28-02-2003 02:24									#1
	AI Subordinates Support Apolyton, buy GURPS/ Alpha Centauri

In theory, the idea of having AI subordinates handle much of the work of running an empire's day-to-day operations in a 4X game makes a huge amount of sense. A national leader is most likely not his nation's most brilliant general, or its most gifted economist, or its most eloquent diplomat. Rather, he is someone who can find the right people for subordinate positions, work with them, and synthesize their expertise into policies that work. Capturing that in a game could free players from a lot of the tedium of managing every little detail.

But with the current state of AI technology, the almost inevitable fact is that top human players will be the most brilliant generals, the most gifted economists, and the most capable diplomats their nations have to offer. That completely demolishes any hope of making the game "more realistic" by having the player delegate responsibilities to subordinates, because the quality of subordinates the player would demand in real life is simply not available in the game.

Thus, the best we can reasonably hope for is AI subordinates that can handle complex orders from us, warn us in advance if our orders are about to get us in obvious trouble, and get out of our way when we want to take over something ourselves. That gives us as much help as the AI technology available is competent to provide without forcing us to put up with incompetent decision-making.

Now the question I'm curious about is how good a job MOO3 did striking that balance. I'll probably try the game sooner or later to find out for myself (in spite of how busy Civ 3 tends to keep me), but I'd be interested to hear the opinions of others who have played the game enough to think they have a good feel for the question.

Nathan

	28-02-2003 03:55									#2
	Support Apolyton or Terrorists Win

game is great... stop theorizing and start playing cya

	28-02-2003 04:57									#3
	Support Apolyton buy from Amazon

I don't have MoO 3 yet, but I can tell you that an AI should be better at micromanagement than a human, because micromanagement is just juggling numbers. The difficulty is in getting the AI to share your goals. For example, say in a game of MoO 2 that you're short on food. An AI should have no problem looking at your entire empire, and finding the population point to shift that causes the least loss to your industrial/scientific capacity. It might even decide that multiple shifts in both directions will increase both food and industrial capacity. The problem with this theoretical AI is that all it is doing is micromanaging, it doesn't see the big picture. Maybe you want the extra manufacturing on planet X even though manufacturing isn't particularly efficient there because you're expecting an invasion.

From what I've read here, understanding how to set the AI goals seems to be where people go right/wrong in dealing with the helper AIs in MoO 3. What I'd like to see on MoO 3 isn't another strategy guide (though I appreciate them ), but a "How to get your Viceroys to do your bidding" guide. Or maybe "101 things not to do because your Viceroys WILL[ do something bad because of carrying out implicit orders."