The study of endgame tsumeshogi—the art of composed checkmate problems in shogi—has evolved through several distinct paradigms, each reflecting the broader strategic and technological currents of the game. The earliest systematic approach emerged during the Edo period (1603–1868), when professional shogi players and composers created collections of practical checkmate sequences. This Classical Tsumeshogi paradigm emphasized elegance, brevity, and direct applicability to real-game endgames, with canonical works such as the "Shogi Zukō" (1716) establishing the foundational conventions of problem composition.
In the 20th century, the Modern Tsumeshogi school arose, driven by composers like Kōzō Masuda and later by the competitive shogi world's need for deeper training materials. This paradigm shifted toward longer, more complex problems that tested not only tactical vision but also strategic foresight. Modern Tsumeshogi formalized conventions regarding piece drops, repetition, and the use of all pieces, creating a distinct genre that separated composed problems from over-the-board endgame play. The school also saw the rise of problem-solving competitions and the codification of difficulty levels.
The late 20th and early 21st centuries brought the Computer-Assisted Tsumeshogi paradigm, as shogi engines and endgame databases enabled composers to verify correctness and explore positions beyond human calculation. This period saw the generation of problems with unprecedented depth and the systematic cataloging of tsumeshogi patterns. Subsequently, the AI-Driven Tsumeshogi paradigm emerged with deep learning and neural-network-based engines, which not only solved problems but also suggested novel compositions and refuted long-held assumptions about checkmate sequences. This paradigm has blurred the line between composed problems and engine-generated endgame analysis.
Today, endgame tsumeshogi is integrated into the broader Tsume and Endgame Theory framework within shogi strategy, serving as both a training tool for players and a field of study for AI researchers. The paradigm continues to evolve as engines become more powerful, with ongoing debates about the role of human creativity versus machine-generated problems. The historical arc from classical elegance to AI-driven complexity mirrors the transformation of shogi itself, making tsumeshogi a microcosm of the game's strategic evolution.