The smallest durable construction

A small construction can be too small. It saves a component name, a file, a check today, and tomorrow creates an unobtrusive dependency in exactly that place. Nobody wants to touch it anymore. Substructure is quieter than superstructure. It does not look like architecture.

The opposite movement is more familiar: a manageable problem becomes a system with layers that have to be explained before the original problem can even be seen. Then the construction no longer only solves the task. It mainly manages itself.

Between those two mistakes sits the more interesting question: which construction is small enough to be understood, and large enough not to break at the first change?

In Sextile, this shows up in the derivation. A single calculation could be cast directly into code. That would be small. But once it should remain visible which factors were applied and what another decision would have changed, the program needs a form for that path. Not necessarily a large architecture. But some durable place where the reasoning can stand.

That is probably the more useful measure than size alone. A construction may grow when the problem has grown. It becomes suspect where it mainly serves the fantasy of what it might later become.