Interfaces for foresight

A forecast nobody acts on is trivia. The gap between a model's output and an organization's behaviour is an interface problem, not a modelling problem.

Probability is not a UI

"73% likely" tells a team nothing about what to do on Monday. Useful surfaces translate probability into a decision boundary: at what threshold does this change our roadmap, our hiring, or our pricing?

Design rules we keep returning to

Rule	Why it matters
Evidence one click away	Trust collapses without provenance
Show the change, not the state	Deltas drive action; totals drive comfort
Name the decision	Every view should answer "so what?"

Restraint as a feature

The temptation is to show everything the pipeline produces. The discipline is to show the three things that would change a mind, and to make the rest reachable but quiet.