have taken several classes categorized as "Artificial Intelligence" that had absolutely nothing to do with natural language processing. Your Turingcentric view is that AI involves nothing but natural language processing. I find this to be a very limiting redefinition of the term.
When you press a key and a 'q' appears on your monitor, that is cause-and-effect logic. If you believe that the human brain uses anything more than really, really, really complex cause-and-effect logic please tell me about it. I will be happy to read about your belief.
Neural networks and genetic algorithms can simulate some very human-like decision making, but they are still cause-and-effect. I classify both as AI.
253. JayAckroyd - 6/28/2000 6:10:20 PM
The notion that the brain works by cause and effect logic is controversial. Since we know for sure that at lowest level, nature doesn't work by cause and effect, why we would we make that the least hypothesis?
As far as "redefinition"s go, it's the AI folk who've been on a steady path of dumbing down the definition as they fail to meet criteria they originally intended to reach. I don't doubt you've taken "AI" courses that discuss brute force methods for various processing problems, or that explore neural net dead ends. It's not AI if it doesn't work,or doesn't involve, in your words, decision-making.
BTW, you seem to concede my point that my laptop producing a q is as much an act of intelligence as the systems you cite, because the result from a cause and effect relationship. Does "intelligence" mean anything beyond that?
254. DocBrown - 6/29/2000 3:18:40 PM
Jay, can you please explain how we know at the lowest level that nature does not work by cause and effect?
Not that it would change my answer. If nature has a process that is not based on cause-and-effect logic then a "strong" AI would simply need to harness that process.
Frankly, I believe that words like intelligence, consciousness, free will, understanding, meaning, and idea are practically useless. We use these words without "knowing" what they mean. We cannot explain how these processes work, yet when we use those words we hope that the person to whom we are speaking "understands" us.
When I say that I understand algebra, do I mean the same thing as when you say you understand algebra? How can we possibly know?
Interpersonal communication would be much more fool-proof if we described our own mental processes in terms of logic and algorithms, instead of "understanding" and "knowing."
255. DocBrown - 6/29/2000 3:21:20 PM
So in answer to your question, "intelligence" does not mean anything beyond really complex cause-and-effect relationships.
256. JayAckroyd - 6/29/2000 6:03:23 PM
So then the appearance of the q when I type it is artificial intelligence. That result relies on a very long sequence of cause and effect events, at the bit level.
So everything is intelligent. A bacterium, by this definition, is intelligent.
257. CalGal - 6/29/2000 6:15:42 PM
Please, god, I sense a Free Will conversation coming on. If we stop it anyway, does that mean it was meant to stop or that we do, after all, have free will?
258. JayAckroyd - 6/30/2000 8:28:21 AM
Jay, can you please explain how we know at the lowest level that nature does not work by cause and effect?
Because God plays dice. Quantum mechanics is probabalistic, not strictly determined by cause and effect.
Penrose argues in The Emperor's New Mind that consciousness may be tied to quantum events. I don't believe that, but do think that there is a point where cause and effect become too complex to model. While in principle a coin toss could be analyzed and predicted, it, like all behavior we consider random, is too complex to be analyzed.
259. JayAckroyd - 6/30/2000 8:33:19 AM
Doc's use of the word "decide" indicates an act of will. Studies of the genome and evopsych studies are making it clear that some behavior is wired (like circadian rhythms), some behavior results from a combination of environment and heredity, and some is entirely in your control.
The free will question has been answered. You have some control over your destiny, but your choices are constrained by heredity, environment and culture.
260. JayAckroyd - 6/30/2000 9:39:15 AM
Another NYtimes story about missile defense.
Alternatives such as hitting (big, slow) missiles before they release(tiny, fast) warheads are discussed. The mechanics of the project are clearer in this discussion than any other I'd read. For example, the warhead killers are positioned by operators:
The goal of this network is not only to aid kill vehicles but to push the defensive battle as close as possible to enemy territory so as to give military officers time to fire more than one interceptor at a specific warhead, raising the odds of success.
The tactic is known as shoot-look-shoot. In theory, antimissile officials say, three or more hits might be attempted against a given warhead.
General Kadish, director of the Pentagon's Ballistic Missile Defense Organization, told Congress in February that if interceptors are 80 percent dependable, two tries will provide 96 percent confidence and three will give 99 percent assurance of a successful kill.
The device itself works by ramming the warhead:
The exoatmospheric kill vehicle is the star of the Clinton plan, a 130-pound wonder just 54 inches long. It is made by Raytheon at a plant in Tucson. In space, it would guide itself toward the target, its tiny computer analyzing sensor readings and firing thrusters.
Its big challenge is to disregard the decoys amid the nuclear warheads, which the sensor tracks through their heat rays and sees as twinkling points of light, like stars.
Zipping along at about two miles per second, the kill vehicle is to slam into the nuclear warhead in space and demolish it by force of impact.
261. JayAckroyd - 6/30/2000 9:42:06 AM
Interpersonal communication would be much more fool-proof if we described our own mental processes in terms of logic and algorithms, instead of "understanding" and "knowing."
Doc, thirty years of research has demonstrated that this algorithmic approach to describing human mental processes doesn't work. AI researchers assumed that intelligence could be reduced to a series of logical steps, but that has proven not to be the case.
262. JayAckroyd - 6/30/2000 9:59:15 AM
I also didn't know this:
A shadow that hangs over all these antimissile proposals is Safeguard, the only weapon the United States ever built against long-range missiles. Switched on in North Dakota in October 1975, it was meant to shield nearby missile fields from a disarming first strike.
Its plug was pulled as military strategists concluded that Moscow could overwhelm it with a rush of new warheads then being deployed. In all, the $25 billion weapon system lasted just 133 days.
Today, its mothballed base is a wilderness of broken asphalt, empty silos, tumbled buildings and a chapel, eerily intact. It all stands as a monument to what can go awry in antimissile plans.
263. DocBrown - 7/6/2000 2:54:47 PM
I asked:
Jay, can you please explain how we know at the lowest level that nature does not work by cause and effect?
JayAckroyd responded:
Because God plays dice. Quantum mechanics is probabalistic, not strictly determined by cause and effect.
So what? I never said that either the brain or the Universe was deterministic. What I said was that these things operated by cause and effect.
Penrose argues in The Emperor's New Mind that consciousness may be tied to quantum events.
Again, so what? We can simulate quantum events with random number generators. True, the random number generators will be deterministic, but we humans cannot tell the difference. QM may make it impossible to create a machine that duplicates the brain of a certain individual human, but it is still possible to make a machine that duplicates certain functions of the brain.
For this thread, the biggest bang for buck return probably comes from pattern classification.
264. DocBrown - 7/6/2000 2:58:59 PM
For example, the shoot-look-shoot tactic JayAckroyd described above could greatly benefit from an AI that recognized missile launches that were threatening as they left the ground. I would be very impressed by such AI. No need for this system to pass a Turing test.
