Tag
A reflection arguing that in multi-model setups, the consensus output is less valuable than the disagreements, which reveal genuinely contested parts of a problem. The post questions whether consensus should be the goal and how to distinguish productive disagreement from noise.