The Flatland Fallacy in Abstraction

Every abstraction drops dimensions on purpose. That is its usefulness: a map that kept everything would be the territory and would help no one. So the act of flattening is not the error. The error is the second step, the silent one — reasoning onward as though the flat result still contained what the cut removed.

Call it the Flatland fallacy. In Abbott's plane, a sphere passing through is not misperceived so much as under-determined: the circle really is all the data the plane can hold. The Flatlander goes wrong only when he concludes that circles are what exist. The move from this is what I can see to this is what there is is the whole fallacy, and it is available to anyone in any number of dimensions.

Why relations go first

Flattening is not uniformly destructive. It tends to preserve quantities and destroy relations. Cut across a braid and you keep the strand count and lose every crossing — over, under, order, accumulated twist. Average a wave field and you keep the amplitude and lose the phase. Rank a class of students and you keep an ordering and lose why any of them thought what they thought.

Since structure usually lives in the relations rather than the counts, the first thing a projection removes is the thing that made the object worth describing. That is why the fallacy is so productive of confident nonsense: the numbers survive, so the reasoning feels supported.

Three failures worth naming

Reification of the slice. The dashboard becomes the organization; the metric becomes the goal; the diagram becomes the field. The proxy is promoted to the object.

Recovery assumed rather than shown. Sometimes several projections do reconstruct the whole — that is tomography, and it is a triumph. But recovery has to be demonstrated, not presumed because more than one flat view exists.

Analogy inflated into result. Contextuality theorems say something stronger than "we lost detail": for some systems no single flat assignment reproduces all measurable relations at once. That is a proof about what cannot exist, not a metaphor about limited vision. Borrowing its authority for the loose case is its own version of the fallacy.

The repair

The remedy is procedural and unglamorous. With every flat representation, state three things: what it keeps, what it discards, and whether the discarded part is recoverable from other views or genuinely unavailable. An abstraction that carries its own loss statement stays a tool. One that does not becomes a plane its users mistake for the world.

This is the same discipline Volume 27 applies to borrowed geometry. A metric tensor used as a fluid analogy, a braid used as a picture of a field, a graph used as a picture of correlation — each is a projection with a validity band. Naming the band is not modesty. It is what keeps the picture usable.