A sizeable minority endorse a plainly wrong judgement when a unanimous group asserts it; one dissenter breaks the effect.
About 37% of Asch's subjects conformed to a plainly wrong line-length judgement. The finding that matters more for institutional design: a single ally dropped conformity by roughly three quarters.
In practice
- A new analyst spots the error in the forecast, sees six colleagues nod it through, and says nothing.
- One person on a jury says "I'm not sure" and three others discover they were not sure either.