Estimating how long a phenomenon will last

A forecast built from age alone ought to be embarrassing. Instead, it is often the least embarrassing number in the room. In 1969, Richard Gott treated his arrival at the eight-year-old Berlin Wall as a random point in its life and guessed another eight years.

The Wall lasted twenty more.

That miss is respectable because the method knew almost nothing. It used one fact—age—and produced a baseline against which cleverer forecasts could embarrass themselves.

Age tells us that something has survived this long. Old institutions, technologies and habits have already endured shocks that killed their forgotten rivals. We usually notice a thing when attention is unusually high: launch, scandal, rapid growth, apparent collapse. Calling today an ordinary midpoint can be a heroic assumption dressed as arithmetic.

Known facts also win. A licence with an expiry date overrides a lifetime heuristic. A component facing withdrawal changes the calculation. A wall surrounded by political pressure contains information that its birthday cannot summarise.

The useful comparison is almost impolite in its simplicity. One forecast says the new platform will dominate for fifty years because it changes everything. The baseline says it has existed for three. Another declares a century-old institution finished by next summer. The baseline asks why this summer.

Neither answer deserves automatic victory. One has a seductive story; the other has the inconvenient fact of survival. Put both on the table before the clever people begin decorating their hunch.

Behavioural principles

Behavioural ideas at play in this post

Short, plain-English explanations of the principles behind this post, with links to related books and examples in the archive.