Sounds pretty logical. i havent analysed the current systems in any depth but it seems to me that the way they currently do things is with a 'generic' rule, where one building is always going to be built first (higher "rating") over another. what would be nice is if the actual benefit of that particular building, in that particular city, could be calculated. things like how long itll take, whether itll be worth it in the long run. for example, no point in building a superhighway in a city with only a couple of grassland/plains tiles nearby. other examples might include randomized or pre-determined (or even dynamic) preferences - for example, using higher shield production cities to build barracks and military units, or using those cities to build economy based buildings and trade resources etc.
Because c-evo is deterministic things like which cities are best for what can be figured out to a decent degree of accuracy.. but like you said it could get very complex. the idea seems to be getting much more in-depth rather then keeping everything generic, and the result would be having much more control over the ai.
So.. if more than one of us understands it, then im all for it. in fact, im all for it even if nobody understands it. heh.
Posted 31-Oct-05 13:10 by Dragonfly