Perhaps I am the geek among geeks that finds this sort of thing interesting, but AI Wars is awesome. To explain what it is here is a post by the designer from over at the Wargamer Forums:
"Hey guys,
As the creator of the game, just thought I would chime in on a few points. KevlarSocks makes some interesting points about Warcraft I versus Warcraft III, and I couldn't agree more with the analysis of those two games. Warcraft I didn't have very good AI, and so it used extra units in order to fake an AI that was better than it was, to keep up with the human opponents. A number of other games include "cheating AIs" that do much the same.
However, the entire premise of AI War is different from those titles. Those titles presume that there are two sides, and they are overseen by a central intelligence, and those two sides each try to best one another by starting with mostly identical -- or at least equivalent -- forces. It's a battle of wits against a similar-ish opponent, basically. As you say, that's the premise of Chess as well, and since it's such a discretely defined sort of game and has been around for so long, you get AI that is very good at just crunching up all the possible paths in that game and coming up with optimal moves. Other turn-based games like Galactic Civilizations II and similar are also able to get a lot of mileage out of human-proxy-style AI based on their turn-based natures, even though they don't reach the sophistication of Chess.
Then enters the RTS genre, with a plethora of games that all largely try to simulate human opponents, but in realtime. There are endless problems with trying to make AI that is very good in this scenario, but the end conundrum is this: at best, the AI will perform like an adequate human opponnet. So, therefore, I might as well just play against a human opponent if there is one; the only reason to play against the AI is if either a) I am not very good at strategy games, or b) I don't know anyone to play against.
I don't know about you, but to me those seemed like pretty lame reasons to code an AI. In film and books there are countless examples of interesting fights that are one-sided, or long odds, or which have some sort of inequity that leads to interesting circumstances. In Chess you only get that sort of scenario if you play poorly and are down material late in the game, struggling to come back. Same with most other strategy games. What I wanted to do was take RTS mechanics and make something completely new with it -- not a battle of wits between likeminded, like-powered opponents. But rather, a David and Goliath scenario. Ender vs. the Buggers. Starfleet vs. the Borg. Humanity vs. the Terminators. Us versus those aliens in The Forever War, whatever they were called. I'm told that it's an apt metaphor for Battlestar Galactica, too, though that's still on my need-to-get-to-that list. Pick your favorite: you can't get that sort of scenario, not really, within the bounds of a traditionally-designed RTS, for the simple fact that none of those scenarios have anything to do with a battle of wits between even numbers of units.
Anyway, so the bottom line is that, yes, the AI outnumbers you literally 10,000:1 at the start of the game. But no, it's not a cheating AI any more than a FPS level is cheating because the bad guys with guns outnumber you 100:1 or more. In Left 4 Dead, you might kill 2,000 zombies in one campaign with your four survivors, but that's not a cheat: it's a scenario. In AI War, you might kill upwards of 100,000 enemy ships at a cost of nearly that many of your own (if it's a long game), with perhaps 40,000 enemy ships still surviving if you win. It's not about even numbers, it's about picking your battlefields and engaging the enemy sub-commanders that you come across at each one. It's about guerilla warfare and grand strategy in a hostile galaxy. It's about being on the wrong end of things but prevailing nonetheless.
That's hard for some people to accept, in the sense that to them a strategy game is defined by being a battle of wits between similar opponents. How you can ever know who is "better" at the game if you don't have two equal starting sides, after all? If you're looking to know whether you are better than the AI or not, AI War will certainly disappoint from that sense because it's an apples to oranges comparison. But, if you're looking to find out if you can come up with clever solutions to what can be very complex scenarios, if you can beat the odds and stamp out the buggers/terminators/borg/whatever, then that's what AI War is all about simulating. AI War and Warcraft III are two very different sorts of game despite the RTS trappings that make them seem more similar than they are, at least at first.
If you want a ton of more info about the AI, you might like this article series I wrote (the part on asymmetry perhaps being the most relevant to this discussion, although part 1 also is quite relevant). If you're a member at AIGamedev.net, I also did a two-hour interview (one of their longest ever) that was very well received. From a lore/story sense, our wiki has a section on How Does the AI Think? that you might be interested in.
Design decisions like a hugely asymmetrical AI are the sort of risk that a big studio would never take, because it's impossible to tell how people will react. Some people think of it as being revolutionary and amazing, others can't really tell the difference at all, others think of it as a cheat. I guess it all depends on what you look for in a strategy game. Fortunately, the overall reaction has been quite positive, despite the fact that no game is for everyone. If anyone reading this is unsure, I suggest the demo; it's a great intro to what is going on here. Thanks for your support!"
