🚀 go-pugleaf

I assume the GOBANG game you are talking about is the one I did write
some time ago. The 'Official' option has never been included. Neither
have I made any other improvement. One of the reasons is that I don't
really know what the 'Official' rules are. (Somebody told me they
existed but I never learned about them.) So if somebody can tell me ...

I must say that I myself am also quite surprised how well the program
plays. Maybe some history. Quite some time ago I implemented the well
known game of putting stones on piles and getting four on a row. The
technique used was straightforward backtracking with only limited
cleverness. A friend of mine told me wheter I could do GOBANG in the same
way. My reply was that this was a lot harder because the number of
possible moves is very big and, hence, backtracking won't work. Because
he continued asking I decided to program the thing completely without
backtracking, just looking at the current situation. I was convinced
that this would result in a program that was trivial to beat. When I
got the first working version ready and started playing for the first time,
to my surprise I was defeated in a few moves. Also the second and third
time the program did win. Since then, only minor changes have been made
to the strategy. Only in the top two level backtracking is added, and only
in the case the computer can make rows of 4 (forcing the player to a
particular move). Also some randomness was added to avoid similar games
all the time.

Since I wrote the program I beat most humans as well. the strategy of the
program is so easy that you can play it by hand as well.

Mark Overmars

P.S. there is a UNIX version of the program as well, although it is an older
version that is a bit less clever. It uses CURSES to handle the screen.