The fifth day after the memo went out: thirty-one replies. Twenty-six acknowledged receipt, three asked about the sharing table, two asked when the next revision would come. Nobody asked about the data.
Julian date 2459517, Sunday, eleven-twenty at night.
Hu Yisu came into the machine room at a run. No USB drive in his hand—his phone, screen facing out, a photograph on it, badly blurred, of a projection screen.
Shijie. He said. I remember.
Junior year, an elective the medical school offered. Introduction to Neuroscience, two credits, no exam—one set of notes to hand in and that was all. He had gone four times. In the fourth session they put up a figure; he photographed it with his phone, and never looked at the photo again.
The figure showed a cultured slice of rat cortex on a sixty-channel microelectrode array, recording spontaneous firing. Cut the firing into avalanches, count how many electrodes each avalanche carried, plot it log-log: a straight line.
The teacher read the slope aloud. Hu Yisu said: minus one point five.
He had handed those notes in; they sat in an old mailbox's drafts folder, and it took him two hours to find them.
She took the phone and enlarged the photo, enlarged it again. Blurry. The axes couldn't be read; only a slanted line and, upper left, a number: 1.5.
She said: minus one point five—you find it in forest fires, in earthquakes, in sandpile models, in solar flares. Self-organized criticality. You hand me a slope and it says nothing.
I know. He said.
She said: you went to that class four times.
Three of the four I was asleep. He said. The last one was because the notes were due. I sat in the front row, took it all in, and thought nothing of it afterward.
She handed the phone back.
That analogy of yours, I remember it. She said. Snoring. Which time did you say it.
The third time. He said. The tower climb. You said no changing the analogy until the thing was verified.
Now it's verified. She said. You may change it.
I'm not changing it. He said. I think it holds.
She turned the paper notebook to a fresh page and wrote four lines.
One: avalanche-size distribution, power law.
Two: long-range temporal correlation.
Three: low-frequency power spectrum.
Four: shape collapse.
The first three we have, and they have. She said. All three, and it still says nothing.
And the fourth.
The fourth is different. She said. The fourth isn't a distribution. It's a shape.
Julian date 2459518, Monday.
She explained the fourth one to Hu Yisu. It took about three minutes.
Events have size, and they also have a course. A small event lasting two hours and a large one lasting thirty: normalize the time by each one's own duration, normalize the amplitude by each one's own peak, then average all the events inside the same size bin—if they are a family born of one dynamics, those mean shapes will collapse onto a single curve. The small one walks this way, the large one walks this way, the walk is the same walk; only the distance walked differs. If they are merely unrelated things that bumped into one another, no collapse.
Why can't those collapse.
Because collapse doesn't ask for a distribution. It asks for a path. Things triggered by noise have no path.
Eight hundred sixty-seven events, split into five bins by duration, a mean shape for each bin, normalized, stacked.
Collapsed. The five curves stacked up; maximum deviation, four point three percent.
She ran two controls. First: keep the amplitude distribution, shuffle the events' time order and redeal a thousand times—collapse residual averaging thirty-one percent. Second: a synthetic sequence triggered Poisson, injected with the same amplitude distribution—residual twenty-eight percent.
Three plots side by side. Into the directory that looked like a cache they went as well.
Hu Yisu stood behind her looking: Shijie.
Mm.
Your hand is shaking.
She said: I know.
Hu Yisu went back and forth over the five curves three times. Shijie, he said, this collapse—do they do it too, over there?
I don't know. She said. I'll go look.
Monday night she did the same to the events of the 1,042 stars.
The optical sampling was too coarse—one point every thirty minutes; events could only be cut down to longer than an hour. She cut out eleven thousand three hundred events. The log-log slope of the size distribution came to minus one point four seven, error plus or minus zero point zero nine. Three percent off the radio side.
Shape collapse: three bins, maximum deviation seven point one percent. Coarse. It collapsed.
In the notebook she wrote the two numbers side by side: −1.52, −1.47.
When they were written she thought: two instruments, two wavebands, two software suites, two orders of magnitude apart in event count—and one and the same number. She did not write that sentence down.
In the afternoon she went at the literature. The university library's proxy let her in. She found the 2003 paper: cultured rat cortical slice, microelectrode array, avalanche-size distribution a power law, slope minus one point five. She found the human MEG and EEG next: the long-range temporal correlation exponent landing between zero point seven and zero point nine, the low-frequency power spectrum running about as one over f.