265. JayAckroyd - 7/6/2000 11:29:14 PM
Yeah, but that's not part of the program. The only role any recognition system is playing in this system is picking out a target after a human has aimed it.
Time Magazine reports that the simulated warhead will be broadcasting its position using a GPS signal. The testers say they won't use the GPS system to targe the warhead unless they have to.
And, Doc, I still have no clue what constitutes an AI for you. Or intelligence, for that matter.
266. angel-five - 7/7/2000 1:04:18 AM
I never said that either the brain or the Universe was
deterministic. What I said was that these things operated by cause
and effect.
Doc, just so we all know where you're coming from, would you illustrate a pure cause and effect system that isn't deterministic?
Again, so what? We can simulate quantum events with random
number generators.
By this logic the giant walk-through heart at the Franklin Institute ought to beat. It doesn't.
True, the random number generators will be deterministic, but we humans cannot tell the difference. QM may make it impossible to create a machine that duplicates the brain of a certain individual human, but it is still possible to make a machine that duplicates certain functions of the brain.
That doesn't follow. What you should have said is "QM may make it impossible to create a machine that duplicates the brain of a certain individual human, but it is still possible to make a machine that models what a brain would do if it didn't have true QM chaos to contend with." Which of course might be interesting in that it might narrow down the possible effects that QM uncertainty has on a central nervous system, but then again, it might not, either.
267. angel-five - 7/7/2000 1:14:59 AM
I should like to say at this point that it's rather useless to model the brain without an equally valid model of the body. Mind and body are not really separable phenomena and their interdependent function is sufficiently complex --it's not just a matter of blood flow and the odd hormone here and there, but gigantic chaotic systems of feedback and response-- that to model one without the other and expect good data is sort of like analyzing trends in herbivore population without accounting for disease or predation.
No one can really say where the seat of consciousness is, or where its cradle may have lain. You can adopt a mystical approach, an agnostic approach, or the ardently reductionist approach that Doc espouses when he starts going on like the geeky love child of Doc from Back to the Future and Ming the Merciless -- but the result isn't even an adequate thought experiment, it's like Adler, Newton and Joseph Kennedy trying to imagine the interior of a singularity after four glasses of wine and a Nova special on PBS. We simply know too little.
We humans have a notable brain fetish and so when it comes time to wonder about the roots of consciousness that's generally where we start thinking. And it's understandable, because the brain's function certainly is responsible for the behavior which we deem to be the observable articulation of consciousness --but that doesn't mean that it's the seat of consciousness itself, or that its actions constitute consciousness.
268. angel-five - 7/7/2000 1:31:38 AM
Getting back to the point:
I will agree with Doc's notion that sufficiently specialized pattern recognition systems will serve just fine for missile defense systems. Indeed, you'd think that 'conscious' AI would be a detriment -- I don't know quite why we need AI for a missile defense system. I suppose it's all how you define AI.
269. mandolin - 7/7/2000 1:08:28 PM
I aplologize for disappearing just when I promised I'd post some info -- the last week at my former job turned into a life-gobbler. New job syarrts next week, and I'm not really sure what will happen, but I expect to be freeer.
I haven't had time to get caught up here, yet, but when I do I expect to get back to reasonable participation.
270. DocBrown - 7/7/2000 2:10:06 PM
Jay said:
Yeah, but that's not part of the program. The only role any recognition system is playing in this system is picking out a target after a human has aimed it.
Then they do not have a very practical shoot-look-shoot system. To make this system really useful, they need to detect missiles close to launch. Whoever runs this syetem must be looking at the entire surface of the Earth at all times, and recognize launches when they see them. In this case, a really sophistocated pattern recognition system would be the cat's meow. Unless the Pentagon has made a neural net breakthrough that is still classified, this technology does not exist yet.
However, I believe that such sophistocated pattern recognition technology is possible. It would not need any quantum uncertainty to function.
271. DocBrown - 7/7/2000 2:15:06 PM
. . . the geeky love child of Doc from Back to the Future and Ming the Merciless . . .
Not there is an image for you! I cannot wait to repeat that one to my wife. If I time it right I can probably get a jet of wine to shoot out of her nose. Perhaps it could intercept an incoming missile.
272. DocBrown - 7/7/2000 2:26:12 PM
Angel asked:
Doc, just so we all know where you're coming from, would you illustrate a pure cause and effect system that isn't deterministic?
I did already. The entire telephone system of the United States of America (or the world) is a great example. Its behavior is completely cause and effect. Yet you cannot possibly predict the state of the system one second in the future. It is not deterministic because you cannot predict the behavior of all the humans with all the telephones all over the system.
The brain can be the same way. It follows the rules of cause and effect, yet quantum mechanics may mean that the brain is not deterministic.
273. angel-five - 7/7/2000 3:14:24 PM
The entire telephone system of the US is entirely deterministic, Doc. It's just when you start letting humans use it that it becomes otherwise. Try again, and good luck with the jet of wine.
274. JayAckroyd - 7/7/2000 6:00:48 PM
A-5,
the reason I've come to believe in the coure of this discussion that you need an AI is that I don't think humans can process the ballistic data fast enough to shoot. And I don't think the warhead hunters cannot make decisions about target acquistion without AI capabilities.
I guess I haven't made the latter case as well as I might. Let me think about it and post another attempt.
275. angel-five - 7/7/2000 6:48:10 PM
Just so the position is clear: Cause and effect implies determinism. Doc's example is observably neither cause and effect nor deterministic because he cites human interaction as the key to the function of the system and humans are, strictly speaking, neither deterministic nor are they rigid cause and effect systems.
And the phone system isn't either, judging by some of the more interesting errors I've had in placing calls and being billed for them.
I'd like someone to demonstrate a cause and effect system that isn't deterministic.
276. DocBrown - 7/10/2000 9:15:40 AM
Angel, cause and effect absolutely does not imply determinism!
You must get past this misconception before you will be able to comprehend what I am saying.
I'd like someone to demonstrate a cause and effect system that isn't deterministic.
I am afreaid that you are straying toward a mystical, supernatural definition of determinism, but I shall make one more attempt.
Consider the weather.
The Earth's weather is a chaotic system. It is impossible to predict its behavior precisely, and it will probably always be so. The best we can do is to specify a range of behaviors. The further we look into the future the wider we must expand that range.
Yet at the same time the weather is a cause-and-effect system. Yesterday the wind blew here, the sun shined there, a little evaporation, a little sublimation, and all of this combined to cause a summer shower on my newly washed car.
The weather is not random, it is a complex system of chaotic causes and effects. Much like the brain.
277. angel-five - 7/11/2000 1:27:39 AM
Doc, I don't know what definition you're using for deterministic. The one I have in mind is neither supernatural nor hard to grasp.
