Questions about a game in progress

[Quakka]

Hi all, first post.

thejeff said:
Well, if you can come up with a good, simple algorithm for "provinces you aren't likely to lose", combine it with the other obvious strategy of targeting provinces more likely to have sites of that path, weighted to keep everyone happy, I'd love to see it.

This is a pretty interesting problem. I'd suggest a simple fuzzy logic algorithm.

The basic idea would be to calculate a score for each province you control, based on n criteria. Each criterion would award points for that province, depending on how well the province matches it on a scale of 0 to 1. Each criterion would also have an intrinsic importance value, which would be used to scale it against the others. The actual points received would be (match) * (importance).

When the scores are calculated, the provinces are sorted according to score, and the highest scoring is selected. Provinces searched to 4 or more in the relevant paths or targeted by an overlapping search are excluded.

In my programming experience, these kind of things work best when you concentrate the weighing on a few things, and keep the rest as "tiebreakers". I'd suggest the following:

1: Ratio of friendly neighboring provinces versus enemy controlled ones. This would cause a tendency to avoid frontier provinces, or provinces you have captured with raids.

2: Previous searches. This would pick unsearched provinces over ones with 1, 2 or 3 level searches. A bit trickier to calculate for Aschaic Record, but possible.

Tiebreakers:

3: Probability of magic sites. Wastelands would get targeted over farmlands.

4: Probability of magic sites for this spell. Gnome Lore would pick mountains over other terrain and so forth.

5: Number of other sites already found. IIRC there is a maximum number of sites that can be found in one province.

It's not a perfect system, and never will be, but it'd be a lot better than going through the province list from top to bottom. Assistive features such as the auto search are not meant to make real decisions for the player anyway, just to remove the chore of making the trivial ones.