Matrix Games Forums

Real dumb question...if I am editing a full campaign scenario, how do I set the AI?...as in full WITP scenario. Thanks fer yer patience with me idiocy.

Good question.

I once asked a commissioner at a hearing - and his reply was

"Good question. Next question?"

AI is not programmable as such for a full scenario. The manual says it is programmable for micro scenarios - but not anything about how to do that. You must answer questions in the Scenario Editor - and the question is the only guide you have. You can set objectives in order - 1 to 4 - for example.

To some extent is is possible to "program" the AI by how you set certain values in the database. In many cases it appears these settings will be changed by the AI on/after the first day - but sometimes they work. For example, you can set the altitude of an air group, and the objective of a land unit. Other values may be set - but not in a normal editor - because the fields required do not appear in a normal editor.

In general, AI is not AI at all. It is achieved mainly by hard code - instructions per slot to do this or that on/after this/that date - apparently. It does this moderately well for Japan - and (according to Joe Wilkerson) cannot work for the Allies no matter what.

To this add that AI becomes worse than horrible late in 1944. All Japanese air groups convert to kamakaze status between Aug and Nov 1944 - meaning you no longer can do recon, search, ASW, fighter missions, land bombing missions, name it - and all your units die on first use. It renders the game unplayable - and the ONLY solution is to NOT let AI be boss - but to enter the Japanese move manually (or lie and say you did).

In general, AI is not AI at all. It is achieved mainly by hard code - instructions per slot to do this or that on/after this/that date - apparently. It does this moderately well for Japan - and (according to Joe Wilkerson) cannot work for the Allies no matter what.

To this add that AI becomes worse than horrible late in 1944. All Japanese air groups convert to kamakaze status between Aug and Nov 1944 - meaning you no longer can do recon, search, ASW, fighter missions, land bombing missions, name it - and all your units die on first use. It renders the game unplayable

Any chance of improving this situation for the Admiral's Edition?

The AI might convert units to kamikaze units, but it can then assign those units to other missions - so it does fly naval search and naval attack. I found this out to my frustration as H8K's started doing naval attack missions. After spending 4 years doing naval search they had incredible experience and hence accuracy. Meanwhile their durability and armament made them the toughest Japanese strike aircraft to shoot down.

I actually prefer the AI as the Allies - no matter what mistakes it makes, it always gets reinforcements to replace its losses. This makes for a good game, so long as you ignore the score.

The official story is there is no money for true AI. However, they have ripped out a lot of hard code from slots, similar to war plan orange.

The AI does not understand strategic thought in any sense. It will permit task forces to sail forever without regard to fuel, ignore deep penetration of space by enemy convoys feeding bases in range of strikes.

So the AI does not work for the allies? I played the first year of the war as Jap vs AI and was apalled by how dumb the AI was when I loaded the game and looked at it as the allies after 12 months. I edited the first turn to make the japanese first moves smarter and thought I had screwed up the AI because I THINK enabled scenario AI (I subsequently erased the scenario inadvertantly, so I cannot be sure).

ORIGINAL: el cid again

The official story is there is no money for true AI. However, they have ripped out a lot of hard code from slots, similar to war plan orange.

The AI does not understand strategic thought in any sense. It will permit task forces to sail forever without regard to fuel, ignore deep penetration of space by enemy convoys feeding bases in range of strikes.

AI is hard. We didn't know how hard until last year.

Intelligence is an emergent property of brains. We now know (since Wu's 2008 paper) that emergence is real, and reductionism can no longer be treated as axiomatic. To get intelligence you need a lot of parallelism, so current computer designs are wrong-headed if you want intelligence.

AI is very hard.

I must agree.

Japan bet the farm on a new form of computer/software that was supposed "to work like a human brain." I said it would not work - and it didn't.

Still - I remember when we said a computer would never be able to play chess and win. Now a chess master can beat a cheap computer program one time in ten or so. That is the sort of AI we might be able to get into a game. It isn't actually intelligent - but it is an expert system - a set of rules "in these conditions do that" - and indeed - expert systems in many fields are better than experts in that they never get tired, the never forget every aspect of the subject, and for most instances they give the correct solution. That sort of "AI" is possible - even though it is really just a set of rules created by the actually intelligent humans.