Main Entry: de·ter·min·ism
Pronunciation: di-'t&r-m&-"ni-z&m, dE-
Function: noun
Date: 1846
1 a : a theory or doctrine that acts of the will, occurrences in nature, or social or psychological phenomena are causally determined by preceding events or natural laws
Nothing complex about that. Cause and effect systems are deterministic. Whether or not they're predictable by human beings with limited information is of course another matter entirely, but that's got nothing to do with determinism.
You propose planetary weather patterns as an example of chaos (which is just another way of saying that the mathematics of it is horribly complex). Well and good; weather is chaotic in function. Yet, you say, it's also cause and effect -- that changes in the weather are effects resulting from an aggregation of minor causes. These causes cannot be exactly predicted due to the sheer size of the system and the impossibility of identifying the location and momentum of every single molecule in the system at any one given time, but they are, according to you, causes which lead to effects in a mathematically and physically sound fashion. Well and good; that's how I understand weather to work as well.
Where you lose me is somewhere between your acknowledgement that weather operates by the principles of aggregate cause and effect, and your insistence that because we can't predict the state of the system at any one given time it must not be deterministic. If your insistence reduces to quantum uncertainty and the resulting Copenhagen interpretation ... well, you are no longer discussing a system which you can demonstrate causality in, and as a result your statements meander into 'I Don't Know What I'm Talking About, Somebody Shut Me Up' land.
278. DocBrown - 7/11/2000 10:10:47 AM
Ah hah! Yes, Angel, we were using different definitions of the word deterministic. The I learned to use the word as (essentially) the opposite of stochastic.
Deterministic: a mathematical model that ignores randomness when making a prediction, assuming it to be unimportant to the decision to be made.
Stochastic: a mathematical model that explicitly tries to capture randomness when making a prediction.
Introduction to Simulaton Using SIMAN, McGraw Hill, 1990, p 6.
Given the same starting state, a deterministic system must always reach the same new state while a stochastic system may reach different states. The behavior of deterministic system may be predicted precisely while the behavior of a stochastic system may only be predicted within a range.
As I said, I cannot fully demonstrate the causality in a complex and chaotic system, such as the weather or the brain. But claiming this as proof that the system must violate causality is another matter entirely. I am not sure if you are making such a claim, but it seems as though you are:
. . . humans are, strictly speaking, neither deterministic nor are they rigid cause and effect systems.
279. angel-five - 7/11/2000 11:47:14 AM
Given the same starting state, a deterministic system must always
reach the same new state while a stochastic system may reach
different states. The behavior of deterministic system may be
predicted precisely while the behavior of a stochastic system may only
be predicted within a range.
No, no, not at all. Stepping away for a moment from Heisenberg... the notion that a deterministic system can be predicted precisely ignores the fact that some deterministic systems are complex enough that the act of measurement disrupts them. As for 'random'...
do you mean 'random' or 'chaotic'? Because weather systems aren't demonstrably random. If identical weather systems receive identical input there's no reason to believe that they'll react in anything but an identical fashion... which, of course, means that they're deterministic by both your and my definitions.
You're doing this curious top-down labelling thing where you map your reality based upon the names of the models you choose to represent it... and then expect the reality to conform to the rules of that model. The model is, however, not the system.
280. angel-five - 7/11/2000 11:53:46 AM
How do you differentiate the concept of causeless from the concept
of supernatural?
It's pretty simple. I don't start by assuming that causality is the be all and end of all of nature.
'Causeless', like most of the other terms that get bandied about in these conversations, are laden with accidental meaning. We think of systems as causal or causeless based upon how we percieve their function. Yet a 'causal' system doesn't necessarily operate by cause and effect (it's just a system in which we believe we've identified a cause and effect) and a 'causeless' system isn't necessarily free of cause and effect (it's just a system in which we can not match driving causes to generated effects).
We see the terms shaping our understanding of the systems in question but really they've got much more to do with our powers of observation than they do with the systems they describe.
281. Indiana Jones - 7/11/2000 11:54:19 AM
If identical weather systems receive identical input there's no reason to believe that they'll react in anything but an identical fashion... which, of course, means that they're deterministic by both your and my definitions.
But doesn't your definition of "deterministic" say...
The notion that a deterministic system can be predicted precisely ignores the fact that some deterministic systems are complex enough that the act of measurement disrupts them.
?
(Which I find to be gobbledygook. If a measurement affects a system, then the measurement is itself input. If the system is deterministic, then the result of this input will always be the same.)
282. DocBrown - 7/11/2000 12:06:12 PM
The definitions of deterministic and stochastic both came from the mathematical modeling textbook I cited above. They included the word random.
I shall henceforth write these messages in the context of mathematical modeling, which I know fairly well and consider to be relevant to the discussion of Artificial Intelligence. I agree that when my writing has flirted with the context of physics, which I consider less relevant to AI, it has been ambiguous.
In this context random really means random, as in a random number generator. That is certainly not chaotic in the same sense as quantum uncertainty, but from any useful AI perspective (eg. a Turing test) a human observer will be unable to distinguish random from chaotic.
Could you please address the question of how you differentiate the causeless from the supernatural? This is a part of your thesis which I have been unable to fathom.
283. DocBrown - 7/11/2000 12:14:20 PM
Angel: Crosspost.
If causality is not the be all and end all of nature, then how can science work reliably?
If we live in a Universe in which causeless events occur, then how will we learn the mechanism by which causal events occur?
284. AytchMan - 7/11/2000 12:31:42 PM
A while back, Jay and I wrestled over our differing levels of confidence in the ability of technology to overcome the problems of developing an anti-missile system.
Putting the AMS aside, I'd like to ask a couple of questions on the pace of technological advance. What will happen if and when the advance of technology outstrips our ability to absorb it? Are there practical limits on the ability of humans to develop technology at an ever-increasing pace? Are we near them? I suppose that these questions are as much philosophical as technical.
285. DocBrown - 7/11/2000 1:01:37 PM
AytchMan, what makes you believe that there is a limit to the pace at which we can absorb new technology?
The limit to the pace of technological advancement is the limit of the money we are willing to devote to it. For example, right now we have the technology to develop many wonder drugs, assuming we are willing to devote a huge amount of money to healthcare. But Americans are getting fed up with the cost of healthcare, so we have invented HMOs and other things to limit the money we devote to healthcare. The less we spend, the slower the progress.
286. AytchMan - 7/11/2000 1:28:28 PM
Doc--
AytchMan, what makes you believe that there is a limit to the pace at which we can absorb new technology?
The limit to the pace of technological advancement is the limit of the money we are willing to devote to it.
Two interesting points. Once again, a Mote response creeps up on me from an unexpected direction.
To take an admittedly poor example, consider consumer electronics. In the last few years, we've leapfrogged through VCR's, CD's and CD-R/W's and a few other things. Now it's DVD. What would happen if Sony began to introduce a new device every three weeks instead of once a year? Haven't the social scientists documented cases of culture shock from people being thrown into a new and radically different environment? Too much change, as it were?
