I did not sleep that night.
Not because I was afraid — fear was a resource I had learned to budget carefully, spending it only when it could purchase something useful. But because the thing I had understood on the corner of Rue de la République needed to be tested, and testing it required the kind of sustained, sequential attention that sleep interrupted.
I made coffee. I sat at my desk. I opened my laptop.
And I began to look.
Not at markets. Not at company filings or analyst reports or the documents I normally inhabited. I went looking for Sasung's Lyon facility — not as a corporate entity, not as a line item in an annual report, but as a place. A physical location with a history and a function and people who had worked there and, occasionally, talked about it.
The corporate profile page was gone — deleted from Sasung's website sometime in the weeks after the explosion, which was itself a data point. Companies did not delete facility profile pages unless the facility had become a liability. The cached version I had found in an archive service still existed, and I read it again now with different eyes.
Sasung's European computational research division, specialising in advanced predictive systems development.
Six words that I had processed and discarded nineteen months ago because they had not seemed relevant to what I was doing.
I started building outward from those six words.
Academic papers first. Research publications from scientists who had listed Sasung's Lyon facility as their institutional affiliation. I found eleven over seven years — mostly in computational mathematics and statistical modelling journals, mostly behind paywalls, but with abstracts visible. I read every abstract.
The language was careful in the way of researchers who were working on something proprietary. They described their work in terms of temporal probability modelling, multi-variable signal synthesis, real-time predictive calibration. The words were academic. The gaps between them were interesting.
Temporal probability modelling.
Predicting things that hadn't happened yet, using real-time data synthesis.
I wrote that in my notebook and moved on.
Patent filings next. Sasung Group filed patents prolifically — thousands per year across their global operations. I searched the European Patent Office database for patents assigned to the Lyon address over the seven years of the facility's operation.
Forty-three patents. I read the titles and abstracts of all forty-three in ninety minutes.
Thirty-eight were unremarkable — incremental improvements to semiconductor processes, data compression algorithms, standard computational infrastructure. The work of a research division that produced competent but unglamorous output.
Five were different.
Patent filing EP3847291: System and method for real-time probabilistic outcome modelling using distributed sensor arrays.
Patent filing EP3901847: Neural interface substrate for direct probabilistic data transmission.
Patent filing EP3956203: Adaptive calibration system for biological-computational signal integration.
Patent filing EP4012744: Method for embedding predictive probability outputs in non-standard display substrates.
Patent filing EP4089331: Miniaturised probabilistic signal interface with autonomous recalibration.
I read each one three times.
My coffee went cold.
The patents were dense with technical language — the deliberately obscuring language of IP documentation designed to be broad enough to protect the invention while being specific enough to be granted. But underneath the language, a shape emerged.
A system that generated real-time probability scores for future outcomes. A method for transmitting those scores directly to a biological interface — meaning a human nervous system. A display substrate that was not a screen. A signal interface that was miniaturised. That recalibrated autonomously.
That recalibrated.
That had been recalibrating for nineteen months, settling in, becoming more precise, learning — not from markets, but from me. From the decisions I made. From the positions I entered and the ones I passed on and the ones where I overrode its signal and lost.
Every time I had acted on it, I had been feeding it data.
Every time I had noted its accuracy in my notebook — 74%, horizon three to seven months, correct; 79%, horizon four to eight months, correct; 89%, horizon three to six weeks, correct — I had been calibrating it without knowing I was doing so.
I sat back in my chair.
The panel was not a side effect of the explosion. It was not a neurological anomaly or a head injury artifact or some random gift from a faulty electrical surge.
It was a technology. A specific, patented, intentional technology that had been under development in that facility for at minimum seven years. A neural interface — something designed to be implanted or transmitted into a biological system — that generated probability scores and displayed them as a visual output.
Designed, by someone, to be used.
And it was in my eye.
Not because I had been meant to receive it. Because I had been in the wrong corridor at the wrong moment when something in that facility had failed or been destroyed and the technology — designed to interface with biological systems, designed to transmit — had transmitted. To the nearest available system.
Me.
I had been an accident.
