A second attempt is not always a repetition of the first. Imagine a person adjusting the temperature of a room: they change a setting, wait, notice the result, and change it again. What they notice becomes part of what they do next. The result has entered the procedure.

Companion demonstration draft · systems and feedback.

A choice that remembers

The counting exercise in The Shape of Possibility deliberately gives each position an independent choice. A feedback process asks a different question. If the next choice depends on the previous result, we cannot carry the independence assumption across unchanged.

Suppose the maker of the tiles decides to alternate marks whenever two identical marks appear together. The first few choices now affect the later ones. Some arrangements become less common, others more common, and the rule itself must enter the explanation. Counting possible arrangements is still useful; treating them as equally likely is no longer justified merely by the count.

Wait before turning the dial again

Timing introduces another distinction. If the room has not yet responded to an adjustment, a second adjustment may be a reaction to an old condition. The person is acting on a report of the past while hoping to influence the present.

This imaginary room illustrates why a model needs more than a list of its parts. It needs a statement of what information travels between them, and when. The Limits of a Useful Model asks how to draw that boundary without concealing what lies outside it.