In a theoretical vein, I've always believed that the technology follows one of those classic S-curves that ramps up and eventually levels off. Are you saying that the curve can increase and even bend upward forever?
287. DocBrown - 7/11/2000 2:04:37 PM
This is simple economics. Consumers will only invest in a technology of they believe it will provide utility for an adequate period of time. If Sony introduced a new standard every three weeks no one would buy their products, even if those products were superior to the competition.
The key I was the oldest. I was also first to take up music, and my siblings never stuck with any instrument for more than a few weeks. The problem was that there was no good way to practice in that tiny, noisy house.
IBM and Microsoft understood this when they started giving PCs and software "backward compatibility." In theory, Sony could introduce new standards every three weeks if those standards were backward compatible with the old standards. So perhaps backward compatibility defines one of the speed limits to technology.
288. AytchMan - 7/11/2000 2:25:59 PM
I may be misinterpreting but I think you're stating that economics is the only real limit to our ability to absorb technological advance. True?
If so, do you reject the argument that there are psychological limits? The point I'm driving at is that there are practical limits (however difficult to define) somewhere below the pure rational ability of the brain to assimilate new concepts. To wax colloquially: beyond a certain point, peoples is uncomfortable with too much new stuff. Yes/no?
289. DocBrown - 7/11/2000 3:15:22 PM
AytchMan,
Perhaps in a cultural context the answer is yes. Some cultures have trouble absorbing new technologies that others embrace. Look at the Catholic culture and birth control, or the Amish and electricity. The government of China is threatened by the Internet. Europeans are suspicious of genetically engineered foods. American public schools struggled for years over whether to allow pocket calculators.
I believe that all of these problems are sociological, not psychological.
I am trying to think of a good example of a psychological barrier to technology, but I cannot. Perhaps the reluctance of my grandfather to use email is a good example. We all know that email is not difficult to learn, he simply has an aversion to changing his communication style.
290. PelleNilsson - 7/11/2000 3:48:15 PM
Too rapid change is not good for business. For two years I've contemplated buying a digital camera but knowing that what I buy today is obsolete tomorrow has held me off.
291. AytchMan - 7/11/2000 7:44:51 PM
Doc--
I am trying to think of a good example of a psychological barrier to technology, but I cannot.
I think your email example serves but an even better example is the computer itself. I personally know a number of people who are intimidated by the beasts. Most of these people could certainly be educated and desensitized but the point is that there is, I believe, a psychological resistance to the newfangled things. It's certainly not economic. Thus, this induces a certain drag on the adoption of new technology beyond your purely economic effects.
292. AytchMan - 7/11/2000 7:57:12 PM
Too rapid change is not good for business.
I agree and I'll go you one better -- too rapid change is not good for people in general. The point I've been trying to carry is that there is a practical limit to the pace of technological advance. This limit is some distance below the theoretical limit of how fast we can actually invent stuff.
I need some help here from the social scientists but I'd wager there are studies that document a growing resistance to the absorption of new technology beyond a certain rate. At the very least, studies of stress must point in that direction.
293. angel-five - 7/11/2000 11:51:40 PM
(Which I find to be gobbledygook. If a measurement affects a system,
then the measurement is itself input. If the system is deterministic, then
the result of this input will always be the same.)
I'm not really sure of your point of contention re: determinism. Can you clarify it?
IAC, the above point is pretty simple. When you try to account for the 'input' deriving from the initial act of measurement, you can only truly measure it by another measurement... I'm sure you can see the infinite regression waiting in the wings there.
As far as "If the system is deterministic, then
the result of this input will always be the same" goes, that would only be true if you had identical systems which you already possessed a full analysis of, and since a) Heisenbergian uncertainty says you can't have that full analysis; b) the only place identical weather systems are going to exist are within the confines of a thought experiment and c) even in a thought experiment you'd only derive that analysis with an act of complete measurement, the whole point becomes sort of moot.
294. angel-five - 7/12/2000 12:19:51 AM
In this context random really means random, as in a random number
generator. That is certainly not chaotic in the same sense as quantum
uncertainty, but from any useful AI perspective (eg. a Turing test) a
human observer will be unable to distinguish random from chaotic.
So? How's that bear on the topic? Illustrate.
If causality is not the be all and end all of nature, then how can science
work reliably?
If we live in a Universe in which causeless events occur, then how will
we learn the mechanism by which causal events occur?
I'm sure you intended these as rhetorical questions, Doc, but they aren't. And there's a simple answer -- science doesn't work reliably, not where we start running into problems with causality, unless it adopts probability models and does away with strict causality as a driving force behind quantum events. This is pretty obvious, no matter how unsettling it is to the Western logical model of scientific philosophy. Have you found many strict-causality-interpretations of quantum behavior being published lately? There's a reason why you haven't.
And as for your question: "If we live in a Universe in which causeless events occur, then how will
we learn the mechanism by which causal events occur?" -- well, the answer here is pretty obvious too. We propose causality mechanisms, but we can't verify them.
295. angel-five - 7/12/2000 12:20:01 AM
The problem of verification isn't really an issue when you're talking about the physics of a Tiger Woods tee shot or a five stage rocket attaining escape velocity -- that's because physical events on the macroscopic level give us very little reason to question causality. Causality is a comfortable scientific heuristic on the macroscopic level, we can all take it for granted (indeed, that's the basis for all classical physics) and indeed if Tiger swings his driver round and intersects a Nike golf ball head on, and the ball takes off down the fairway, no one will bother trying to tell you that the two acts aren't causally related.
Yet quantum events are another matter entirely, and it's a matter which often seems to ignore our notions of causality. That's something that's bugged many people since Einstein's time -- even though the outcomes of quantum reactions will, in sum, tend to align with statistical predictions, individually they can't be predicted at all.
There are still people like you, you know, who like Einstein want to insist that 'God doesn't play dice with the Universe' and instead propose some underlying and hidden causality which dictates the paths that individual quantum reactions may take. Since there's absolutely zero evidence for such a causality the statement requires a leap of faith as large as that which any Christian must make in order to embrace their faith.
But, to answer your question -- if causeless events do occur in the universe, then (at least when dealing with the levels of physics where causeless events take place) we can't distinguish causal mechanisms from causeless mechanisms. No doubt that leaves a sour taste in your mouth, and you'll reject it, but to my way of thinking it isn't worth it to deny that causeless events may happen just to make our mathematics tidier.
296. angel-five - 7/12/2000 12:25:55 AM
The limit to the pace of technological advancement is the limit of the money we are willing to devote to it.
Oh, come on, Doc. There are little things like the laws of physics to contend with. Technological advancement is not an endless procession. Matter has limits.
