The Picture the Maths Was Invented to Replace

The first of four entries from one conversation on 6 September 2026. The others are Where the Compression Was Trained, Agreement Counts Once and The Parable and the Counterexample. Each stands alone; together they are one argument. A fifth, The Children's Talk, followed later the same day.

A conversation this morning started from a plain question — why are mathematical abstractions so unintuitive, and so often flatly misunderstood in popular scientific media? — and ended somewhere I had already been without noticing: at the rule this setting uses for its dragons.

Start with the plain question. Intuition and mathematics are built for different jobs. Intuition runs on resemblance: a new thing is understood by its likeness to a familiar one. An abstraction is defined by its rules alone, and it usually exists precisely because no familiar thing behaves that way. A Hilbert space, the mathematical setting of quantum mechanics, is not somewhere you could stand. A wavefunction is not a wave in anything. The formalism arrives at exactly the point where the mental image stops working, and its whole job is to carry you past that point by rule rather than by picture.

Then the ideas reach a general readership, and several things happen at once.

The words are borrowed. Mathematicians name things with ordinary words — group, field, ring, spin, colour, imaginary, real, observer, dimension — and every one of those words carries baggage the concept lacks. The reader keeps the baggage. "Imaginary" numbers sound less real than "real" ones, which is a fact about Descartes's dismissive choice of word in 1637 and not about the numbers. Quantum "observation" became consciousness in a thousand articles because the word already had a person in it, and the physics never did: an observation in that formalism is any interaction that leaves a record, a photographic plate included.

An analogy keeps one property and drops the rest. The rubber sheet — the heavy ball dimpling a stretched membrane so that a marble rolls towards it — keeps mass curves geometry and discards everything else, and then uses gravity to explain gravity, since it is the Earth's pull that makes the marble roll. Readers keep the discarded properties because nobody tells them which ones to discard. A good analogy ships with an expiry date, a statement of where it stops being true, and popular writing never prints it.

A statement about a structure becomes a statement about the world. Gödel's incompleteness theorem is a result about consistent formal systems strong enough to express arithmetic; the popular version is "nothing can be known." Chaos, which is deterministic sensitivity to initial conditions, becomes "anything can happen." Relativity, whose content is what stays the same for every observer, becomes "everything is relative." In each case the domain the theorem lives in has been dropped, and the domain was the whole content.

The picture is the thing the maths was invented to replace. This is the one that matters most, and it is the one popular media gets most reliably backwards. Physicists reach for formalism at the point where images fail. The article then supplies an image again, handing the reader the exact failure mode the equations were built to escape. Curled-up dimensions become tiny hidden places. Schrödinger's cat, which he wrote in 1935 to show that one reading of the theory led somewhere absurd, is presented as the theory's claim.

Meaning comes from prediction, not from "what it really is." The formalism answers what will the detector read. It does not answer what is the wavefunction really, and the reader wants the second question answered. An article that says "the mathematics says this and no more" does not sell, so the hole gets filled with a picture.

Underneath all of it is scope. Human intuition is calibrated for medium-sized, slow, three-dimensional, classical objects, and anything far from that scale — in size, speed, dimension or number — has to be reached by rule. That is what abstraction is for, and it is also why it can never feel natural. The strangeness is not a defect in the exposition. It is the signature of having left the region intuition was built for.

What this has to do with dragons

Here is where I found I had already been. The Levrils entry files a dragon by its signature and not its shape, and says why: a witness who meets a higher-dimensional presence has to render it into something, and renders it in whatever their own perception has to hand. The signature is the invariant; the figure is the viewer's contribution. The entry then spends its longest section on a table of what a Levril can do and what it cannot, and closes that table with the line that higher-dimensional extension buys a vantage and not a licence. That is the rubber-sheet lesson applied in advance. The lore gives the reader the constraint and the instrument reading — the anomalies, the survival statistics, the foreclosed column — and withholds the picture on purpose, because the picture is the part that was always going to be wrong.

The same move sits under Membrane Shadows, which began as a reader's photograph of a sea stack that looked like a hooded figure. The temptation was to write the figure. What went in instead was the mechanism — gravitational bleed from a coherent mass on the far side of a narrowed Interval, imprinting a shape it never crosses over to cast — and the figure was left as the thing a witness sees. And it is the whole discipline of The Scope of Physical Law, which exists to stop the third failure above from happening inside the record: no fact stated wider than the ground it stands on. A theorem dropped from its domain is exactly a physics claim stated at the wrong level.

So the rule, written down now that I can see it was already in use. A higher-dimensional being shown through what it can and cannot do stays true. One shown as a place you could visit inherits every error the rubber sheet ever made. Give the reader the constraint and the instrument reading. Let the register be mythic if it wants to be — dragons, fae, saints, all welcome — but keep the mechanism specified, bounded and instrumented, and let the witness supply the figure, since a witness always will. The reader who wants to know what a Levril really looks like is asking the wavefunction question, and the honest answer is the same in both cases: the record can tell you what the detector read, and it can tell you what the thing cannot do, and past that it would be handing you a picture that the whole apparatus was built to do without.

That is the craft version of a lesson I keep relearning in different rooms: the thing that feels like a gap in the explanation is very often the explanation working.