https://whamp.github.io/blog Meandering Curiosity en-us (Will) (Will) Wed, 06 May 2026 00:00:00 GMT https://whamp.github.io/blog/2026-05-06-calling-it-memory-made-agents-better-at-memory https://whamp.github.io/blog/2026-05-06-calling-it-memory-made-agents-better-at-memory Napkin was already strong on LongMemEval_S. Renaming the model-facing surface from napkin to memory improved one Qwen3.6 run from 87.6% to 90.6% accuracy, with the biggest gains on multi-session and temporal reasoning. Wed, 06 May 2026 00:00:00 GMT (Will) agentsmemorylongmemevalbenchmarksnapkin https://whamp.github.io/blog/2026-06-13-i-tried-to-disprove-the-tdd-benchmark-i-reproduced-it https://whamp.github.io/blog/2026-06-13-i-tried-to-disprove-the-tdd-benchmark-i-reproduced-it I expected a better TDD skill to beat Kun Chen's ProgramBench result. It did not. A behavior-first TDD variant tied the original TDD arm and still trailed the no-TDD control. Sat, 13 Jun 2026 00:00:00 GMT (Will) agentsbenchmarkstddcodexprogrambench