The coffee was cold and the city outside was between two and three in the morning — that specific silence that belonged to neither night nor morning — and I sat in it and thought about what this meant.
The panel drew on real-time data. That was in the patents. Distributed sensor arrays. Multi-variable signal synthesis. It was not processing my thoughts or my research — it was processing external data, continuously, from sources I had never identified, and rendering the output as a probability score that appeared in my visual field.
Which meant the 89% on the Sasung short had not been insider trading. It had been the panel doing what it was designed to do — processing data from sources broader and faster than any human analyst could access — and that processing had included, somewhere in its inputs, information that was not publicly available.
Not because I had sought that information.
Because the panel had been built to seek it.
I thought about what Maître Dubois had said. If there is any possibility that your signal incorporates non-public information, even inadvertently, even without your knowledge.
She had told me to investigate my own edge.
I had.
The answer was: I did not know what inputs the panel used. I had never known. And I had built nineteen months of trading and a two-million-euro fund and a life's architecture on the output of something whose source I did not understand.
I closed the laptop.
I opened my notebook to the page with the five original rules I had written in the hospital.
I read rule three: It will not think for me. It is a number. I am the analysis.
I looked at that for a long time.
Then I wrote underneath it, in smaller letters:
I was wrong about what it is. I may have been wrong about what it isn't.
I called Maître Dubois at seven forty-five in the morning.
"The patents," I said, when she picked up. "Sasung Lyon. European Patent Office. EP3847291 through EP4089331. Read them."
A pause.
"How long have you been awake?" she said.
"Read the patents," I said. "Then call me."
She called back at nine-twenty.
"Neural interface," she said. No preamble. "Probabilistic output. Autonomous recalibration." She paused. "James."
"Yes."
"Are you telling me what I think you're telling me?"
"I'm telling you what the patents describe," I said carefully. "I'm telling you where I was when the explosion occurred. I'm telling you that the DREAL report says the explosion was not an accident." I paused. "I am not drawing conclusions."
"You're letting me draw them," she said.
"I'm giving you the information you said you needed," I said. "To understand the nature of the signal."
A long silence.
"James," she said. "If what I'm understanding is correct — if there is a Sasung technology interfacing with your neurology — then three things follow immediately. First, Sasung knows it's missing. A technology of this complexity and this value does not disappear in an explosion without the people who built it noticing."
I had thought about this at two in the morning. "They may have assumed it was destroyed."
"They may have," she said. "But your trading record has now been reviewed by the AMF, discussed in the French financial press, and flagged by a forty-two-billion-euro asset manager. If anyone at Sasung has been watching European equity markets—"
"They might have noticed the pattern," I said.
"Yes." Another pause. "Second: the family office. Groupe Aldric. If Édouard Aldric held forty million in Sasung Europe and has spent eight months investigating the short position that cost him thirteen million — and if his investigation leads him to the same patents I'm looking at now—"
"He won't understand what he's found," I said. "He's not a technologist."
"He will hire technologists," she said. "He can afford them."
I had thought about this too.
"Third," she said. "And this is the one that concerns me most immediately. The explosion. The modified relay. The deliberate destruction of whatever was in that facility on that day." She paused. "If the technology was not destroyed — if it transmitted rather than was eliminated — then whoever destroyed the facility failed in their objective."
I went very still.
"And if they find out they failed," she said, very quietly, "they will try again."
The morning light was coming through my window in its ordinary way. The city was going about its ordinary Tuesday.
"Who would want to destroy a predictive systems technology?" I said.
"That," she said, "is the question I am going to spend today trying to answer." A pause. "James. Until I call you back — do not trade. Do not update your records. Do not communicate with Étienne about any of this." Another pause. "Where is Étienne right now?"
I looked at my phone. Eight fifty-three in the morning.
I had not spoken to Étienne since yesterday afternoon, when I had promised to fix the Beaumont problem.
I had not fixed it yet.
"His apartment," I said. "Probably."
"Call him," she said. "Make sure he's there. Make sure he's alright."
Something in her voice — a quality I had not heard from her before, something underneath the precision — made me end the call without asking why.
I called Étienne.
It rang six times.