All three matched.
Then she went back and checked on her own claim that you find it everywhere. The sandpile model's avalanche exponent sits around minus one point one; forest fires come out flatter; solar flares have too few samples to plot four points. The minus-one-point-five family is the neuron family—largest sample size, most often repeated, repeated for twenty years.
She laid her figure beside the figure in the paper. Her x-axis ran in Julian dates; theirs, in milliseconds. Her nine pulsars, the nearest four hundred light-years out, the farthest eleven thousand; theirs, a slice of cortex zero point four millimeters thick. The two curves stacked: like two pulse readings taken off the same blood vessel, a layer of skin between them.
Three unread messages on the phone. The earliest, Friday at seven-odd in the evening: I've written it. Saturday: did you look. Sunday: never mind.
She opened Friday's photograph. Under the worksheet, one line in pencil: someone with open eyes has to look at one place. I don't know where that is.
She read it three times. Typed a line: that's a good line. Deleted it. Typed: what did the teacher say. Deleted it. In the end she sent back two words: saw it.
Back at the dorm that night she changed the water in the tank. Halfway through, Lao Wu came out from behind the driftwood, claws feeling the way ahead, touched the tweezers once, and backed in again. She went still, and waited until the water had settled before she turned off the light.
Julian date 2459519, Tuesday, ten at night Beijing time. Three in the afternoon in Vienna.
Four squares on the screen. Kramer was there. Hu Yisu was in the machine-room square, microphone on.
Ursula Kovacs was fifty-eight, hair cut short, gray, undyed. Behind her a white wall, and pinned to it an A4 printout of formulas, the characters small, three pushpins. In front of her a glass, with water in it. While she talked her hand turned a pencil, turned it fast.
No courtesies. Her first sentence was: one question first. Can you modulate it.
Zhou Yan said: no.
How do you know that.
I've tried. She said. Strip out the amplitudes, keep only the event times, and the mutual information between two stars drops to zero. The information is all in the amplitudes, and the amplitudes are drawn from one and the same power law. There is no code.
Ursula stopped the pencil for a moment: good.
She said: those two sentences you will weld shut. Correlation carries no information. It cannot be modulated. Whoever takes it to send signals is lying to himself, and lying to everyone else too.
Kramer said: does that go into the memo.
Ursula said: into the header of every page.
Then she said: you call it correlation. I call it organization.
Zhou Yan said: where is the difference.
The difference is you are counting the wrong units. Your nine stars are not nine sources; they are nine sampling points. You think you are watching nine things. You are watching nine positions of one thing.
And the one thousand forty-two.
One thousand forty-two sampling points. She said. The one thousand forty-two you counted are not one thousand forty-two participants; they are one thousand forty-two probes stuck into the same thing.
Zhou Yan said: stuck with what.
Ursula lifted her head: with you. With your hydrogen masers, your CCDs, your ephemerides. You think you are outside it, looking in. Your clocks are in the solar system, and the solar system is inside it. You are not outside.
She heard the first half of that sentence clearly; the second half scattered in her right ear. She switched the earphone to the left and said: again.
Ursula said it. This time it came through. Beside her, Hu Yisu wrote the sentence down.
Hu Yisu said suddenly: Shijie. That analogy of mine.
Zhou Yan said: say it.
Snoring. He said. The one snoring doesn't know he snores.
The screen was quiet a few seconds.
Ursula said: right. So don't expect it to answer you.
Hu Yisu lifted his head again: may I ask a layman's question.
No. Ursula said.
He asked anyway: the avalanches in my head—are they the same thing as this.
Ursula looked at the screen, looked two seconds. She said: not the same thing. Your brain is inside it. They are not the same size, not the same age, not the same thing. But your brain is one part of it measuring another part—and that sentence, and the pulsar sentence, are one sentence.
Hu Yisu said: then when I look at these numbers, what am I.
You are one part reporting on another part. She said. Don't suppose this grants you any standing.
Zhou Yan turned the paper notebook back. In the margin of the week-six page stood one word: organization. Circled heavily—the pen tip had pushed a ridge into the back of the page. She rubbed the ridge with a finger.
