You Cannot Flatten a Sphere

Take a sheet of paper and try to wrap it smoothly around an orange. It creases. It always creases, and no amount of care prevents it.

That is not clumsiness. It is forbidden, and Carl Friedrich Gauss proved it.

His result was so pleasing he called it the Theorema Egregium — the remarkable theorem. It says that a surface's curvature is intrinsic: it belongs to the surface itself, not to how the surface happens to sit in space, and no bending that avoids stretching can change it.

A flat sheet has curvature zero. Roll it into a tube — still zero, which is why that works. Roll it into a cone — still zero, which is why that works too. A sphere's curvature is not zero, and no bending will get you there. To wrap the orange you must stretch, tear, or crumple.

You have used this theorem today without noticing. Fold a slice of pizza lengthways and the tip stops drooping. By giving it curvature in one direction you have forced it — by Gauss's theorem — to stay straight in the other. The fold is not stiffening the crust. It is a proof, held in one hand.

Everything in this zine follows from that one fact, so it is worth being clear about its status: this is not a rule of thumb, an approximation, or a limit of current technique. It is a theorem. It was true before anyone made a map and it will be true after the last one is thrown away.

The Earth is roughly a sphere. A map is flat. By the theorem, no map can preserve everything — not as a practical difficulty, but as a matter of proof.

So every projection ever drawn has had to choose. The properties on offer are:

You may have some. You may never have all. Preserve area and you must warp shape; preserve shape and you must warp area. The theorem does not care how good your mathematics is.

This is worth stating carefully because the ordinary complaint about maps is that they are inaccurate, as though a better cartographer might fix it. That is not the situation.

Which means every map you have ever read was the answer to a question — and the map does not tell you what the question was.

Gerardus Mercator published his world map in 1569 with a specific job in mind: getting a ship where it was going.

His projection preserves angle. That has one glorious consequence for a navigator — a course of constant compass bearing is a straight line on the paper. You lay a ruler between here and there, read the angle, hold that heading, and you arrive. For four centuries of sail, this was worth almost any price.

The price was area, and it grows without limit toward the poles. On a Mercator, Greenland looks about the size of Africa. Africa is roughly fourteen times larger.

Now the part that matters, and it is not that Mercator was wrong. Mercator was right, for sailors. The trouble is what happened next.

A projection is a decision about what matters. Once it is printed, the decision looks like the world. Generations grew up with a picture in which the tropics are small and the high latitudes enormous, and no one told them a trade had been made on their behalf, or by whom, or for what.

The distortion is not the scandal. The silence is.

In 1973 Arno Peters presented a world map that preserved area, and argued that the familiar one had been quietly shrinking most of the world's people for four hundred years.

On his map every country occupies its true share of the page. Africa is vast, as it is. The high latitudes shrink back to size. Shapes, in exchange, are stretched — countries near the equator pull long and thin.

It became a public fight, and it is worth telling honestly, because the tidy version of this story is wrong in both directions.

And here is the part that survives the mess. Both sides were arguing about which distortion to accept, because the theorem left them nothing else to argue about. There was never an option on the table that distorted nothing.

Which is the useful lesson, and it generalises far past maps: when reduction is forced, the honest move is not a better reduction. It is to say what you gave up.

Nicolas Auguste Tissot could not remove the distortion. Nobody can. So in 1859 he did the next thing, which turns out to be better.

Imagine a tiny circle drawn on the globe. Project it onto the flat map and it will not come out a circle — it comes out an ellipse, squashed and turned by exactly however much that projection mangles that spot.

So: draw that ellipse. Draw one at every intersection of the grid. Now the map carries its own error bars on its face — you can see at a glance where shapes are true, where areas swell, and by how much.

It is called Tissot's indicatrix, and it is one of the most quietly radical instruments in the history of representation, because of what it concedes and what it refuses in the same gesture.

