The Shadow Is Bigger Than the Thing

Start with the thing everyone gets wrong, including most of the headlines in April 2019.

A camera works by catching light that left the subject. That is the whole mechanism — the apple sends photons at your eye, and you see an apple.

A black hole sends nothing. Not a little. Not faintly. The event horizon is the surface where the geometry has tipped so far that every path leading outward has stopped existing; there is no direction that is away. Light does not struggle out slowly. It does not get out.

So the ordinary way of knowing about a thing is simply unavailable, and no better camera will fix it. This is not a limit of our instruments. It is a fact about what the object is.

Which sounds like the end of the enquiry, and is instead the beginning of it. Because a thing that emits nothing still does a great deal — to gas, to stars, to the paths of passing light — and all of that is visible.

The rest of this zine is about what you can learn from the second-best option, and about the one number that makes it dangerous.

Gas falling toward a black hole does not go quietly. It compresses, shears against itself, and gets hot enough to shine — so the object that emits nothing is generally surrounded by something that emits a great deal.

Now put that glow behind a lens made of gravity. Light travelling near a massive body does not go straight; its path bends, and the closer it passes the harder it bends. At a particular radius the bending becomes exactly enough to curve a beam right around into a circle — the photon sphere, at three GM/c² from the centre. Light there is orbiting.

So what arrives at a telescope on Earth sorts itself into two kinds. Light that passed far enough out escaped and reached you. Light that came in closer than the critical approach spiralled in and never left.

The result is a bright ring around a dark middle — and the dark middle is not an object. It is the set of directions in which any light you might have received went into the hole instead.

The astronomers' word for that dark middle is the shadow, and the word is well chosen — a shadow is defined entirely by what it happens to, never by itself.

Here is the fact this zine is built on, and it is a piece of exact geometry rather than an impression.

For a non-spinning black hole, all the relevant radii can be written in one unit — GM/c², which is just the object's mass expressed as a length:

Divide the last by the first. 3√3 / 2 ≈ 2.598.

The dark circle in the picture is about two and a half times wider than the object casting it. Gravity does not merely hide the black hole; it magnifies the hiding, throwing the darkness outward well past the horizon's true edge.

One more thing makes this usable, and it is close to a fluke. Real black holes spin, and spin drags spacetime around with it — so you would expect the shadow's size to depend on how fast it turns and how it is tilted toward you.

It barely does. Across the whole range of spins and viewing angles, the shadow's angular radius stays at 5 ± 0.2 GM/Dc² — a spread of about four percent, with the non-spinning case (3√3 ≈ 5.196) sitting near the top of it.

That near-constancy is a quirk of the equations rather than a principle, and it is what turns a dark smudge into a measurement: if the shadow's size is fixed by mass over distance and almost nothing else, then measuring it tests the geometry itself.

The first one they got was not the nearest. It was the one whose shadow happened to be biggest in our sky.

M87* sits at the heart of a giant elliptical galaxy about 55 million light years away, and it is enormous — the ring measurement implies a mass of 6.5 ± 0.7 billion suns. Distance shrinks it; size restores it. The two nearly cancel, and what is left is a ring 42 ± 3 microarcseconds across.

A microarcsecond is a millionth of a second of arc, and a second of arc is already a 3,600th of a degree. Put it in terms of a thing:

Nothing on Earth resolves that. A telescope's sharpness is set by the wavelength it works at divided by the width of the dish, and to see 42 µas at the 1.3 mm the hot gas shines at, the dish would have to be about as wide as the planet.

So they used the planet. That is spread seven.

The image released on 10 April 2019 came from data taken two years earlier, over six nights in April 2017 — the gap is not delay, it is the checking. And what the checking looked like is the next spread, because it is the most quietly admirable thing in the whole story.

Here is the awkward part of the method, and the reason the result is trustworthy.

An array of separate telescopes does not record a picture. It records how much the incoming waves agree between each pair of dishes — a scatter of measurements across a plane, with enormous gaps where no pair happens to sit. Getting from that to an image means filling in what was never measured, and there are infinitely many images consistent with the data.

So the choice of method is doing real work. And every person involved knew exactly what general relativity predicted they should find.

That is the classic way to fool yourself, and they took it seriously. The collaboration split into four imaging teams, each kept blind to the other three, working with two quite different families of algorithm — the long-established CLEAN and the newer regularized maximum likelihood.

The blinding is not bureaucracy. It is the only available answer to a question nobody can answer about themselves: did I find this, or did I draw it?

This is the same instinct as the five-sigma rule in No. 39 and the whole of Too Good to Check. Wanting a result is not a disqualification. Failing to build a guard against wanting it is.

You cannot build a dish the width of the Earth. You can, it turns out, use the Earth.

