[FarAway Pretender]

Mercurycs, I play Dom2 exclusively single-player (just don't have the time to make a commitment to play MP with folks I've never met, and am still trying to get friends to play it with me!), and I am finding, after about 9 months of an hour or two a day that I'm only finding the limits of the AI. Some of that came from reading stuff here where people talk about the limits of the AI, I probably never would have figured it out on my own.

The gameplay is different enough for each faction (and each Pretender design) that there's a significant learning curve, which is fun whether you're playing AI or not.

There are certain tactics in use among MP (primarily involving a "Rush" to get certain spells and/or magic items fast) that the AI doesn't rely on. But I'd play it yourself and see if any problems with AI performance slow you down.

Bottom line: Most good AI is not a function of programming, it's a function of game design. Game systems built to allow easy modeling and decisionmaking by a computer allow for good tactics. On the other hand, computers have a hard time modeling wacky and non-statistical combinations and permutations (e.g., "Am I better off casting Haste and Giant Strength on the same unit? What if it's a flyer?").

Just one example from another TBS: In Heroes of Might & Magic 3, the AI did a lousy job with spell selections. BUT, if I faced a Barbarian opponent (Barbarians don't use spells, but rely on combat troops that are more magic-resistant than others), the AI "got" a lot smarter.

One benefit of not allowing super-scripted combat is that the AI doesn't enjoy such a huge handicap in combat. I'm hoping that, in Dom 3, the designers might be able to tweak the Castle-building system in a way that subtly changes game play, but allows for more gradual castle building (and thus allows for more effective modeling of Castle-building decisions in the AI). At this point, that's just a dream of mine, though, and I don't think it's shared by that many.