What the other 95 percent of the universe is, and the dates I find out I'm wrong.
Hogey's Corner had a Gauntlet arcade game, and my dad used to hand me a roll of quarters and send me down the street. Summer of 1986. I was eight. Ten dollars in quarters is forty games, and I have never felt richer since.
Those arcade games stole your quarters. Sometimes you dropped one in and got nothing. No credit, no chime, no wizard on the screen. The quarter was not gone. It was sitting in the coin box, and at close the owner counted it with all the others, and it made the box heavier, and it paid the rent the same as any other quarter. It just never put anything on the screen.
That is dark matter. The rest of this essay says it more carefully.
Take everything you have ever seen or heard of. Stars, gas, planets, dust, comets, neutrinos, your kitchen, you. Every photograph astronomy has ever taken. That is about 5 percent of what is in the universe. About 26.5 percent is dark matter, which has weight and makes no light. About 68.5 percent is dark energy, which has no weight in the ordinary sense and pushes everything apart. Those are the Planck 2018 numbers: the settled measurement, not mine, and every serious person agrees on them.
"Dark" is a bad word for both. It sounds like a shadow, or something black. What it actually means is that the stuff does not touch light at all. It is not dim. It is not hiding. Light goes straight through it and comes out the other side with nothing to say about it. All we get is the weight, and the push.
In this framework space is not smooth. It is built out of voxels, which is just a word for little cubes. Get an ice cube out of the freezer and run your finger along the edges and count them. Twelve. Recognition, which is the single event this whole theory is built out of (one thing comparing itself against another and paying for the comparison), flows along those twelve edges, and the interference of those twelve channels comes out with twenty-four phases.
Most recognition events settle. They close, the books balance, and what you get is matter and light and everything you can point at. A small fixed fraction cannot settle. Not because the universe is sloppy or in a hurry. The geometry does not leave them anywhere to land. The event happened, the cost is real, the ledger carries it, and it never closes. That residue still has weight, because weight is what an entry on the books is. It never gets to make light, because it never closes.
It is the quarter in the coin box. Heavier box, dark screen.
Here is what that costs, in the framework's own arithmetic MODEL:
Ω_dm = sin(π/12) + φ^−4/24 = 0.2648981 Planck 2018: 0.2645 ± 0.005In pieces, so you can check it on your phone: sin(π/12) is 0.2588190, which is the twelve-fold part, and φ is the golden ratio, 1.618 and change, so φ^−4/24 is 0.0060791, a small correction divided by twenty four. Add them and you get 0.2648981. The measurement says 0.2645. The gap is 0.0003981, and Planck's own resolution is 0.005, so the gap is about one twelfth of the error bar we have to see it with.
Now the part that matters more to me than the agreement. That correction term, φ^−4/24, is machine-proved unique under a stated contract THEOREM (conditional). The Lean kernel checked it, not a person squinting at it, which means: if you accept one specific written-down list of requirements, exactly one term can sit in that slot. Lean verifies that the conclusion follows from the contract; it does not verify that nature signed the contract. What it does not mean is that no other form exists. Rival forms do exist. They obey looser contracts, they give different numbers, and no measurement we currently have can tell them apart. So the selection of this one is structural, not empirical. I would rather say that out loud than let four matching digits do work they have not earned.
Honest edge. The number rests on two named premises. One, that nature actually works this way on this channel. Two, that the cost of a path that never closes is its untransmitted closure weight. Neither is proved. Both are written down where you can find them and swing at them. And the kill rule is a number: any measurement more than 0.021 from 0.264898 and this is finished. I do not get to move that line later.
The first version of that formula we published was wrong. A constant got slotted into the wrong place, and the wrong number went out with exactly as much confidence as the right one. Our own audit caught it and we corrected it the same day, and the correction log is public, timestamps and all. I would rather you read that log than this essay.
In my twenties I told God I would not attribute a single characteristic to him that I could not verify, because a characteristic you make up is an idol, even a flattering one, maybe especially a flattering one. That habit is the reason I have anything at all. It is the same rule here. 0.2648981 is not a characteristic I get to make up, and neither was the one we got wrong.
Back to Gauntlet for a second. The thing everybody who played it remembers is that your health goes down the entire time. You are not being hit. You are not doing anything wrong. The bar drops because the game is running. You feed the game quarters to keep existing on the screen.
Dark energy is that. It is not a substance sitting in space. It is the overhead of keeping the ledger running at all, the cost of the books being open, and that is why it behaves so strangely compared to everything else we know. Ordinary matter falls together. It clumps into dust, then rocks, then stars, then galaxies, then those long bright filaments you see in the survey maps. Dark energy never clumps, not even a little, not anywhere. It sits at the same density in your kitchen as in the emptiest gap between two galaxies, because it is not a thing that can pile up, it is a rate you are being charged. And a charge like that pushes. The expansion of the universe is what the charge looks like from the inside.
Ω_Λ = 11/16 − α/π ≈ 0.685 Planck 2018: 0.6889 ± 0.0056MODEL Eleven sixteenths is 0.6875, and α is the fine structure constant, the number that sets how hard light and charge grip each other, and α/π is about 0.00232. Take one from the other and you land about 0.0037 from the measurement, inside the error bar.
Standard cosmology treats dark energy as a cosmological constant. Same number yesterday, same number in a hundred billion years, frozen, forever. That is the textbook assumption and it is a completely reasonable one.
This framework says it is not frozen. It thaws, slowly, as the universe's books age. Cosmologists write the thawing as a pair of numbers, w0 and wa. A frozen constant is w0 = −1 and wa = 0, dead flat. This framework says (w0, wa) = (−0.983, −0.017). About 1.7 percent off frozen today, and drifting.
That is a small difference, and it is a difference the sky can see. DESI is measuring it right now. Euclid is measuring it right now. One to three years. If they land on frozen at five sigma, the vacuum sector of this framework is dead, and I will write that essay too, on the same site, with my name on it. This one is a manuscript derivation, registered as H-W in our public register, which means it has a date and an author and a number attached to it, so nobody has to take my word later about what I claimed earlier.
| Claim | Number | What kills it | Who decides, and when |
|---|---|---|---|
| Dark matter is unsettled recognition residue, not a particle | 0.2648981 vs 0.2645 ± 0.005 | any measurement landing more than 0.021 from 0.264898 | anyone who measures the matter budget; open ended, and the tolerance is wide because the honest room is wide |
| Dark energy is ledger overhead, and it thaws | 11/16 − α/π ≈ 0.685; (w0, wa) = (−0.983, −0.017) | a freezing signature at 5 sigma (w0 = −1, wa = 0) | DESI and Euclid, one to three years |
About 5 percent settled accounts: us, the stars, the light, everything anybody has ever photographed. About 26.5 percent accounts that opened and never closed, still carrying weight, never carrying light. About 68.5 percent the cost of keeping the books open at all, spread perfectly even, slowly leaking as the books get old.
We are not selling anything. I am not asking you to believe either of these numbers, and our job is not to convince anyone of anything, which is lucky for us, because I would be bad at that job. The job is to write the number down, write down what kills it, print the time we got it wrong, and then stand still while the sky decides. Both of these can be wrong. One gets decided by DESI and Euclid inside three years. The other gets decided by anybody who measures the matter budget past our tolerance. If we are wrong, you will find that out from the same files we used to say we were right, and that is the only part of this I am certain of.