Mockylock

First impressions: Really nice! Played the demo till late night. It will be hard to wait for the final product. Played the d-day scenario, it was like going back to the good PG days. One thing though... It would have been nice if you could see your opponent's units which was revealed during his (human or AI) turn, in your own turn. So that you don't have to follow his turn in order to remember where the air fleets are. And thereby make in easier to do the alt-tab a.k.a. gf-lurking-around-button to a greater convenience. But it's a little thing Keep up the good work! Glad I preordered

How do you guys win that overlord scenario?? I can get a favourable draw almost each time. But a win.... Can't do it time. I can get russia surrender, but not UK and definately not US. What I do is to move all AFs west and take out allied Afs. And almost all armies west too, to deal with the landings. Meanwhile I send HQ and two german corps to protect Finland and bind USSR troops there. Then swift east and beat USSR. Italy is defended by 3 corps. Is that a totally wrong approach?

Defender of the crown (1987, commodore 64) http://en.wikipedia.org/wiki/Defender_of_the_Crown

I'll give it a try (j_nygren@email.dk)

I'll give it a try (j_nygren@email.dk)

You are right Kuniworth, that was very cute indeed.

What Gavrok said. In particular point 4. Retreat should be done manually in your own turn, IMO.

I dont know about the naval ratings. But I like the idea of differentiate HQ's on more than one attribute. I first thought about: Attack rating Defence rating supply rating And my guess it that implementation wise this should not be that much work. A simpler version could be, keep it as is, but give some or all the generals some kind of special ability. Fx lower Eisenhower overall rating but give him a bonus to his supply value. Give Kesselring a bonus when commanding air units or when fighting defensive trench warfare. And maybe some attack bonus for Rommel. Guderian (whom should be included) might get a hige panzer bonus ... and so on. That would be a simple way to differentiate the generals (game wise and historically) and put some more thought to which one to buy.

hehe no, I havn't played Rambo. But actually the AI could benefit, seeing that Rambo is totally random, it would maybe develop a more cautious playing style against him? The point is, to have the AI act as the human brain, reacting to pattern recognition in the observed environment. To Hubert and alike -> Neural Networks, Evolutionary algorithms etc...

Yes I agree on that - only games played after patches have been applied should be logged. Imagine if all top player games after SC1 patch 1.07 were logged and analysed - I am pretty sure AI performance would increase significantly. Even player styles could be found, and used by the AI to counter a specific player. Etc.. Edit: SC1 was really great, the best game I have ever played, maybe except civilization. So I will buy SC2 no matter what. Very good job Hubert. [ December 06, 2005, 09:05 AM: Message edited by: Mockylock ]

I agree that it almost impossible to create a good AI in a complex setting. That is why you need to have the AI learn from experience/already played games (as stated in another thread). If not, it's just a matter of time before our 'intuitive brains' have learned the underlying AI function, and then the AI is toast for good. The better the implemented AI, is just a matter of delay on our learning of the AI (in the no-learning case). Therefore there is only one way as I see it - give the AI information on what strategies leeds to victory, and take actions accordingly (which is log the games, and do pattern recognition on the strategies etc). So, when a new strategy paradigme between players occurs which will lead to a lot of victories, the AI will also learn it and (in time, based on the frequency of observations) react accordingly. It is very complicated to implement yes, but the theory is there. And I personally would really like an AI that evolves over time. Not saying the AI would be perfect - but definitely better - and most important more intuitive and better suited to react on new strategies (and current strats) I am no AI expert, but I beleive the concept is already implemented in shoot 'em up games and in games like AOE (but there the AI only learns in-game (I think) and that is a different story. Sorry if there are some overlap with the other thread.

There seems to be a lot of discussions on the board how to implement the AI in SC2. But as I see it, the general consensus is that the AI should be event driven, that is, some predefined strategies triggered by some predefined propabilities. And then I read something about 'A program can never have intuition like humans', think it was JerseyJohn. At first I agreed, but now thinking about it, I believe there is a way of giving the AI (at least some) intuitive skills. The idea, on a more abstract level, is that the AI has to learn from its experience. But the only way it can do that is to observe games (or part of at least). So, in order to do that it would be great if there were an option in SC2 to automatically upload games when played to some server/database. Here a program could analyse the games, with some pattern recognition techniques, and then learn from these observations. Then if this was 'checked' in the option panel of SC2, the AI could ask the program on the net for guidance on what to do in some situations. With some clustering techniques the players (if every player had a unique player id) and/or stragtegies could even be segmented, and the AI could use that information to apply the most suited strategy. There are several ways an AI can learn from these logged games. A simple example could be to optimize the free parameters in the scripted AI, e.g. the propabilities for taking certain tasks/strategies when the predefined events occur. This was just an idea, and not something Hubert should go into before release, but IMO I believe it has potential. For the solo players it would be great with a continously adaptive AI who gets better and better (as more games are logged). And the program that analyse the data could be upgraded independently of the game, making it more sophisticated progressively over the time. The idea is partly based on the way they made the chess program Deep Blue, which analysed more than 700.000 grandmaster games in order to optimize the evaluation parameters of game board positions. Just an idea - what do you think? Jakob

Just email: j_nygren@email.dk

Are you really all tired of this game? I think it is a great game... Have played against the AI since 2002 (or 2003), and have recently begun playing humans. I think it is great, and I think I'm quite addicted to it. What we need is some more AARs! The recent Zapp-Terif game#1 was really exciting, do some more of that. Try some new systems (no tech game and/or blashys new high US/Russia readiness systems or whatever). If this forum dies... where can I find some AAR's? And I'm not that good at german (panzerliga)

I think this new system is a good idea... The huge bid-systems just seems ridiculous, and makes the game very static in the east. I have started a game where I play axis with russia+usa readiness at 60% and bid 120 at 1-5-10. We are both equal intermediate (I think) players... So far interesting..