The strange thing about the second opinion is not that it exists but that it works, most of the time, for most people. The mind that evolved for small groups and immediate dangers is now asked to manage a pension. It does its best. Put a number on risk and it becomes a target; put a target on it and the number stops meaning what it meant. Institutions know all this, of course, which is why the second opinion is so rarely left to chance.
Where the model breaks
Notice how rarely anyone is argued out of the second opinion, and how often they are simply moved past it by a change in circumstance. In the studies we have, the effect is modest, reliable and almost entirely ignored by the people it describes. The lesson is modest. Build for the person who exists, not the one in the model. It is tempting to call this irrational. It is more useful to ask what it is for.
