What looks like risk from the outside is usually something simpler from the inside: tiredness, a deadline, a wish not to look foolish. The honest conclusion is uncomfortable for reformers: people are not waiting to be corrected. The lesson is modest. Build for the person who exists, not the one in the model. There is a version of this essay that ends with a policy prescription. This is not that version; the mechanism is interesting enough on its own.
What follows
The remedy is rarely exhortation. It is a change to the situation so that the easier choice is also the better one. A better account of risk would begin with the ordinary case rather than the laboratory one. The strange thing about the queue is not that it exists but that it works, most of the time, for most people. The measure that matters is not what people say about risk but what they do when nobody is scoring.
