Nerdy Doings

[tracher]

In reference to my previous Scotland Yard post...

So I decided to give it a start.  I spent way too much time on a fancy map editor, and learned all sorts of things about the game in the process:
The board I have must be very old (1985).  There's no 108 on the board.  Later editions fixed this by getting ride of 200 and shifting a few numbers

Comments

[zanfur]

Have you considered using a simple alpha-beta pruning tree, using "possible movements of fugitive" as an evaluation metric to minmax? There is a *huge* multipliciy of moves for each "detectives" turn (I just consider the detectives to have one large "supermove"). I think, in general, an evaluation function that minimizes the number of potential moves the fugitive has is a good start

[tracher]

I was thinking much along these lines, but I'm not sure on the specifics of how to coordinate 5 detectives in such a maneuver. I think this coordination is one of the more interesting problems for both AI and real players.

[sonofzeal]

When using the defined set of starting locations, do you have the detectives work out what spots Mr. X could possibly be in?

[zanfur]

If you're asking me, no, I don't bother. It rapidly becomes "the whole board" after the first couple turns. I treat that as "unknown" and head for the underground stations.