And even if it didn't, it's doubtful to me that increasing funding for research can never be counterproductive to the overall level of appliable science derived as a result of that research. You and your scientific-utopian stuff need to have a little lie down and a drink.
You vote Natural Law, don't you?
297. angel-five - 7/12/2000 12:36:28 AM
Aytch:
There are many reasons to believe that the majority of a culture will only absorb new technology at a limited pace. It's pretty plain to me, though, that if technological advancement heavily outstrips this rate of absorption one of the main results will be increasing differentiation between economic classes. New technology costs money. In a capitalist system it will be sold for as much money as the market will offer for it. Richer classes will therefore be much more likely to harness and benefit from newer technology.
This is as it's always been; it's just that as the pace of technological 'progress' hastens, and new technological advancements spring up that are only of use to those who are already at a certain technological level (computers are pretty useless to people without electricity, you know) we see the potential technological gap between classes growing larger and larger.
To me, that's much more of an interesting notion than whether the 'masses' will continue to embrace new technology or whether some Butlerian jihad will sweep through future society at some point in time.
I suppose someone could always invent cold fusion or a Star Trek replicator or whatever favorite scientific deus-ex-machina-Utopian-dreamscene-look-honey-it'll-
make-us-all-equal advancement you personally favor, but it seems to me more than likely that the public would never be presented with the full benefit of such a discovery. There'll be more money in it otherwise.
298. angel-five - 7/12/2000 12:48:46 AM
Too rapid change is not good for business. For two years I've
contemplated buying a digital camera but knowing that what I buy
today is obsolete tomorrow has held me off.
Well, that's certainly true, Pelle. Yet a lot of that sort of thing doesn't reduce directly to the rate of change, but the marketing strategy of technological companies -- like Intel having the next six generations of chip already designed but instead of jumping ahead and releasing their best product, instead marketing each chip in succession in order to make the most money they can. I imagine that the field is a little more competitive in digital cameras, but there's still the same sort of mechanism at work.
Of course, part of the justification of that mechanism is that, well, software lags behind hardware and even if they released the first 6000 mHz chip tomorrow it'd still take years for commercial software to be developed which could take full advantage of that blinding speed. (I don't know how true that is, but it's often said.)
If it's true, it'd be pretty much in line with what Aytchman's saying anyway. But for me the situation is more indicative of what I've been talking about, albeit in a subtly different way -- increasing the pace of technological development makes it likelier that the bulk of that technology will be held by a select few and consequently controlled and dispensed in a manner most likely to bring them the most benefits, financial or otherwise.
299. Indiana Jones - 7/12/2000 9:03:18 AM
When you try to account for the 'input' deriving from the initial act of measurement, you can only truly measure it by another measurement... I'm sure you can see the infinite regression waiting in the wings there.
Not so. For example, you can know how much resistance a VOM will add to a circuit before you use it. Then when you put it on the circuit, you just account for that.
Anyhow, it boils down to that you're using "deterministic" in a way that saps the word of its meaning. This is plainly evidenced in the contradiction in post 281 (drawn from your post 279). On the one hand you criticize Doc ("No, no, not at all") for saying that
Given the same starting state, a deterministic system must always reach the same new state...The behavior of deterministic system may be predicted precisely
but OTOH you in turn say
If identical weather systems receive identical input there's no reason to believe that they'll react in anything but an identical fashion... which, of course, means that they're deterministic
So apparently deterministic means whatever you choose it to mean at the time (i.e., predictable or not predictable).
300. Indiana Jones - 7/12/2000 9:06:02 AM
One more thing about measurement: the VOM is an example from application. If you want to just talk theory, then I stick with my original statement--a measurement that affects the output is itself an input. It does nothing to change the definition of "deterministic" (that is, the same inputs will result in the same output).
301. angel-five - 7/12/2000 11:53:14 AM
Jones:
I've sat here for the last five minutes reading your post again and again trying to find the contradiction you speak about, and failing. I suggest you read and reread the highlighted texts you selected; in the first I say that deterministic systems have identical output for identical input, and in the second I do too. Have at it.
Re: the circuits:
If you're really going to equate adding resistance to a circuit with measuring every particle of a chaotic fluid, then you know very little about the subject at hand. I'll try this again, though, just for you.
What we call 'chaos' mathematic is really just complex mathematical shorthand for predicting the possible courses a system of indeterminate state but known composition may take. We know, for example, that fluid will behave in a turbulent fashion under certain conditions -- that the nice and even flow patterns will be disrupted as smaller bits of the fluid strike out in different directions as a result of inner tensions and forces within the flow. We just don't know the precise form that the turbulent disruption will have, because we don't know the location and momentum -- or even good approximations of them --of every particle in the system.
Now, bear with me -- if modern science and a guy named Heisenberg is to be believed, we can't ever know both the location and the momentum of any single particle, let alone every one in the set, and Heisenberg hasn't been disproven yet. His science seems strange, but also seems to be spot on. But for the terms of the thought experiment, let's pretend that we can somehow get around Heisenberg's uncertainty principle, or at least we can get such good approximations of locations and momentums by some trick that we learn what we need to know.
302. angel-five - 7/12/2000 11:53:36 AM
So here we are with this weather system and we have this wonderful way of measuring it. And once we measure it and gather the location and momentum of every particle within it we can exactly predict what form the turbulence will take and all the other nice things in a chaos system which otherwise would give us fits. Except it's in an unknown state before we measure it, and we don't know where the particles are, and so when we go putting our instruments into the system to measure it we will get chaotic turbulent flow as a result as the system starts bouncing itself off the instrument, and that flow will disrupt the system in an unknown fashion. And you'll end up with the infinite regression of measurement, because, remember, the point is to possess a complete analysis of the system at any one time and you'll never get that from a single measurement. And this is even assuming that we can get round Heisenberg, which, I'd venture to guess, we never can.
303. AytchMan - 7/12/2000 2:33:59 PM
angel--
I'm in pretty solid agreement with your 296-298 although I was looking at it more from the perspective of the individual, not society and class. My goal was simply to establish that there were more serious limits to technological advance than just the amount of money thrown at it. Specifically, that people as individuals could only absorb it so fast. While I don't think we're at those limits, perhaps we're starting to get close.
304. AytchMan - 7/12/2000 2:40:51 PM
angel--
...even if they released the first 6000 mHz chip tomorrow it'd still take years for commercial software to be developed which could take full advantage of that blinding speed.
As stated, I think that's true but the key word in your sentence is 'full'. The increased speed of the chip would be of benefit immediately. But then software developers would slowly begin to code new software to take advantage of the new architectural features of the chip. Right now, it's about a five-year cycle to really wring all of the throughput juice out of a new chip. By then, of course, we've all moved on to the next one.
305. Indiana Jones - 7/12/2000 4:06:29 PM
a-5: The weather system you describe is pure theory, as you yourself admit. There's no realistic way of measuring every particle of a system that complex. That's why I plainly distinguished between theory and a real-life simple system that is measurable.
