open-source research

The 'More Agents = Smarter' Myth — A Detective Experiment Debunks It

12 small AI Agents + a reasoning hub = 96% cases solved; Agent self-voting only 38%. A cold shower for the 'more Agents = better' narrative.