ORIGINAL: herwin
AI is very hard.

Shoot Harry, AI is impossible. Nobody even understands the mental process entry points for intuition, much less the psychological reinforcement aspects of cohesion, bonding/gathering, will, confidence.

All AI can do is process variables and algorithms that can be better, and better defined. Then, it’s just a matter of doing the math to see who wins or loses. Very valid in the general situation. And very valid as a basis.

But, every now and then, the human mind comes up with something out of the box. Maybe that’s what intelligence truly is; the ability to appreciate and react in a unique and (okay) intelligent fashion to a stressful and unusual situation.

You know, and I know, that situational analysis is structural and formalized. But you know, and I know, that situations are dynamic, and the effects of one man, doing something utterly stupid and utterly courageous will effect the course of an entire battle.

AI stuff, if sufficiently well understood, is great for comparing nominal doctrinal impulses to opposing nominal doctrinal impulses. Wonderful for planning purposes. But in the nitty-gritty, AI breaks, because it does not subsume the intelligence factor. AI is structural. It is not and cannot be intelligent (human).

ORIGINAL: JWE

ORIGINAL: herwin
AI is very hard.

Shoot Harry, AI is impossible. Nobody even understands the mental process entry points for intuition, much less the psychological reinforcement aspects of cohesion, bonding/gathering, will, confidence.

All AI can do is process variables and algorithms that can be better, and better defined. Then, it’s just a matter of doing the math to see who wins or loses. Very valid in the general situation. And very valid as a basis.

But, every now and then, the human mind comes up with something out of the box. Maybe that’s what intelligence truly is; the ability to appreciate and react in a unique and (okay) intelligent fashion to a stressful and unusual situation.

You know, and I know, that situational analysis is structural and formalized. But you know, and I know, that situations are dynamic, and the effects of one man, doing something utterly stupid and utterly courageous will effect the course of an entire battle.

AI stuff, if sufficiently well understood, is great for comparing nominal doctrinal impulses to opposing nominal doctrinal impulses. Wonderful for planning purposes. But in the nitty-gritty, AI breaks, because it does not subsume the intelligence factor. AI is structural. It is not and cannot be intelligent (human).

Amen, brother.

I was going over the evidence about the genetics of male obligate homosexuality with my DNA forensics students on Monday. I explained why we now suspect there are at least two genes involved on the X chromosome, and discussed how the mechanism seems to be increased preference for males in both sexes, with females having increased fitness and males decreased fitness. Then they asked me how it worked, and I had to admit how little we know about brain development.

I think intelligence and consciousness are closely linked, having to do with creating and maintaining a model of the world. This model is updated based on various kinds of sensory data and is used to control behaviour. It appears to be distributed--my advisor, Karl Pribram, would say 'holonomic', which makes it very hard to model. So intelligence is hard, very hard, and AI is completely wrong.

Curious about something. Given that the Homeric Greeks understood concept, as well as us, efficiency, materials, etc.; and further given that wars cull the best of us; I wonder if we as a race are as smart as we were 3,000 years ago?

Maybe the Greeks had 1 or 2 % shit-hot dudes in the mix. Can we find that many now? I know this is also a Darwinian question, but it does have to do with intelligence. Does intelligence randomization decrease as the +3 sigma population is eliminated?

I've seen comments to the effect that such things appear to be held sort of 'in reserve' in the gene pool and re-emerge later. Very interesting question regardless of the answer.

ORIGINAL: JWE

Curious about something. Given that the Homeric Greeks understood concept, as well as us, efficiency, materials, etc.; and further given that wars cull the best of us; I wonder if we as a race are as smart as we were 3,000 years ago?

Maybe the Greeks had 1 or 2 % shit-hot dudes in the mix. Can we find that many now? I know this is also a Darwinian question, but it does have to do with intelligence. Does intelligence randomization decrease as the +3 sigma population is eliminated?

Mean IQ has been rising fairly rapidly for the last 100 years. This is probably due to diet + schooling, so your average Homeric Greek, lacking both, was probably as dumb as Homer Simpson.