Notice this is not humility about whether the map is useful. A Mercator with indicatrices on it is still a Mercator, still gets your ship home, still hangs on the wall. What has changed is that the reader can now see the price.

Hold that shape — the summary, and the record of what the summary cost — because the rest of this zine is about a domain where we do the first half and have almost entirely refused the second.

A person is not a sphere. But a person is many-dimensional, and that is the property the theorem is actually about.

Here is a partial list of axes along which one human being varies, none of which reduces to any other:

And every one of those bars moves — by the day, by the room, by how much sleep there was, by whether the lights are humming. This is not a fixed shape. It is a shape in weather.

No. 21 already showed why this is not a line, and made the case with spectra: a spectrum is a fingerprint, not a ladder. That argument is not repeated here.

This zine asks the next question. Granted it is many-dimensional — what happens, exactly and provably, when somebody flattens it to one number anyway?

An IQ score. A support level. A functioning label. Each takes that many-axis shape and returns one value.

That operation has a name, and it is not a metaphor: it is a projection. Exactly like flattening a globe onto paper, it maps something of higher dimension onto something of lower dimension — and exactly like flattening a globe, it cannot be done without loss. Not "in practice." At all.

So the theorem tells us three things about every such number, before we know anything about the person:

Look at the shape from the last spread and ask what one number could possibly do with it. Take the mean? Then a person with two ferocious peaks and five troughs scores the same as somebody flat across the middle, and those are not the same life. Take the minimum? The maximum? Each answers a different question, and each throws away a different six-sevenths of the picture.

And here is what makes it worse than a map. A Mercator at least keeps every place on it. Greenland is distorted but it is there. A single-number summary of a person does not distort the other dimensions — it deletes them. There is nowhere on the score for the sensory axis to be wrong. It simply is not represented.

Mercator kept angles because his readers were crossing oceans. Every projection is optimised for somebody's journey, and you can read the somebody off the choice.

So put the question to the summaries that get applied to us. What was this optimised for — and therefore, who is it for?

None of these is villainous. A budget really does have to be allocated; a service really does have to decide. Projections are not crimes. They are tools with users.

But notice what is common to the bottom three rows: the person being summarised is not the user. The dimensions kept are the ones the institution needs to navigate by. The dimensions deleted are the ones only the person needs.

And when your own map is drawn for somebody else's journey, the sensation is very specific and very familiar: the thing you most needed represented is the thing that isn't on it.

The theorem does not say never summarise. Cartographers did not respond to Gauss by refusing to make maps. That would have been useless, and nobody would have got anywhere.

They responded by making the distortion visible. That is the whole move, and it transfers.

So: if a number must be produced — and sometimes one must, because budgets and services and doors are real — then produce the distortion alongside it.

None of this is exotic. Every one of those is ordinary practice somewhere already: error bars, confidence intervals, a methods section, a "nothing about us without us" review. We know how to do this. We simply do not do it here.

And the standard objection — that's more complicated than a number — is exactly the objection Tissot faced, and it is answered the same way. Yes. It is more complicated. The simplicity of the bare number was never real; it was purchased by hiding the complication, and the person carrying that complication is still carrying it whether or not it is printed.

A number with its indicatrix is a usable map. A number without one is a Mercator with the legend torn off, handed to somebody who was told it is the world.

Here is the last thing the theorem gives us, and it is the gentlest.

On a sphere there is no edge and no corner. There is no margin. Every point has exactly as much sphere around it as every other; no place is peripheral, because peripheral is a property of flat things with borders.

The margins arrive with the projection. The edge of the map is not a feature of the world — it is where the cartographer decided to cut, and every world map has one, and somebody always ends up split down it or shoved to the rim.

So when a summary puts you at the far end of a scale — the low end, the difficult end, the end that gets the shorter appointment — that position is a fact about the projection. It is not a fact about the sphere.

And the sphere is the real object. The map is the thing we made because we needed something that would fold into a pocket.