Chess Aesthetics Lab

Explore how Iqbal’s three models score a chess composition.

Test Positions

Use a FEN for either side to move, without castling or en passant rights. Automatic search looks for a forced mate whose shortest length matches the selected depth.
Feature guide · P1–P7 and T1–T10
Scoring method, source code, and references

The app uses full precision for intermediate calculations. The Loyd example scores 6.675 in the 2008 model. The thesis reports 6.674 using three-decimal rounding at intermediate steps.

The 2010 and 2012 results are expected values from this implementation’s top-k aggregation. The 2012 model scores the main line.

Source code: scoring engine, web adapter, interface, and worker. Scoring engine SHA-256 5acf94bb654b30c0de648552e89e79471d213a3821cd146ff2898d054f478744.