What Chess Engines Miss When Humans Explain a Game
Andy Lee reflects on annotating five games and seven fragments from two Chess Olympiad rounds under a compressed deadline. Using Stockfish to locate critical moments, he describes choosing memorable narratives, judging moves within match context, resisting impractical computer variations, and favoring short human lines. Examples from Caruana, Aronian, Abrahamyan, Niemann, and Wesley So show annotation as translation rather than raw evaluation.
Opens on litandchess.substack.com · Curated by GlobeRead
GlobeRead's Take
Powerful analytical tools create a curious editorial problem: they can show more than a reader can use. Chess engines expose hidden tactics with astonishing precision, yet precision alone does not explain why a human chose a move under time pressure, team obligations, and incomplete knowledge. As machine assistance spreads through writing, research, and creative work, the craft of selecting and translating becomes more valuable, not less.nnLee’s account gives that craft a memorable formula: human games, machine variations, human explanation. The “battle of the buried bishops” in Caruana–Ivanchuk turns two positional decisions into a narrative a reader can retain. That choice is not decorative simplification. It is an argument about attention: among thousands of possible lines, the annotator must identify the pattern that reveals the struggle without pretending the reader wants a database dump.nnA second tension appears when engine judgment collides with competitive context. Aronian’s 34…f5 was not the computer’s preferred defense, but it kept winning chances alive when the team situation rewarded activity. Lee’s restrained annotation acknowledges both evaluation and intention. His treatment of long tactical lines follows the same principle: show the mating net or the compact queen sacrifice, not every branch required to prove it. Accuracy remains essential, but relevance determines which accuracy belongs on the page.nnWe picked this essay for chess players, editors, teachers, and anyone using AI-assisted analysis. It offers a candid look at the decisions hidden behind apparently authoritative commentary, including the humility required to judge stronger players from a comfortable desk. The piece also reminds readers that explanation is not the residue left after computation; it is a separate act of understanding. That insight travels far beyond chess, especially wherever automated output still needs human judgment, audience awareness, and editorial restraint. When a machine can calculate everything, who decides which truth is worth noticing?
We curate the internet's most interesting articles — science, psychology, history, philosophy and more. Always free, no algorithm.
