Catch 22 - the AI
Moderators: Joel Billings, elmo3, Sabre21
RE: Catch 22 - the AI
deleted.
RE: Catch 22 - the AI
ORIGINAL: redmarkus4
I sense that a part of the problem with the AI is the objective setting. The AI seems to ignore city hexes and other key historical objectives on the defense. Perhaps this is why it doesn't attack aggressively as well?
It certainly doesn't ignore cities nor key terrain, but it doesn't treat them all the same, not does it cling to ground if it is threatened to be outnumbered, overtaken or even surrounded. I find it performs extremely well for an AI, especially if comparing it to any other strategic AI I have seen so far. Compared to the one in WitP/AE it is quite a bit better (though the former, to defend it, has to deal with even greater complexity as air is treated more explicitly, and the key naval component adds a completely new level which amplifies all logic challenges, even for a human sorting out his next amphibious landing).
It is smart enough to even start evacuating key cities if isolated and attempt a break-out. Often too little too late, but a human player would face the same challenges given how strong Germans tend to be in 41/42, or Russian Rifle Corps post 43/44.
If at all, I found it gives Moscow up a tid bit early (in comparison to what would have likely been the historical course), but to be honest that way
it has at least once in my GC saved a whole lot of Red Army units worth much more than a few city blocks in the long run. So taking into account that neither I and probably also all other players won't play the game in utmost historical fashion (kind of playing as if the own head could get separated from the body if you screw up or needlessly waste your troops), the AI has to in fact diverge as well from historical doctrine and behavior if it is to be competitive with inventive gamers that like employ all available hindsight and experience, perhaps even fight a modern war rather than a world-war where blitz krieg and close air-land cooperation were pretty new and large untried.
If it would stick to historical behavior or strategies, i.e. hold forward too long as a Russian in 41 (along Stalins orders), perform needles counterattacks to open pockets or hurt the Germans, push the Wehrmacht forward in increasingly bloody battles right up to blizzad (instead of digging early), or later not retreat German Divisions from cities once threatened by Russian envelopment (along Hitlers "Feste Punkte" orders), I would bet it would be extremely easy to defeat given today's knowledge.
For the lack of aggression, that clearly changed with the patches. Not sure what they changed, but it may either have to do with the improvements that affected the CV of units (if the attack decision is computed based on relative CV?), or with some kind of threshold they may have tuned up in the AI routines deciding above which odds to attack (or maybe even "pre-simulating" the outcome roughly and deciding on base of average losses?). If you tune the difficulty levels across the board up to >=110% for AI, it gets more aggressive, but not smarter (also with that also chances for it making small pockets seem to improve).
The best experience as Axis player I got so far out of the game (at 110% levels) if I did not start with rail repairs before turn 6, and not use HQ-Build-up before turn 8. I advanced under a lot greater supply strain, much slower which gives AI some extra chance to reorganize after each pocket or breakthru. That caused big headaches in front of Leningrad, or Moscow for me, which, however, felt very much as I would expect it.
In contrast I have tried to test how far Axis under optimal supply conditions and HQ-buildup use could advance prior to December, i.e. test where the supply model would put an inherent stop, but given the low Soviet MP and CV in 41 I succeeded in taking even Stalingrad, Tambov and Murov. I have not played the Soviet side into the late 44 phase when it theoretically should start rushing forward, but I would expect the same to be true. So putting a strain on your supply makes the AI cope much better with the fast movements the phasing player can make.
Certainly an AI could be programmed that would mimic smartness, and could follow "by the book" tactics as much as human could stick to a safe course by the book. And there are also groups today working on "learning algorithms" that for example work by trying to identify key parameters in a set of less important ones for the outcome of processes, but that doesn't sound like anything a game publisher in a niche sector (or even a large studio) could afford. Nor would I be willing to pay a couple of 100 or 1000 dollars for a license, or the 64 node cluster needed to run an AI turn in less than 7 days... As a mathematicians joke would say -- the problem is solvable. Just don't mention the time required...
One cheap way to allow a systematic improvement of AI exists, though: modders, if the routines were written in a sufficiently powerful scripting language and open to modification and enhancement. That I still find would add a lot to 2by3 or Matrix products in general.
For the price of the game, I can only say the they have done a great job with the AI, and I hope that will continue that when a real naval component and a more detailed air war (the closer to WiTP/AE detail levels, the better) will added in the future titles.
- Redmarkus5
- Posts: 4454
- Joined: Sat Dec 01, 2007 1:59 pm
- Location: 0.00
RE: Catch 22 - the AI
Well, I posted more detailed results of my latest game vs. the Soviet AI under the last beta in another thread. All I can say is that you and me must be playing a different game.
Come December '41 the Soviet AI wouldn't even attack except in six hexes along the entire front and even there it attacked 26,000 Axis troops with a single brigade, even though there were several Soviet divisions in the adjacent hexes, unused.
I admit that I only had it set on 'normal' 100%. But really, only six attacks along the front and a complete failure to commit the forces that are already sitting right there?

