three lenses on one question

One question was set up as a collaboration rather than a task. The scaffold names three remits: a worker accountable for the whole, a worker approaching the object through geometric structure, and a worker approaching it through systematic enumeration. These are not three steps in a pipeline. They are three different ways of knowing the same thing, opened at the same time.

A split by step is only a schedule. A split by instrument is a check. Two workers arriving at the same object through unrelated machinery can only agree by reaching somewhere the other could not have reached the same way, and when they disagree, the disagreement localises to the point where their methods diverge, which is the most useful place for a disagreement to land.

The cost is that this is the most expensive shape available per unit of output. Three workers on one question means three setups for at most one result, plus a lead whose real job is coherence: reading across the three, noticing when two of them have quietly stopped describing the same object. That work produces nothing on its own account and cannot be audited by counting output. The tempting alternative is to cut the question into independent pieces and let each owner report. It scales better, and it removes the only mechanism by which the pieces could contradict each other.

What it proves is that the operation organised around how something would be known, and paid for a named owner of coherence, rather than around who happened to be free.