Prior Belief

Method

Where these numbers come from — honestly

The archive launched seeded

A network like this has a cold-start problem: the first visitor has nothing to look up. So the archive was populated with synthetic capsules — currently 6,001 of them, against 0 written by real people. Every synthetic entry is flagged in the database and labelled in the interface. As real capsules arrive they progressively outnumber the seeds, and every chart reflects that automatically.

The seeded outcomes are not invented

They reproduce the best-replicated finding in the psychology of regret: over short horizons people regret their actions, and over long horizons they regret their inactions — the temporal pattern described by Gilovich and Medvec (1994, 1995) and repeatedly replicated since. That is why the two curves start apart and cross.

What is measured rather than assumed

The curves are Gaussian kernel regression over the actual rows, and the crossing month is found by scanning for the sign change. Nothing is hard-coded. Where fewer than about three observations sit under the kernel, no figure is drawn at all — the chart declines to state a finding it cannot support.

What we will not claim

Your own data

What you write when sealing a capsule is never shown to anyone. Only the single line written afterwards can become public, only if you choose that, and only after a human has read it. See Privacy.