Oriel — AI & technology worth noticing, every day.

Solving the fox game

Web page · https://blog.sesse.net/blog/tech/2026-11-13-19-51_solving_the_fox_game.html

First seen by Oriel on .

AI summary

Steinar H. Gunderson describes a roughly 1,000-line, dependency-free solver for the Fox Game and reports a complete result under the stated rules: sheep win with perfect play.

  • The analysis uses Christoffer Stjernlöf’s rules, omitting forced fox jumps, multi-jumps, and the 40-move draw rule.
  • Sheep need 19 pieces for a forced win, or 18 when allowed to move first; nine sheep can force a draw, while 11 suffice against one fox.
  • Game states use two 64-bit bitboards plus a side-to-move bit; manually entered move tables define sheep moves and fox captures.
  • Combinatorial ranking, left-right canonicalization, and memory-mapped two-bit results represent more than 2^41 states in 329,870,110,455 bytes.
  • Iterative propagation across 64 cores required 40–80 passes per chunk and about 36 hours; retrograde analysis might have been ten times faster.

#fox game #game solving #bitboards #tablebases

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Points

Score reported by the source.

  • Lobsters: 24, latest observation ; +23 over observed period since

Comments

Comment count reported by the source.

  • Lobsters: 5, latest observation ; +5 over observed period since

Aliases

SourceExternal IDURL
Lobstersq81fi4http://blog.sesse.net/blog/tech/2026-11-13-19-51_solving_the_fox_game.html Read discussion