BEER runs contests of algorithms. The owner, Marcos D., describes a game, allows people to send algorithms that choose a move from any given position, and then makes the war begin. Each algorithm plays 100 games with any other algorithms, gaining 1 point for each algorithm it defeats more than 50 times.
I love the idea, and I hope one day people will compete to find the best "LUA_AI" for abstract games implemented in Nomignolov!
I decided to compete for the current game ("Intervalo"). With the help of Nomignolov. Together with my sister, we thought of 11 different algorithms, and I used Nomignolov to see which one was our strongest warrior.
It was a bit tricky, because:
- players in Intervalo move simultaneusly;
- the algorithms choose a move instead of evaluating a position;
- Nomignolov does not support repeated games at present.
At the end I was able to select one algorithm (it was one of my sisters') and sent it.
More details later, but the big news is that Nomignolov started doing what it was designed for: study abstract games and collect data (still, someone may use it to play...).