The protocol handler is complete. Enough of the XBoard protocol has been implemented to play a complete game. Move selection is completely random. A 20,000 game match was run between identical copies of the program (which didn’t take long since it moves instantly), and as predicted the outcome was almost exactly 50%. Interestingly, 85% of the games resulted in a draw.

The next area of focus will be implementing the evaluation routine, which gives a static analysis of a position. This will be a direct port of the evaluator from Prophet3. The evaluation needs a lot of work, but forward progress will come after the rewrite is complete. Once the evaluator is in place I can make the engine just slightly smarter than a complete random mover. I’m also considering a JNI integration with chess4j, as a proof of concept for future work.