The real-life system, you can account for the impact your measurement has on it. And if you want to have a pretend weather-predictor system in your imagination, well, just pretend that you're accounting for the influence of measurement on it as well.
As for not being able to understand the contradiction, the first statement is not yours, but Doc Brown's, which you replied to with "No, no, not at all." That's why it's a contradiction: you disagreed with what he said and then said the same thing yourself.
(Which further convinces me you disagree with people just to disagree with them.)
306. ChuckSezdotcom - 7/12/2000 5:21:55 PM
Future story one
Future story two
Future story three
307. angel-five - 7/13/2000 1:27:29 AM
The real-life system, you can account for the impact your
measurement has on it.
How?
And if you want to have a pretend
weather-predictor system in your imagination, well, just pretend that
you're accounting for the influence of measurement on it as well.
You're missing the point. It's a thought experiment where you change one impossibility to ponder another.
Thanks for clarifying what you were talking about. In the instance you're speaking about, I wasn't arguing against Doc's statement about determinism except when he said that deterministic systems could be exactly predicted --that's pretty obvious given the context. You know, the response was as follows --
No, no, not at all. Stepping away for a moment from Heisenberg...
the notion that a deterministic system can be predicted precisely
ignores the fact that some deterministic systems are complex enough
that the act of measurement disrupts them. As for 'random'...
All of this, of course, makes me believe that you just disagree with people just to disagree with them, because I know you can actually read. I'm thinking little else is going to bring you into a conversation about chaos and uncertainty, because you aren't conversant with them if you can brush off the effects of measurement upon a chaotic system.
308. Indiana Jones - 7/13/2000 9:19:24 AM
How?
Refer to previous statements.
Re chaos: I prefer not to discuss my personal background on any subject because you should be able to respond or not respond to arguments as given, without any need for appeal to credentials. I therefore won't be goaded into a dick-measuring contest. Suffice it to say, in this instance I believe mine would prove every bit as long as yours.
Moreover, my point of contention with you was about the definition of deterministic. And on that, you still have done little to shore up a self-contradictory usage other than fling lots of sand.
In fact, you have now used weather as an example of both a deterministic system that can be predicted and cannot (compare your statements in
279: "If identical weather systems receive identical input there's no reason to believe that they'll react in anything but an identical fashion."
302: "Except it's in an unknown state before we measure it, and we don't know where the particles are, and so when we go putting our instruments into the system to measure it we will get chaotic turbulent flow as a result as the system starts bouncing itself off the instrument, and that flow will disrupt the system in an unknown fashion. And you'll end up with the infinite regression of measurement, because, remember, the point is to possess a complete analysis of the system at any one time and you'll never get that from a single measurement."
Lots of verbiage in the second, but you're essentially undercutting what you said in the first. 279 does reconcile with 302 if you accept my position that to the degree measurement affects a system the measurement is itself input, but accepting that would require admitting someone else might be right for once.
Naw.
309. angel-five - 7/13/2000 11:55:29 AM
Indiana, you're being dumb. The difference between whether a system is deterministic and ought as a result to be predictable given the right initial data, and whether a system can actually be predicted, ought not to be so easily lost on you. In the first case I'm speaking of theory and in the second I'm speaking of reality.
I believe that weather systems act in strictly mechanistic ways and therefore could, if you had a thorough analysis of them -- let's propose your noninvasive method of analysis just for grins, something that gives an accurate 'snapshot' of a weather system, and say we've used that -- predict exactly what the weather system is going to do, leaving aside unpredictable input from the earth's surface. So, yes, in fantasyland where instruments don't bias the experiment, or you can measure a turbulent system with instruments that will somehow have an exactly calculable effect upon the system, you can pull it off.
Outside of Fantasy land, though, it's just a fantasy.
310. angel-five - 7/13/2000 11:57:12 AM
You know, I actually was going to make the point earlier stressing that on the one hand I was talking about a thought experiment and on the other I was talking about the real world. (Hell, I thought I did anyway.) But I figured that there was no need to go on any further when it was already obvious.
311. angel-five - 7/13/2000 12:03:32 PM
Refer to previous statements.
Your previous statements just say that we should do it and can do it, and you offer, astonishingly enough, an example from electrical physics as some kind of proof. That's not a how, Indiana. I've already explained why it's not that simple -- you won't touch that, though, and instead you just say it can be done. Once again, how? Give me something to work with, here.
Re chaos: I prefer not to discuss my personal background on any
subject because you should be able to respond or not respond to
arguments as given, without any need for appeal to credentials. I
therefore won't be goaded into a dick-measuring contest.
That's okay, I made my point. And it didn't have much of anything to do with chaos itself (though you've made some pretty amazing statements for someone who claims at least to be somewhat conversant with chaos theory) but rather targeted your ironic presence in the conversation at all. Remember? You'd just had the incredible glass house inept chutzpah to chastise me over what you called disagreement for disagreement's sake.
312. Indiana Jones - 7/13/2000 12:06:41 PM
In the first case I'm speaking of theory and in the second I'm speaking of reality.
But your "weather system analyzer" exists only in theory (which you have repeatedly conceded). Incidentally as a reminder, here's what I said in post 300:
"If you want to just talk theory, then I stick with my original statement--a measurement that affects the output is itself an input. It does nothing to change the definition of 'deterministic' (that is, the same inputs will result in the same output)."
Anyway, I think we've beaten this to death by now, so you can have the last word should you so desire.
313. angel-five - 7/13/2000 11:37:51 PM
I take it you aren't going to explain how we can calculate the exact effect a measuring device will exert on an unknown chaotic system? with only one measurement, to boot? Oh, well, have fun, Indiana.
314. JayAckroyd - 7/14/2000 9:46:06 AM
On chaotic systems, A-5 is leaving out the most interesting part. While these systems are very sensitive to initial conditions, they generally can be expressed in simple, albeit non-linear terms. For example, the Mandelbrot set is a quadratic mapping z --> z2 = c where c is parameterized by x and y in the complex expression c = x + iy. (Complex expressions involve the square root of -1, the i in the expression).
On Aytch's question, there's no doubt that technical changes is often constrained by social patterns and previous investment in technical change. Network economies alone stop innovation--just try to introduce a new keyboard layout. I'm about to finish rereading Anna Karenina, and a running theme in the novel is the difficulty of introducing effective technical change in agriculture among the peasants (formerly serfs) that actually do the work.
