Page 1 of 1

Andrew Brown's maps and AI

Posted: Mon Oct 27, 2008 6:03 pm
by Dereck
Reading the notes included with Andrew Brown's maps it states that the Allied AI may not recognize the new bases. Does that also go for the Japanese AI if one wishes to use the new map and play against the Japanese AI?

RE: Andrew Brown's maps and AI

Posted: Mon Oct 27, 2008 7:40 pm
by rockmedic109
I believe so.  I don't think Andrew did anything to the AI, so the AI would not know anything about the new bases.  I think I read one of his notes that confirms it, but I cannot find it now.
 

RE: Andrew Brown's maps and AI

Posted: Mon Oct 27, 2008 8:12 pm
by Dereck
Then I suppose my next question would be if there weould be anyway to get the AI to recognize the bases or would that be too hard or just impossible?

RE: Andrew Brown's maps and AI

Posted: Tue Oct 28, 2008 3:00 am
by el cid again
It is more complicated than that.

What Andrew probably means is that the AI is hard code script - "this unit appear at base x" and "that unit go to base y" - and in that sense - no - there is nothing you can do about it.

But AI DOES recognize bases for some purposes - and will for example tend to feed them and defend them.

RE: Andrew Brown's maps and AI

Posted: Tue Oct 28, 2008 4:33 am
by Andrew Brown
ORIGINAL: el cid again

It is more complicated than that.

What Andrew probably means is that the AI is hard code script - "this unit appear at base x" and "that unit go to base y" - and in that sense - no - there is nothing you can do about it.

But AI DOES recognize bases for some purposes - and will for example tend to feed them and defend them.

Yep. That's about it.

To elaborate on the original question, neither side will specifically target the new bases for attack because they are not in the AI "target lists" of bases, which are in the game code and so not moddable (unlike AE).

I mentioned the Allied AI specifically, as many of the new bases are in outlying parts of the map, and I didn't think there would be many players that had problems with a Japanese AI seriously threatening Papaete or Panama. Most people don't have great problems in defeating the Japanese AI, or at least containing it to the historical Japanese boundary of conquest. The problem is greater for the Allied AI, because a human Japanese player is much more likely to threaten these outlying areas when playing vs Allied AI.

This is mainly why I state than scenarios using my map are best played against Japanese AI (or PBEM of course) as opposed to against Allied AI.

Andrew

RE: Andrew Brown's maps and AI

Posted: Tue Oct 28, 2008 12:39 pm
by el cid again
To which add that Joe Wilkerson said somewhat adamently that AI is never able to play the Allies - even in stock. I have run hundreds of AI vs AI tests - some for years of game time (which is pretty fast when automated) - and I cannot contradict him. I say the same thing Andrew does - except more so: RHS cannot be played vs Allied AI. Even working for far too long on AI oriented scenarios only yielded modest results - it works OK for a few days - then poorly - and eventually breaks down entirely in late 1944. I think it is almost tragic - but it is expensive to work out a real AI. One programmer and I have a dream - we might create a utility to wash each turn through which fixes the AI - be even then - it will be AI as Japan. Trying to do the Allies is daunting - and would be a second generation project. We estimate a year after AE is out we will START to work on that utility - and it will require another year or so to write. Which means it may never happen.