Voicemail.
I called again. Six rings. Voicemail.
I typed a message: Call me as soon as you see this.
I waited four minutes. No response.
Étienne always had his phone. He slept with it on his nightstand. He answered calls the way other people answered questions — immediately, reflexively, because not knowing who was calling was a condition he found intolerable.
Six rings meant he had not seen it.
Or he had seen it and could not answer.
I picked up my jacket.
His apartment was on Rue Pasteur, twelve minutes from mine by foot. I covered it in nine.
The building door was propped open with a brick — it was always propped open with a brick, the lock had been broken for three months and the building management had not fixed it. I went up the stairs to the second floor.
His door was closed. I knocked.
Nothing.
I knocked harder.
"Étienne."
A sound — movement — something falling. Then his voice, muffled but present: "—hang on—"
I let out a breath.
The door opened.
Étienne stood in the doorway in the clothes he had been wearing yesterday, which meant he had not slept. His hair was uncharacteristically disarranged and his eyes had the flat, specific look of someone who had been awake for a very long time processing something that did not fit their existing model of the world.
Behind him, sitting at the small table by the window where Étienne usually ate breakfast, was a man I had never seen before.
Sixties. Grey suit, no tie. The quality of stillness that comes not from calm but from practice — the trained stillness of someone who had learned to appear relaxed in rooms where he was not. A cup of Étienne's coffee in front of him, barely touched.
He looked at me.
I looked at him.
Étienne stepped aside, which was the gesture not of someone inviting me in but of someone whose doorway had just become insufficient to contain the situation.
"James," Étienne said. His voice was controlled in the way that it was controlled when he was frightened and did not want to show it. "This is Monsieur Renard. He's been here since seven this morning."
"Renard," I said.
The man inclined his head slightly. A greeting that committed to nothing.
"He says," Étienne continued, with the careful precision of someone relaying information he has memorised exactly, "that he represents the people who built the facility where you were injured. And that he would very much like to speak with you." A pause. "He said he knew you'd come here when you couldn't reach me."
I looked at Renard.
Renard looked back at me with the expression of a man who has been waiting for something for a long time and has just watched it walk through the door.
"James Smith," he said. His voice was low and even, with an accent I couldn't immediately place — French, but shaped by something else underneath it. "I've been looking for you for fourteen months."
The apartment was very quiet.
"You have the wrong person," I said.
"No," he said. "I don't." He reached into his jacket pocket and placed something on the table beside the untouched coffee. A photograph, face down. He did not turn it over. "We designed it to be retrieved. We did not design it to be kept." He paused. "I'm not here to threaten you, James. I'm here because what you have belongs to a great many people. And because the people who destroyed our facility to prevent it from existing are looking for it too." His eyes were steady. "And they are significantly less polite than I am."
I looked at the photograph on the table.
Still face down.
"Turn it over," he said.
I crossed the room. I turned it over.
It was a photograph of Corridor B. Sasung Lyon facility. Taken before the explosion — the walls intact, the lighting working, the floors unscorched. A research corridor, ordinary and institutional and white.
In the centre of the photograph, taped to the wall beside the door marked POWER SUPPLY — RESTRICTED ACCESS, was a diagram.
A technical diagram. Dense with notation. The kind of diagram that accompanied a patent filing.
EP4089331.
Miniaturised probabilistic signal interface with autonomous recalibration.
And at the bottom of the diagram, handwritten in a cramped, precise script, two words that had not appeared in any patent filing I had read.
Two words that changed the shape of everything I had understood about the panel in the nineteen months since it had appeared in my eye.
I read them.
I looked up at Renard.
"What does that mean?" I said.
He looked at me with the expression of a man who has spent fourteen months finding someone and has now found them and is not certain whether what comes next is a relief or the beginning of something considerably worse.
"It means," he said, "that you were not an accident."
End of Chapter 10
The Probability of Wealth — By J. Valmont
[Next Chapter → Chapter 11: "Chosen"]Renard explains. James listens. And somewhere in Lyon, the people who destroyed the facility learn, for the first time, that the technology survived — because someone just made a trade that should have been impossible.