This document describes using genetic programming to solve chess problems related to the King-Rook-King (KRK) endgame. Specifically, it evaluates genetic programming's ability to learn basic chess rules and the number of moves to checkmate in various KRK positions without direct programming of chess knowledge. The author defines the "grand KRK problem" as predicting the number of moves to checkmate for any given KRK position, and "petite KRK problems" as classifying positions by number of moves to checkmate. Genetic programming is able to solve some petite KRK problems and shows promising but limited results for the grand problem, suggesting this approach may be able to learn chess-like rules from examples.