315. JayAckroyd - 7/14/2000 10:16:42 AM
Here's a cia report on the foreign missile threat. CIA
It includes this fascinating quote:
"North Korea could convert its Taepo Dong-1 space launch vehicle (SLV) into an ICBM that could deliver a light payload (sufficient for a biological or chemical weapon) to the United States, albeit with inaccuracies that would make hitting large urban targets improbable. North Korea is more likely to weaponize the larger Taepo Dong-2 as an ICBM that could deliver a several-hundred kilogram payload (sufficient for early generation nuclear weapons) to the United States. Most analysts believe it could be tested at any time, probably initially as an SLV, unless it is delayed for political reasons. "
The 2005 date you hear quoted is not the date the Koreans will pose a threat. It's the earliest date the pentagon thinks they can have a defense. So, first, the Koreans aren't crazy, because they are not accelerating on their deployment plans to get us as soon as possible. Second, if they suddenly turn crazy, use 'em or lose 'em is a real threat--if this CIA report is to believed.
This again points out that this is a purely political issue in the US, with nothing to do with actually protecting the US from any missiles.
316. angel-five - 7/14/2000 11:09:08 AM
Most analysts believe it could be tested at any time, probably initially
as an SLV, unless it is delayed for political reasons. "
I sort of agree with Jay's analyses here but a few things also strike me. The first is that NK finds itself in some pretty precarious circumstances these days. It is in desparate need of Western (and, in its own case, Southern) aid and is terribly short of cash, and it's likely going to be at least half a dozen years before it can get its footing. These are some fairly pressing political reasons for NK to not push its WMD program at this time. So while the CIA's prediction track record the last decades has pretty much sucked spectacularly, I'd imagine that these factors are what have given them the 2005 estimate more than any AMD factors (2005 seems a little early for that, to me, anyway).
I also think it's important to keep in mind that nuclear weapons will not suddenly become useless to other nations if the US fields a working missile defense system. IT just means that other nations will not be able to use them on the US. Unfortunately or fortunately, depending on how you see the issue, there are plenty of other potential targets for these weapons and therefore they still have use. IT seems likely that the first generation of AM technology will be leaky enough and expensive enough that the US will likely not be extending its AM aegis anywhere else but the homeland.
317. jayackroyd - 7/14/2000 2:48:06 PM
You find this ambiguous:
"Most analysts believe it could be tested at any time, probably initially as an SLV, unless it is delayed for political reasons. "
It seems to be that you're switching the argument from "they're in desperate straits and will try anything" to "their straits are way too desperate for them to do anything."
I don' think either is true. The CIA analysis says they could test at any time. If that's so, the latter statement is false. And if it's not so, then the former statement is also apparently false--they're showing restraint right now.
318. AytchMan - 7/14/2000 2:52:02 PM
As a general rule, never expect the leaders of any nation to act rationally.
319. ChuckSezdotcom - 7/14/2000 6:35:53 PM
Not sure I got the hang of the subject here. Future War bout WWIII type nukes, "A Boy and His Dog," Sci Fi, or futuristic phyiscs. Got two today, first is the top feature of ChuckSez, Darkmatter; and then a piece by a professional writer on how he feels bout Missile Defense Systems.
320. angel-five - 7/14/2000 11:25:40 PM
I don't know that I'd characterize the NK position as being too precarious to do anything, Jay. All the same, they are in terrible shape and are in need of SK's assistance (and ours). That obviously has to weigh in their decision-making.
Of course, they are showing restraint one way or the other, whether it's out of a genuine desire for peace or sheer pragma. I personally wouldn't bet a bent nickel on the former, given NK's aggressive track record, but the fact remains that they aren't testing that delivery system and they could be.
Still. Let's be blunt -- North Korea is screwed. Their economy has been broken and they are in a food production deficit. They absolutely cannot afford to have the timetable for outside assistance put on hold, and I think they also realize that they can't beat their opposition, even with nuclear weapons -- not with their economy in ruins.
Don't get me wrong -- the fact that the North Koreans are making peace gestures, instead of planning for a full-out offensive against the South as a means of resolving their desperate situation, is very encouraging. It indicates that some things are changing in North Korea. Yet we also have to realize that they're in a bind, one that can reasonably be expected to affect how they choose to proceed with WMD research. Either they go flat out and hope for luck, or they stop it and accept the outside assistance they'll lose if they test missiles. They don't appear to be going flat out...
which, for me, when combined with the fact that it isn't a short process to test boosters for ICBM capability, and conversion isn't a quick process either, makes the 2005 figure entirely feasible without calculating in a Pentagon timetable for AMS tech.
321. DocBrown - 7/17/2000 11:38:49 AM
Me: The limit to the pace of technological advancement is the limit of the money we are willing to devote to it.
Angel: Oh, come on, Doc. There are little things like the laws of physics to contend with. Technological advancement is not an endless procession. Matter has limits.
Again our interpretation of words clashes like my wardrobe in the 1977.
I consider "the pace of technological advancement" to be akin to the velocity at which a group of automobiles might explore the roads of an unknown land. They will follow different routes and reach different destinations, if they reach a destination at all. We may imagine shortcut routes, but the explorers may only travel on the roads. We may imagine many exotic destinations, but the explorers can only discover places that actually exist.
I do agree that the laws of nature impose limits on what technology we may develop. However, they do not impose limits on the speed at which we may create that technology.
And I never vote Natural Law. I might vote Green, if Ralph Nader wasn't such a jerk. This November, I am leaning toward Captain Morgan.
322. DocBrown - 7/17/2000 11:58:14 AM
Angel said:
There are still people like you, you know, who like Einstein want to insist that 'God doesn't play dice with the Universe' and instead propose some underlying and hidden causality which dictates the paths that individual quantum reactions may take.
You keep bringing the discussion of AI back to quantum uncertainty, and I still do not understand why.
I am perfectly satisfied that God may "play dice with the Universe," so to speak. I think I said so about two weeks ago. If you want to equate uncertainty with causelessness (I asked you about this in 278) I will give you a wary nod.
Now please explain how quantum uncertainty limits our ability to create practical Artificial Intelligence, especially for straightforward military applications.
323. angel-five - 7/18/2000 2:15:33 AM
The mention of quantum uncertainty comes in at two points, Doc; since you made one of them yourself, I'd have at least thought you'd have gotten that one.
The first is the notion brought up by Jay that consciousness is an emergent property of quantum mechanics.
The second is your contention that a random number generator should do just as well in providing quantum 'randomness' in a neural net system. That point mystifies me, still. Random numbers are neither quantum in origin nor truly random in effect.
As to whether quantum uncertainty limits our ability to create artificial intelligence or something close to it -- I don't know that it does, neither do you. Where are you getting that notion from my posts? MY original comments were aimed at your 'mimicking the brain with RNGs' silliness.
324. DocBrown - 7/18/2000 9:44:07 AM
Angel, Jay's point about consciousness (whatever that is) is not necessarily relevant. Useful AI decision making, such as discriminating between a missile launch and an industrial accident, does not require consciousness. If you want to go back to the argument that the only proper use of the term Artificial Intelligence is in the context of Turing tests that is fine with me. But tiresome.