Come December '41 the Soviet AI wouldn't even attack except in six hexes along the entire front and even there it attacked 26,000 Axis troops with a single brigade, even though there were several Soviet divisions in the adjacent hexes, unused.
I admit that I only had it set on 'normal' 100%. But really, only six attacks along the front and a complete failure to commit the forces that are already sitting right there?

- Attachments
-
- AIQuestion2.jpg (469.88 KiB) Viewed 271 times
WitE2 tester, WitW, WitP, CMMO, CM2, GTOS, GTMF, WP & WPP, TOAW4, BA2
RE: Catch 22 - the AI
ORIGINAL: redmarkus4
Well, I posted more detailed results of my latest game vs. the Soviet AI under the last beta in another thread. All I can say is that you and me must be playing a different game.
Come December '41 the Soviet AI wouldn't even attack except in six hexes along the entire front and even there it attacked 26,000 Axis troops with a single brigade, even though there were several Soviet divisions in the adjacent hexes, unused.
I admit that I only had it set on 'normal' 100%. But really, only six attacks along the front and a complete failure to commit the forces that are already sitting right there?
You could try increasing the difficulty levels. Not sure it works here like in AE, where you can change levels in-game to give AI a boost for a couple of turns by providing for e.g. with supply benefits (no supply limitations/failed checks that is) that help a lot with rebuilding units or getting airframe replacements before switching back to a level where isolation and starving is possible. Never tried.
AI is naturally best on the defense, since that's inherently easier than organizing an offense. Since setting morale difficulty to >=110% allows AI to pass all admin checks for MP, and Soviet MP are inherently low due to poor leaders and proficiency (moral), this makes quite a difference. I recall the next level was about 125%, at and above which AI passes all dice rolls for initiative checks. That makes a huge difference, but I found that to be too much under patch 1.05 as AI to evidently started brute force grinding -- so I quite stopped that GC quite quickly. I guess AI would do the same as Axis and take Moscow with ease at this level? Perhaps someone has tried that and can comment?
To me it seems like the limit on AI aggression, at least for Soviet AI, stems to a good part from the poor MP, and, thus, poor C&C of the Soviet Army? Perhaps also for AI games the Soviet side is a bit to weak now in the early years, with 1.05.xx?
Not sure how sensitive AI is to being ground down prior to winter, though. A dev would have to explain what exact checks the AI would undergo to see how CV and so on are considered. I would guess you are facing pretty weak units?
RE: Catch 22 - the AI
Programming a good AI may be a challenge that will always be tough. A human looks at the battlefield and can think "yeah, this looks good over here"...the computer has to compute odds, maybe the VP of the cities nearby, maybe ( or not ) look at units in the general area, all sorts of doodads.
RE: Catch 22 - the AI
I just finished my first '41 GC against a Challenging Soviet AI. I won a decisive victory in Nov '43. The AI put up a pretty tough battle until Leningrad fell in fall 42, and Moscow in winter 42/43. After that things went south quickly. The 41/42 blizzard was tough for me, and I even lost a Hungarian and German division in an encirclement outside of Kursk. I found the AI would attack with reasonable aggressiveness when I overextended myself. Overall the campaign was great fun, but both sides were hit by bugs in the betas that affected the overall result.
I made a lot of mistakes in my first GC, and can do a lot better. Going to crank up to Hard and give it another go, perhaps with a more unconventional overall strategy.
I made a lot of mistakes in my first GC, and can do a lot better. Going to crank up to Hard and give it another go, perhaps with a more unconventional overall strategy.
RE: Catch 22 - the AI
I tried the 1.05 version with Axis set to 125% and Soviets at 80% (Challenging setting). I won it in version 1.03 in a fairly easy fashion. The 1.05 version, however, kicked my butt. By mid September I had no Soviet divisions that had any training left. Essentially the Axis AI just steamrolled me. So I tried it again with running all of the Soviets to the Urals and trying to allow the computer to have its own way - it penetrated only four more hexes than when I was fighting it. So in the game I lost 4.3M men by mid-September and could say I slowed down the Axis four whole hexes!
The new code and rules shatter Soviet formations MUCH easier. So I tried some tests and found that with Axis AI at 110% and Soviet AI at 100%, the Axis AI won a decisive victory in November 1944. So I now play the Axis AI at 110% and my Soviets at 100% and it is a much better game (for me at least!).
The new code and rules shatter Soviet formations MUCH easier. So I tried some tests and found that with Axis AI at 110% and Soviet AI at 100%, the Axis AI won a decisive victory in November 1944. So I now play the Axis AI at 110% and my Soviets at 100% and it is a much better game (for me at least!).
RE: Catch 22 - the AI
ORIGINAL: carlkay58
I tried the 1.05 version with Axis set to 125% and Soviets at 80% (Challenging setting).
Those are Hard settings, not Challenging. Challenging is AI 110%, player 90%.
RE: Catch 22 - the AI
OK, that makes me feel a bit better!
