Move Ranking and Search-Free Playing

Abstract

Quick and effective ranking of moves based on their rough, initial estimation can be used either in a stand-alone mode, i.e. in a search-free playing program, or as a training tool for human players, or - most frequently - as a preliminary step before more expensive analysis is performed. In particular, initial sorting of possible moves according to their potential strength is often used in alpha-beta pruning algorithms, thus leaving more time for exploration of the most promising paths. Certainly in competitive tournament play, due to time constraints, efficient move pre-ordering must be either implemented in a search-free way or rely on shallow search in order to devote the remaining time to deeper, selective search.