Dōbutsu shōgi
Shogi variants · all families · data
Also known as: Let’s Catch the Lion!, Doubutsu Shogi, どうぶつしょうぎ.
Gote (the second player, ‘white’) wins
| Result | Second player wins |
| Strength | Strong |
| Method | Retrograde analysis |
| Year | 2009 |
| Solved by | Tetsurō Tanaka (田中 哲朗), University of Tokyo |
| Source | Tanaka 2009 |
| Board | 3×4 (3 files × 4 ranks) |
| State space | 246,803,167 |
| Distance to result | 78 plies from the start (the theoretical win takes 78 moves) |
Solution history
Section titled “Solution history”| Date | Size or opening | Result | Strength | Credit | Source |
|---|---|---|---|---|---|
| 2009 | standard game | second player wins | Strong | Tetsurō Tanaka | source |
One row, derived from the record’s year, solver and primary source; no fuller history is recorded yet.
Engines
Section titled “Engines”- dobutsu (Clausecker) (engine): perfect play, source
References
Section titled “References”- Tetsuro Tanaka, 「どうぶつしょうぎ」の完全解析 (An Analysis of a Board Game “Doubutsu Shogi”), IPSJ SIG Technical Reports 2009-GI-22(3), pp. 1–8, 2009
- Tanaka original paper PDF (author’s University of Tokyo page)
- Dōbutsu shōgi (Wikipedia)
- clausecker/dobutsu — independent C tablebase and engine (perfect play from any position)
Resources
Section titled “Resources”Tablebases
- brianhliou/dobutsu-shogi — from-scratch Rust tablebase (333 MB .ctb via MPH)
- clausecker/dobutsu — C reference tablebase
Explorers
Write-ups
Dōbutsu shōgi (3×4 board, “Let’s Catch the Lion!”) was invented in 2008 by professional shogi player Madoka Kitao. It was STRONGLY solved by Tetsurō Tanaka and published June 2009 as a non-refereed IPSJ SIG Game Informatics technical report (2009-GI-22, No. 3). Using retrograde analysis, Tanaka computed the theoretical value (win/loss/draw) of every position reachable from the start. The initial position is a second-player win (後手必勝, gote / “white” wins) and the win requires 78 plies. The computation ran on a 2.6 GHz Opteron with 16 GB RAM; enumerating reachable positions took ~19 min. state_space = 246,803,167 is the count of REACHABLE positions and is the exact figure reported in Tanaka’s paper (“局面数が 246,803,167 と求まった”); it was independently recomputed from scratch by Liou. The often-cited 1,567,925,964 is the upper bound on TOTAL piece configurations ignoring reachability (after folding left-right symmetry). Note: capturing the opponent’s chick on move 1 hastens the first player’s loss to 76 plies; every other first move delays the loss to the full 78. The clausecker/dobutsu repo is an independent C tablebase reproducing perfect play and corroborates the result, but its README does not itself state the game-theoretic value, so it counts as secondary corroboration. brianhliou/dobutsu-shogi (by this list’s maintainer) rebuilt the retrograde → tablebase → probe → explorer pipeline and checked it against the clausecker reference (50,000 positions, zero mismatches).