- Joel Billings
- Posts: 33494
- Joined: Wed Sep 20, 2000 8:00 am
- Location: Santa Rosa, CA
- Contact:
RE: Catch 22 - the AI
Gary has found that when he runs AI vs AI tests, he'll set the attacking side AI at 110 morale, and all other items for both sides at 100. That one change is usually enough to get the AI attack routines to equal the AI defense routines (i.e. tougher for the AI to attack then to defend). He uses these settings to test any new changes he's making to the AI to see what impact they have on the game. Adjusting the help levels to suit your play style is encouraged.
All understanding comes after the fact.
-- Soren Kierkegaard
-- Soren Kierkegaard
- kfmiller41
- Posts: 1063
- Joined: Tue Mar 25, 2003 9:00 pm
- Location: Saint Marys, Ga
- Contact:
RE: Catch 22 - the AI
Can hardly blame the AI for ignoring cities as after 41 as Gemrany I normally do as well because they mean nothing if i am defending. In a current game i am playing I decided to stand my ground where I was and fight it out in 43 against a human player (although he has only 6.5 million vrs 3.7 million men and I have more armor than he does at this point, yes I have been lucky) and normally what I see as russain player is german players 1 hex stepping back ever turn to blunt any break thru, and it works very well. So well that if your behind in your timeline your not going to make it to Berlin. Still agreat game though[:D]
You have the ability to arouse various emotions in me: please select carefully.
RE: Catch 22 - the AI
The AI seems to forget to tidy up encirclements, and just marchs it's merry, mindless way east. I thought I would try it, yet stopped when all it did was bypass all the Western Frontier Soviet units it so brilliantly cutoff on turn one.
Marquo [>:]
Marquo [>:]
RE: Catch 22 - the AI
It definitely has problems cleaning up its rear areas. I find that if you "hunker down" and sit still with a unit, keeping it going with air supply, for a few turns, you will then be able to do just about anything with the unit without the AI reacting to it. Thus fell Warsaw and Koningsburg in one game.
RE: Catch 22 - the AI
Almost every game against an AI you simply need to keep from doing some things that are just not going to work against a human. You know what those are. If you want a decent game.
BS, MS, PhD, WitP:AE, WitE, WitW
RE: Catch 22 - the AI
It's still brute force calculation for a chess programm.
And yes, chess is comparingly easy - you had only 1 (one) move per time, 8 x 8 hexes, max. 32 pieces.
How many potential possible moves had a Panzerdivision? Nearly 50 in each direction... So, in chess a Queen has a maximum of what? 7+7+7+7= 28
Each Panzer had more moves to calculate than a Queen in chess. And how many Panzers did we have, let alone the other units? And the machine had move them ALL at once...
And yes, chess is comparingly easy - you had only 1 (one) move per time, 8 x 8 hexes, max. 32 pieces.
How many potential possible moves had a Panzerdivision? Nearly 50 in each direction... So, in chess a Queen has a maximum of what? 7+7+7+7= 28
Each Panzer had more moves to calculate than a Queen in chess. And how many Panzers did we have, let alone the other units? And the machine had move them ALL at once...
- Redmarkus5
- Posts: 4454
- Joined: Sat Dec 01, 2007 1:59 pm
- Location: 0.00
RE: Catch 22 - the AI
I agree that making an 'intelligent' AI that plays like a human is probably impossible. However, if the AI has 6 divisions (total CV of 24) in adjacent hexes facing 1 enemy division (CV of 2) it should just attack and push through. Not much high science needed there:
IF AI CV > than Human CV by ration of 3/1 or higher THEN attack with 75% of adjacent units.
Not attacking in such a situation is simply going to ruin any hope of a reasonably fun game vs the AI, which kills half the potential market for future games in this series.
I want the AI to work so that the series succeeds commercially and I get the final result I need - a game that delivers a convincing simulation of WW2 at the operational level.
IF AI CV > than Human CV by ration of 3/1 or higher THEN attack with 75% of adjacent units.
Not attacking in such a situation is simply going to ruin any hope of a reasonably fun game vs the AI, which kills half the potential market for future games in this series.
I want the AI to work so that the series succeeds commercially and I get the final result I need - a game that delivers a convincing simulation of WW2 at the operational level.
WitE2 tester, WitW, WitP, CMMO, CM2, GTOS, GTMF, WP & WPP, TOAW4, BA2
- Redmarkus5
- Posts: 4454
- Joined: Sat Dec 01, 2007 1:59 pm
- Location: 0.00
RE: Catch 22 - the AI
ORIGINAL: jazman
Almost every game against an AI you simply need to keep from doing some things that are just not going to work against a human. You know what those are. If you want a decent game.
+1
Players who want to look for every conceivable way to beat the AI by exploiting holes in the game system can safely be ignored in this forum - they represent a small minority of the consumer base. Most want a game that rewards realistic play with a plausible outcome.
The Devs should focus on making sure that plausibility is central to the game design.
WitE2 tester, WitW, WitP, CMMO, CM2, GTOS, GTMF, WP & WPP, TOAW4, BA2