Hello there! If you follow the AI forum here, you may have seen ags1’s thread about Hnefatafl, a Norse board game popular from the British Isles through Northern Europe from the 5th century through the 11th century. It has seen something of a renaissance lately, and as a student of the game and a hobbyist student of games and artificial intelligence generally, I thought it was high time that tafl games join the club comprising abstract strategy games with AI tournaments.
You can read my (long) intro/rules/FAQ post over at my website, or I can sum up here.
-
What’s tafl?
It’s an asymmetric board game. The king and his defenders, at the center of the board, are outnumbered by the attackers, who start around the edges. The king must reach a corner space, and the attackers must stop him from doing so. (There are a large number of variations on the rules, since no single version survived. The most common variants are played on 7x7, 9x9, and 11x11 boards.) -
How does the tournament work?
You write an AI to plug into OpenTafl, the tafl engine I’m working on, using this protocol, which in turn uses this notation.
Your AI is due by December 15 of this year, and the tournament will play out over the next few weeks. AIs will play two-game matches, one as each side, to account for imbalances in the game. There will be a round-robin stage to establish ratings, then a single-elimination tournament seeded by the ratings established during the round-robin stage. If I can convince a high-level player to play, there may be an AI-vs-human exhibition match at some point after the AI tournament.
The rules are a slight modification of a well-known tafl variant. They’re on the page linked above.
There are prizes.
- Anything else?
Nothing I feel like I have to reproduce here—hopefully, this is enough to pique your interest. Thanks for your time. I hope to see you on the field of battle!