adhd Publishes Cost-Adjusted Ideation Benchmarks — 5.2× Better Trap Detection at 2.3× Latency and 1.9× Tokens

GitHub·medium signalXBlueskyLinkedInCopy link

UditAkhourii/adhd added 983 stars this week to 2,478, implementing tree-of-thought with an architectural fix for premature convergence: a divergence phase spawns N isolated parallel agents under 15 distinct cognitive frames (economic-incentive, async-control-surface, gamification, redundancy-race and others) with zero shared context so branches cannot anchor on each other, followed by a focus phase where a critic scores on novelty, viability and fit, flags traps with reasons, clusters by pattern and deepens top-K survivors. Across six open-ended engineering problems it reports breadth 9.0 vs 4.83, novelty 7.83 vs 2.67, trap detection 9.50 vs 1.83 and actionability 9.50 vs 6.50 against single-shot baselines — and, unusually, publishes the cost alongside: ~2.3× latency and ~1.9× token usage from independent reviews. MIT.