The Netflix documentary "Untold: Chess Mates" premieres April 7, 2026, reigniting the biggest question in competitive chess: did Hans Niemann use engine assistance against Magnus Carlsen?
The public debate has relied on opinions, accusations, and platform algorithms. What's been missing is an independent forensic analysis using behavioral signal detection — not just average centipawn loss.
Why ACPL Alone Doesn't Work
Most cheating discussions focus on ACPL (Average Centipawn Loss) — the average error per move. The problem: a SuperGM playing a quiet positional game can post an ACPL under 10 without any assistance. ACPL measures accuracy, not behavior.
Forensic detection requires analyzing how errors are distributed, not just how many there are.
The Signals That Actually Matter
Modern forensic analysis uses 75+ analytical signals across 35 behavioral dimensions:
-
Error Tail Shape (Pareto Alpha)Humans produce heavy-tailed error distributions — long clean stretches punctuated by sudden collapses. Engines produce thin, uniform tails.
-
Crisis CorrelationHuman accuracy degrades under tactical pressure. Engines maintain constant performance regardless of complexity.
-
Sequential PatternsHumans break precision chains through tactical oversights. Engines produce unnaturally long consecutive optimal moves.
-
Recovery DynamicsAfter a blunder, humans tilt — their next moves degrade. Engines snap back to perfect play immediately.
What Does an Independent Analysis Look Like?
A full forensic report evaluates every move against these behavioral dimensions and produces three independent scores: Engine Likelihood, Human Authenticity, and Assessment Confidence.
Every verdict is fully explainable — not a black box. Every flagged signal includes the measured value, the human baseline, and the statistical significance.