She said: is it waking.
Ursula said: the question is badly put. Awake or not is not a switch; it is a quantity—global phase synchrony. I call it S. Zero is each going its own way; one is all at once.
How much counts as awake.
Zero point five. Past zero point five, the local can no longer be shut in; one twitch carries everything with it, and no one gets to stay where they are.
How much now.
Ursula said: tonight I'll run a rough estimate on your numbers. I will not give you an error. Thirty years of data estimating a process on a million-year scale—one significant figure is all I can give you.
My method is the stupidest one there is: normalize the nine series, take the global mean at each moment, see how much of the total fluctuation the global fluctuation has eaten. Nine stars is too few; the estimator leans low.
Zhou Yan said: why did you take this call.
Kramer says you have a number you don't dare release. Ursula said.
It isn't that I don't dare.
Then you won't. She said. The two, I don't distinguish.
And that figure.
Tomorrow. She said.
Before the meeting broke up, Kramer asked Zhou Yan for the table of one thousand forty-two, saying he meant to do an internal preprint. She said: when I've finished the polarization.
Kramer said: what has polarization to do with this.
Ursula answered for her: if it is an organization, there are pathways. Stars do not join through vacuum.
Kramer said: then I won't do that preprint.
Ursula said: you will. You only want to wait for someone else to go first.
Kramer laughed once, and didn't deny it.
Zhou Yan said: I finish the polarization first.
Polarization won't give you a name. Ursula said. Polarization gives you pathways. If it's a name you want, go do something else.
I know.
You don't. Ursula said. Your kind all want two things at once.
After the meeting Old Tang's call chased her down, one matter only: the Miyun station office's deadline was five working days. Friday.
She said: I know.
I've already blocked for you once—told them the raw voltage records were being sorted.
Thank you.
I block once. He said. Zhou Yan—those two lines. Are you filling them in or not.
I am. She said. Forty-one terabytes; exporting means redoing the scrubbing. I'm doing it.
Old Tang said nothing at the other end. After a few seconds he said: that's half of a true sentence, isn't it.
More than half. She said.
One more thing. Old Tang said. The station office asked me why ours are the only two lines on page three left blank. I said the data was bulky. They said, then write us a schedule.
What did you say.
I said I'd ask.
She said: five days.
Old Tang said: I said three.
He hung up.
In the machine room Hu Yisu waited for her to hang up before he spoke: Kramer wants that table. Do I give it to him.
The public version. She said.
That isn't the version he means.
Then the six thousand three hundred eighteen version, not the one thousand forty-two.
Hu Yisu said: he'll ask again tomorrow.
We'll see tomorrow. She said.
She took the tape out of the drawer and set it on the desk, and looked at it a while. The crack in the corner of the plastic case was still there, the mended transparent tape gone yellow. She pried the case open; the reel lay wound neat, and near the rim the two-fingers-wide stretch where the magnetic coating had wrinkled, shining under the desk lamp.
She set a finger against the teeth of the reel and turned it half a circle. It turned. A little stiff.
Hu Yisu watched from the doorway: Shijie, still no machine.
I know. She said.
Until when.
Until there is one.
Julian date 2459520, Wednesday, eight-forty in the morning.
The mail held one attachment and one number.
The attachment was a one-page derivation. She understood the first line; from the second line on, she did not.
The body was one sentence: S ≈ 0.31.
Below it, a new line: It is already awake. The eyes have not opened.
Below that one more line, added as if in passing: go look at the interstellar polarization archive. Look at the orientation of the magnetic fibers. If it is isotropic, I am wrong.
Hu Yisu came in at eight-fifty and looked at the number: what is zero point three one.
She said: a third of all-at-once.
Is that much, or little.
I don't know. She said. I don't know what to set it against.
Against last year.
We don't have last year. She said. We don't even have yesterday.
She printed the attachment and pressed it under the first page of the paper notebook.
In the notebook she wrote: S ≈ 0.31.
When it was written she added no explanation. Below it, blank.
At eleven a letter came from Kramer, one question only: who do you plan to show this number.
She replied: first, the names on the memo.
The driftwood in the tank had been chewed to a third notch. She logged the water temperature: twenty.
She hung the earphones on the back of the chair. In her right ear the hiss still sat at its new height. Day fourteen.