Can someone refresh my memory on the difference between hard and very hard AI?
manual is not specific on the topic, it just says
historical -- balanced, no advantages
hard -- some logistical advantages
very hard -- some logistical and combat advantages
if my memory doesn't fail me, I believe you can't starve the AI under "very hard", am I right?
my question goes more for "hard"; does the AI still needs supply under "hard"; e.g. would the Allies AI starve at Bataan, if left unattended, or would it remain in supply forever?
Lotta bad info on the AI here, sorry for that.
Alfred provided a link to a prior thread and all of your questions are answered there.
The reason the manual is not specific about the AI is the manual pre-dates the AI. Worse, after it was written there were some very SERIOUS changes made to the game that adversely impacted the AI. These changes were made in favor of the PBEM "purists". I won't dig those discussion up, but feel free to search them up and read them. Anyway, those changes in some ways broke the AI rather badly and it has been in catch up mode ever since. The Ironman series puts many of the changes back in (in terms of the TOE's), but it can't completely fix all of the changes that were made. My last point is that, AFAIK, as part of the big change just before release, the effects of the AI difficulty settings were changed to be consistent with the other changes. (If that makes sense.)
That said, the newer AI scripts that Andy just released are really pretty good. I've been testing them and the AI does quite well for itself.
Further to PaxMondo's comments.
There was a very conscious decision made by the devs to not disclose in any great detail, the computer benefits (and disadvantages too) which result from the different difficulty levels. Forum searches will find a few dev comments on the subject, but they are couched in very general terms. This dev policy does make it hard for the general player to determine accurately what is the exact situation. Which is exactly what the devs wanted and as such it would be most inappropriate for me to breach the dev policy.
I will expand slightly on PaxMondo's observation of "breaking" the AI. The word "breaking" is too harsh for it implies that the computer opponent is useless. That is a claim often made by PBEM players who tend to talk nonsense about the AI, usually based on their limited experience of deliberately playing ahistorically implausibly which the historically plausible scripts cannot be made to react to such human behaviour. If the situation were reversed everyone would complain of (a) the AI cheating and (b) they had bought a game which was supposed to be based on the historical record. Imagine the outcry if there were a script for 7 Dec 1941 where the Japanese could use the first turn magic move bonus to land on the West Coast. Or to land sufficient forces to capture Pearl Harbor on 7 Dec 1941.
It was one of the dev aims to improve the quality of the AE AI compared to that of classical WITP. It had been one of the loudest complaints of classical WITP players that the AI was not sufficiently challenging. Originally it was thought this widespread complaint could be mainly addressed via the medium of the new scripts. However a couple of years into the development it became obvious that the other AE improvements were unmasking even further the classical WITP AI limitations. To fix this necessitated throwing out the old AI and basically writing a new one which would better support the AE improvements and scripts. Needless to say, the available time to accomplish this task was less than the ideal.
To further complicate the AI development was the constant tension between making the AI better and players complaining they couldn't beat it. Yes the AI is not as strong as it could be because there are too many forum cry babies who simply can't handle being defeated by the computer. These cry babies are the posters who create the loop from hell:
- do anything to beat the computer (includes repeating until memorised the exact computer moves so that once learnt the human can take advantage of this knowledge, or redoing the turn until a new favourable outcome results, or using hindsight and disregard of historical political limitations to hit enemy weak points)
- then complain the AI is too weak
- when improved AI/script/computer OOB is provided, complain the AI cheats/is unhistorical, often then abandoning the game not accepting they have been defeated by the computer, (but had a human opponent similarly abandoned the game, would bask in the glory of having won as their human opponent had resigned)
- return to start of loop of doing anything to beat the computer
With all the unfair restricti8ons placed on Andy Mac et al, it is remarkable how well he has succeeded in making the AI challenging for those who don't engage in the loop from hell.