Eight radio telescopes at six sites — Arizona, Mexico, Hawai'i, Chile's Atacama, the Sierra Nevada in Spain, and the South Pole — pointed at the same patch of sky on the same nights and recorded what arrived, each stamped against an atomic clock.

Nothing was combined at the time. There is no network that could carry it: the array wrote roughly 350 terabytes per telescope per day, about 3.5 petabytes in all, onto racks of helium-filled hard drives. The drives were then flown to two correlator supercomputers and played back against each other, and only there did the separate recordings become one instrument.

Consider what this instrument is. Its aperture is a planet. Its resolution comes not from any one dish being good but from the dishes being far apart, and knowing precisely how far. No site saw the ring. No site could have. The image exists only in the relationship between them.

Three years after M87*, the same array released the one at home: Sagittarius A*, about four million solar masses, roughly 27,000 light years away, at the centre of our own galaxy.

Its ring came out at 51.8 ± 2.3 microarcseconds — slightly larger in our sky than M87*'s, because although it is around 1,600 times less massive it is also vastly closer.

It should have been the easy one. It was much the harder one, for a reason that is worth sitting with.

Everything near a black hole orbits on a clock set by its mass, and a small black hole runs fast. The innermost stable orbit around M87* takes something like thirty days to go round. The same orbit around Sagittarius A* takes about thirty minutes.

An interferometer builds its image out of the Earth's own rotation, sweeping the pairs of telescopes across the sky over hours. That method assumes the subject is the same subject from the first hour to the last. For Sagittarius A* it plainly wasn't, and the methods had to be rebuilt to cope with a source that changes while you are still looking.

Being nearer did not make it easier to see. It made it faster, and fast is its own kind of invisible.

Everything anyone knows about a black hole is a fact about something else.

The heat of the gas. The bending of the light. The orbits of stars whipping around the galactic centre. The width of the dark patch. Not one of those is a measurement of the object; every one is a measurement of what the object does to its neighbourhood — and together they pin it down to within a few percent.

This is not an exotic epistemic situation. It is an extremely ordinary one, and most of us live inside it.

And the response this usually gets is the one the EHT refused: if it can't be seen, treat it as absent. Every unphotographable thing in a person's life gets met with that sentence in some form — you don't look Autistic, you seem fine, you managed last week.

The astronomers had a far stronger version of the problem — a thing that emits literally nothing, not merely something inconvenient to observe — and they did not conclude it was absent. They built an instrument out of a planet and measured what bends around it.

Most versions of this argument stop at the previous spread, on the warm note. This one doesn't, because the physics has a second half and the second half is the useful one.

Go back to the number. The dark circle is not the same size as the object. It is about 2.6 times wider. Measure the black hole by its shadow and forget the factor, and you will conclude it is two and a half times the thing it is.

Now carry that across, carefully.

What an observer sees of an unmet need is almost never the need. It is the disruption — the shutdown, the meltdown, the missed deadline, the sudden absence, the job that ended badly, the reply that came out wrong. Those are real, and they are genuinely caused by the thing. They are also systematically bigger than it.

They are bigger because they include everything that had to bend. Years of compensating, the effort of masking, the accumulated cost of a hundred small accommodations made privately and never counted. All of that is bent light. None of it is the horizon.

Two things follow, and they cut in opposite directions, which is how you know it isn't a slogan.

For anyone doing the observing: do not size the need by the disruption. You are looking at magnified darkness. The person is not as large a problem as the wreckage suggests, and treating the wreckage as the measurement is how support plans get built for a person who does not exist.

And for anyone being observed: the disruption is not the measure of you either. It was inflated on the way out — by the bending, by the years, by an environment that made a small thing throw a long shadow.

In physics the magnification is exactly 3√3/2 and you can divide it out. In a life there is no such factor, which is precisely why you ask the person rather than measuring their shadow.

There is a comfortable misreading of everything above, and it needs closing off on the page rather than in a footnote.

The misreading is: since it can't be photographed, no claim about it can be checked, so any claim will do. That is not what happened here, and the EHT is its refutation.

They could not see the thing. They measured it anyway — Sagittarius A*'s ring at 51.8 ± 2.3 microarcseconds, matching what general relativity predicts for an object of about four million solar masses at that distance, to within a few percent. Not being able to see something is not the same as not being able to know anything about it. It only means the evidence arrives sideways.

So the claim this zine actually makes is narrow, and it is worth saying in one sentence: an unphotographable thing can still be measured, argued about, constrained, and got wrong.

Which is also the honest version of the human half. Self-report is not magic and it is not beyond question; it is the instrument available, and like every instrument it is checked against consequences, against other people's accounts, against what happens next. Asking is not a lowering of standards. It is the measurement.