Notes from the margin.

When sources disagree, publish the disagreement

Two reputable outlets can report the same funding round differently. The temptation is to pick the more convenient figure. We publish neither.

The counter he switched off.

Reputation is not a source

An organisation writing about its own director is not independent evidence, however accurate it is. We hold cases back on that basis, including ones we would like to publish.

The counter he switched off.

One case is an observation, not a pattern

A reading drawn from a single case can be worth publishing. It is not worth calling a pattern, and the difference is not cosmetic.

The counter he switched off.Leave While It Is Still Working

An inferred constraint has to stay inferred

Some links in the archive are our reasoning rather than anyone’s account. They are marked as such permanently, and no amount of confidence promotes them.

She did not ask farmers to move. She moved to them.

The record shows what people did, not what stopped them

Reporting is consistently good on decisions and actions and consistently thin on problems and constraints. The gap is not random — it sits at the front of every chain.

No Uzbek founder had done it.The paper that became a patent.

An impressive result is not evidence of agency

A world title is thoroughly documented and tells us almost nothing about the thing this archive looks for. Outcomes are what agency sometimes produces, not what it consists of.

A case without a constraint is incomplete

We can publish a case that shows what someone did and still not know what they were up against. Those cases are weaker than they look, and we mark them rather than round them up.

Twenty-five years away, then a roadshow through nineteen cities.A shop where buying and learning are the same visit.Construction, Not Disruption