Tool-assisted game movies
When human skills are just not enough

Submission #2768: Weirwindle's NES Battle Chess "Fastest Mate" in 04:19.54

Console: Nintendo Entertainment System
Game name: Battle Chess
Game version: USA
ROM filename: Battle Chess (U).nes
Branch: Fastest Mate
Emulator: (unknown)
Movie length: 04:19.54
FrameCount: 15598
Re-record count: 40
Author's real name: Timothy Cookson
Author's nickname: Weirwindle
Submitter: Weirwindle
Submitted at: 2010-07-30 01:24:27
Text last edited at: 2010-08-01 06:23:50
Text last edited by: Weirwindle
Download: Download (1396 bytes)
Status: decision: cancelled
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:
The priority of this run is to finish a game of chess against the default level opponent in the least number of moves possible.


  • Emulator used: FCEUX 2.1.4a
  • Aims for fastest time to checkmate
  • Uses a suboptimal opening move to be made so that any opening books preprogrammed into Battle Chess will be obsolete.
  • Manipulates computer analysis through blunders
  • Minimizes piece exchanges as to cut out fight scenes

My first tactic was to open with a move to quickly render the engine's opening book void. There were several candidates for this including the moves 1. g3, 1. h3 etc. Using these lines I opted for quick advancement for one of my bishops and queen due to the large amount of board that they need to cover quickly.

During the first tryout of this game I was able to checkmate in less than 12 moves using Rybka 4 (the best chess engine software to date) and repeatedly trying to blunder the opponent until a quick checkmate.

One thing that limited and quickened a last exhaustive search for a win is that the opponents choice of a move never changes for a certain situation. I limited my search tree to 5 branches per move (choosen by Rybka) so 5 choices ^ 8 moves = 400000 possible lines.

I pruned out those moves that held many piece exchanges, those that did not advance either the queen or bishop, and those that led to useless moves like castling my king or advancing both knights, or advancing more than 3 pawns. This left me with about 13 hours of searching by hand with a little lua script help to take away action scenes.

When I was finished I had three possibilities for a mate in 8 moves. One consisted of 4 piece exchanges() and the other two had 3. Those two were

1. g3 e5 2. Nf3 Nc6 3. d4 Nxd4 4. Nxe5 Bc5 5. e3 Nf5 6. Bc4 f6 7. Qd5 fxe5 8. Qf7# in 15883 frames and 1. h3 e5 2. Nf3 Nc6 3. d4 Nxd4 4. Nxe5 Bc5 5. e3 Nf5 6. Bc4 f6 7. Qd5 fxe5 8. Qf7# in 15598 frames

Any room for improvement will be by increasing the number of choices per move and a lot of lua scripting.

Similar submissions (by title and categories where applicable):