I thought the point about random number generators was straightforward. A human observer cannot distinguish between a macroscopic manifestation of quantum mechanics and a macroscopic manifestation of a simulation of quantum mechanics. Likewise, a human observer cannot tell a manifestation of a truely chaotic phenomenon from the output of a random number generator.
Note that I do not propose that the output of simulations and actual systems will ever be the same. My contention is limited to their ability to fool a human observer.
A Turing test, relies on the judgement of a human observer. Thus I believe a sufficiently complex simulation should be able to fool a human observer 50% of the time. I make no claims about the complexity of such a system . . . it might be greater than the entire Internet today.
For the purpose of any practical decision making, from driving my car to sorting my laundry to guiding my missile interceptors, I believe that a simulation of intelligent behavior can perform to my satisfaction.
325. JayAckroyd - 7/21/2000 3:33:09 PM
Some clarifications.
1) I don't believe that human intelligence is a result of quantum processes. Roger Penrose makes that argument in The Emperor's New Mind, which I may have mentioned.
2) I believe that the Turing test is a fair test for AI. I also believe that a less difficult test might also be fair, but have not yet heard of one. I note that this was the standard test until it turned out to be so hard.
3) I don't believe you can simulate consciousness with a Von Neuman machine. I don't think an OR gate is enough to simulate the complex interactions of neural firings and continuous neurotransmitter flows.
4) I think that a successful implementation of a star wars defense would require something close to consciousness on the part of the defense system. It is likely that any launch against such a system would involve a novel situation (since the tests would all be public knowledge, it would be easy to introduce novelty.)
5) It's also worth noting that the question of whether it would work is only one of several points I made earlier about the usefulness of a missile defense system. I happen to think "It won't work." is the weakest of those arguments.
Finally, I really think Doc is abusing the language when he characterized specialized mechanical activity by computers as "intelligent." Intelligence involves learning, dealing with novelty adjusting strategies, and some degree of consciousness. The cognitive ability that intelligence tests attempt to measure is very different from a robotic painting system.
326. DocBrown - 7/24/2000 2:04:08 PM
JayAckroyd said:
2) I believe that the Turing test is a fair test for AI. I also believe that a less difficult test might also be fair, but have not yet heard of one. I note that this was the standard test until it turned out to be so hard.
Jay, would you be willing to accept Turing as more of a pH test than a Litmus test? That is, the results need not show "this computer is artificially intelligent and that computer is not artificially intelligent." In this case you could conclude that "this computer is more intelligent than that computer."
Also, there is a definition for the word 'intelligence' that is not ambuguous at all. In PC parlance, the intelligence of a microprocessor is the number of operations it can perform in one clock cycle.
4) I think that a successful implementation of a star wars defense would require something close to consciousness on the part of the defense system.
Perhaps it would, although "close to consciousness" is awfully difficult for me to comprehend. Can you describe your vision further? If you are correct, then I would support a missile defense system for purely academic reasons. The technological leaps required to make it work would be akin to those the space program gave us in the 60s. Bring them on!
Yet politically I cannot support a missile defense system. Its only real effect would be to make the world angry at us.
327. Jenerator - 8/1/2000 11:24:57 AM
Take the Internet Addiction Quiz in the Internet Thread!! 328. marshame - 8/1/2000 11:51:18 AM
329. SheRex - 8/3/2000 6:41:33 PM
ChuckSezdotcom #319.
Do you think that once dark matter begins to be quantified and understood, there would be some "helpful" defense contractor ready to tech-transfer this knowledge into weaponry?
re BMD: My own employer made billions off of this boon-doggle in the 80's. They are salivating at the idea of NMD. I am especially concerned about the latest rage ... taking the human decision-making out of the loop in order to id & target the pre-boost (?) phase. To me it looks like the nightmare scenarios of Colosses the Forbin Project and Skynet from Terminator, all rolled into one.
330. concerned - 8/3/2000 10:52:45 PM
SDI/BMD - For now and the future. The Left Wing Talking Points against it are, for the most part, abysmally stupid and shortsighted.
331. Wombat - 8/4/2000 11:09:03 AM
The unworkable to defend against the improbable.
332. DocBrown - 8/4/2000 1:33:08 PM
Yes, Wombat, BMD may turn out to be an elephant repellant. Every day we would know that it keeps our enemies from attacking us with missle because no enemies have attacked us with missiles today.
I wonder if the hue and cry from international community would die down if America did the research, but never actually deployed the system?
That way we would get the technological benefits, without wasting money on deploying a useless system and angering the world.
333. Wombat - 8/4/2000 1:40:16 PM
Doc:
I'm all in favor of continuing research. In a decade or two, they might come up with a reliably workable system. Of course, by then Saddam may be dead, the Koreas united (and democratic), Iran pro-west and democratic.
334. foxfourgib - 8/11/2000 6:01:46 PM
DocBrown
"Every day we would know..." You know the same argument could be made for the deterrent value of MAD (Mutual Assured Destruction). Or that the only reason we never attacked the Soviets that way was that they had a missile defense and we didn't. After all, we never did. (In the glad old days of the original ABM deployment - we retired ours after 1 day in operation, Moscow still has theirs (and it's been upgraded a mite over the years)) I do think that research should continue, but it should also be accompanied by a concerted educational campaign designed to cure the polity of their mistaken belief that we really do possess a defense now. They should know the truth and not the urban legend.
335. Indiana Jones - 8/11/2000 6:07:41 PM
Howdy, foxfourgib, long time no see. Things have been kind of slow around here lately, but seem to be picking up the last couple of days.
Hope you'll stick around.
336. foxfourgib - 8/12/2000 11:29:37 PM
Off and on, no problem. I lurk mostly - keeps the BP more in line.
337. Cellar Door - 8/24/2000 12:15:14 AM
GAYS RULE!
338. Indiana Jones - 8/30/2000 2:33:33 PM
U.S.-Israeli laser knocks out rockets in arms test
339. DocBrown - 8/30/2000 3:35:17 PM
Interesting news, Indiana.
The desert is probably the perfect place to use this weapon. Both New Mexico and Israel have good weather conditions for firing lasers. If I tried to fire one from my house on most days, the beam would have to cut through clouds or rain or snow.
I wonder how expensive it would be to implement polished chrome plating as a countermeasure?
340. Indiana Jones - 8/30/2000 3:43:59 PM
Doc: I don't pretend to have an interpretation as to what this actually means...just reporting the news.
341. benwolf - 8/31/2000 2:02:02 PM
Maybe reflective armor would save a missile. Even if it could save a tank or an infantryman, it would just make them easier to see for conventional weapons to hit them. If this fire control system can direct a laser, it can also direct other energy streams or even hard projectiles.
342. Indiana Jones - 9/1/2000 4:52:52 PM
Notice: This thread needs host volunteer.
Clinton passes on missile defense decision
The Mote | Mote Archive
back to top