Main idea

Optimising a timetable — assigning rooms, resources and slots under multiple constraints — is a combinatorial problem. One of the most effective methods comes from elsewhere: in 1983, Kirkpatrick, Gelatt and Vecchi showed in Science that a metallurgical process, heating a metal and then cooling it slowly so its atoms find a stable configuration, transposes into an optimisation algorithm. That is simulated annealing. Between a foundry and a room schedule there is no obvious connection — and yet that transfer changed an entire field of research.

It is not the quantity of accumulated knowledge that produces this sort of result, it is having two sufficiently precise domains in mind and seeing that a mechanism from one applies to the other. The more distant the ideas brought together, the more the connection is worth: between two neighbouring ideas, the relationship was already known.

That is what Luhmann expected of his slip box, which he described as a "communication partner". Not a stock, but a device built so that precise and distant units of thought would cross paths — a note on a physical phenomenon and a note on a scheduling problem, a field observation and a hypothesis read six months earlier.

Why it matters

This changes what you ask of a knowledge system. If the point were to accumulate, completeness would be the right criterion and search the right tool. If the point is to connect, what counts is the precision of the units and the chance that they meet — two properties a complete base does not have by construction.

It also says why narrow expertise plateaus. Going deeper into a single domain raises the quality of the answers inside that domain, not the number of possible connections.

Nuances and limits

A distant connection cannot be commanded and cannot be measured: you know neither how many are missing, nor whether the one you made was the right one. It is a design that makes things possible, not one that guarantees them.

And reasoning by analogy misleads as often as it illuminates. Simulated annealing worked; countless seductive transpositions produced nothing but metaphors.

Open questions

• What is it, in a connection between two domains, that distinguishes an operative transfer from a mere analogy of vocabulary?