Variance (randomness) is driven by the size of the +/-3 sigma population, so if you eliminate them, randomness goes down. The variance of male IQ is about twice that of female IQ, so despite the mean IQ being about the same, you get more male geniuses and Darwin Award winners.

ORIGINAL: witpqs

I've seen comments to the effect that such things appear to be held sort of 'in reserve' in the gene pool and re-emerge later. Very interesting question regardless of the answer.

Epigenetics. It's real. Many molecular biologists are unwilling to accept that the central dogma (DNA->mRNA->protein) fails sometimes.

If human intelligence was due to the accumulation of a large number of small genetic effects, then we would be better able to explain it with reductionist thinking (as the molecular biologists want to do) and it would be susceptible to change by truncation selection (killing off the smartest or the dumbest).

However, human brain size and intelligence changed extremely rapidly in early humans whose interbreeding populations were very small so that stochastic effects play a more important role in genetics. Then it is more likely to involve the sort of changes described by Sewall Wright's shifting balance theory of evolution. From this standpoint human intelligence can be viewed as an emergent property not just functionally, but genetically as well. Reductionist arguments aren't nearly enough.

Complex traits that appear in this fashion have large amounts of non-additive genetic effects and the genetic variability is quite resistant to change by truncation selection. For this reason, eugenics was not only morally reprehensible but doomed to failure. Likewise, the practice of tyrants for millenia of killing the "smart" members of a subjugated population in order to breed a population of slaves is also pointless. The underlying genetic variation in intelligence now is probably the same as for the ancient Greeks. The genetics of intelligence involves potential intelligence and Herwin pointed out some important ways that environment affects realized intelligence.

Human intelligence, for better or worse, is what it is and has been for a long time.

This seems to have gone thoroughly off-topic, but yes, you're not massively wrong. Good night [>:]

I'm just trying to fool my boss into thinking that I'm working rather than gawking at the WITP forum.

I admit to being a bit out of my league here (and one should never attempt to discuss intelligence after a couple glasses of wine)
But I think JWE may have a valid point here regarding the ancient Greeks of Homer's age.

I cannot believe that a race that defined mathematics, reason, and science of the natural world as the ancient Greeks did - could have been dumb as Homer Simson ...just because they did not have an established mandatory publicly funded state-run school system.

Also, I believe that IQ tests are valid - but not necessarily the last word on defining intelligence - hence the recent rise in IQ scores may not, after all, represent a real growth in intelligence - so much as an increased standardization of test results? I don't know. But warfare dose cull the herd by robbing the gene pool, and intelligence, collectively, may have suffered over the generations.

What's wrong with a Darwinian approach anyway?

B

ORIGINAL: herwin

ORIGINAL: JWE

Curious about something. Given that the Homeric Greeks understood concept, as well as us, efficiency, materials, etc.; and further given that wars cull the best of us; I wonder if we as a race are as smart as we were 3,000 years ago?

Maybe the Greeks had 1 or 2 % shit-hot dudes in the mix. Can we find that many now? I know this is also a Darwinian question, but it does have to do with intelligence. Does intelligence randomization decrease as the +3 sigma population is eliminated?

Mean IQ has been rising fairly rapidly for the last 100 years. This is probably due to diet + schooling, so your average Homeric Greek, lacking both, was probably as dumb as Homer Simpson.

Variance (randomness) is driven by the size of the +/-3 sigma population, so if you eliminate them, randomness goes down. The variance of male IQ is about twice that of female IQ, so despite the mean IQ being about the same, you get more male geniuses and Darwin Award winners.

I try to teach college history and this is WAY beyond my pay grade...

Duck in John, it has way wandered off topic - but it's a worthy point. Besides, if you are familiar with the writings of Homer, Herodotus, Xenophon, and Thucydides - you can at least argue they were quite intelligent citizen-soldiers (without going into the philosophers)[;)]

As for me - I'll admit I didn't finish college, but I have been a passionate reader of history.

ORIGINAL: John 3rd

I try to teach college history and this is WAY beyond my pay grade...