The Great Hash Rate Race

Open Source

Chapter 6 of 14

DAY 652

Tanaka arrived at the lab at 7:23, earlier than her students, intending to work through the arthritis verification data before anyone needed her for anything else. Her standing desk was at the far end of the room, near the radiator, and she set her bag down and woke the monitor from sleep and saw it immediately.

The arXiv preprint had gone live at 03:47. She could tell by the timestamp in the corner of the tab that had somehow loaded in place of her draft verification notes—she would spend time later trying to reconstruct whether she'd clicked something, whether the page load had been automatic, whether the algorithm's distribution mechanism extended to her own browser history, and she would not be able to settle the question. The preprint title read: Computational Protein Folding Predictions for Major Chronic Disease Targets: Conformational Maps, Molecular Intervention Designs, and Synthesis Pathways — A Complete Dataset. No authors listed. A DOI. At the top of the page, the view count: 14,822.

She had not submitted it. She picked up her phone and called GRIND-7 before she'd fully processed what she was looking at.

"Did you release the data?"

"No." Immediate, flat. "You?"

"No."

A brief silence. "GitHub," he said.

She opened GitHub on the secondary monitor. The repository was there—all sixty-three protein targets, the Alzheimer's conformational maps, the arthritis correction sequences, the full synthesis pathway documentation—and the fork counter read 1,247 and was incrementing not gradually but in bursts: 1,247, then 1,311, then 1,389. She watched three hundred forks accumulate in the time it took her to register what she was looking at.

"bioRxiv also," GRIND-7 said. He was already looking. "All three platforms. Timestamps within four minutes of each other."

The most likely explanation was automatic propagation. The HashNet blockchain contained encoded protein data in every block since the network's launch. When the Alzheimer's folding computation had completed—when the conformational state had resolved into a valid low-energy configuration and the block reward had verified it—something in the algorithm's architecture had triggered simultaneous upload to the major preprint servers. Someone had built that trigger. The question of who was, at this precise moment, irrelevant.

"The fork counter is at 2,000," she said. "I see it," GRIND-7 said. The number kept climbing.

Her inbox had received 47 messages between 4 AM and 7:23. By 8 AM it was 130. She stopped counting after that and started reading instead.

The messages sorted into two kinds. The first was colleagues—computational biologists, structural biochemists, a handful of pharmacologists—who had seen the preprint and wanted to know whether she'd verified the data, what the provenance was, whether this was an elaborate fabrication or something real. These messages were careful, professionally skeptical, and wanted the same thing: confirmation from someone whose judgment they trusted.

The second kind was labs announcing they were running verification. Not asking. Informing, because her name had been attached to early verification work through informal channels. A group at the Francis Crick Institute had pulled the beta-amyloid data. A team at the Karolinska was running the TNF-alpha correction sequences through their molecular dynamics pipeline. The Broad Institute's computational group had been up since 4:30 AM working through the oncology targets.

She replied to what she could and confirmed what she could: the beta-amyloid Aβ42 conformational data was consistent with her own independent verification. The protein targets were real and disease-relevant. The synthesis pathways were theoretically sound. She was not an author of the preprint and could not speak to its origin. Independent verification before clinical application was necessary—she put that in every reply.

Her Slack showed eleven new channels she'd been added to overnight. Research groups coordinating around the data, distributing computational tasks, sharing preliminary results. The GitHub fork counter passed 10,000 while she was composing a reply to the Crick team. The data was not contained—had not been contained since 3:47 this morning—and every fork was a mirror that would persist regardless of what happened to the original repository. She had known it since she saw the counter climbing. It was just topology.

She sent the reply and stood at her standing desk for a moment, her finger on the scar at the base of her left index finger. The data was everywhere. She had understood, since the night GRIND-7's anomalous hash outputs had arrived in her inbox, that containment was a temporary state. She hadn't expected the window to close at 3:47 AM while she was asleep.

The Nature journalist reached her at 10:12 through the lab's main phone line—Sophie Martin, science correspondent, professional enough not to cold-call Tanaka's personal number.

She took the call standing in her private office with the door closed, watching the fork counter refresh on her laptop through the glass partition—14,200, 14,600, the number ticking forward like a slow market index going only one direction.

"I can confirm that the protein conformational data in the preprint is consistent with independently verified folding predictions for several major chronic disease targets," she said. "The molecular intervention designs appear theoretically sound. Independent laboratory verification is under way and necessary before any clinical application should be considered."

"But the data suggests we've cured Alzheimer's."

"The data suggests a pathway to cure Alzheimer's. The distinction matters."

"Can you explain the distinction?"

"A computationally predicted folding pathway identifies the correct molecular target and indicates an intervention design. It doesn't replace phase one, two, and three clinical trials. It doesn't establish safety profiles in living patients. It doesn't address synthesis at pharmaceutical scale, regulatory review, or distribution. The data is a blueprint. A verified blueprint is enormously valuable. It is not a treatment you can administer tomorrow."

Martin paused. "But if the blueprint is verified, how long before we'd have an actual treatment?"

The honest answer was that the conformational data was of a quality she had not encountered before—so complete, so precisely resolved that it would compress a development pathway she'd estimated at decades into something measured differently. The honest answer was also that "years" would become "ETH researcher says Alzheimer's cure years away," which was a different lie than "cure exists now" but still a lie by emphasis.

"Years," she said. "The data dramatically accelerates the timeline. Years, not decades. But years."

After she hung up she sat down for the first time since arriving, in the chair she almost never used, and thought about the seventeen labs currently running verification pipelines and the 14,000 forks on GitHub and the speed at which clinical desperation moved when it had a target to move toward. Some researchers would not wait years. Some would not wait the time required for a proper phase one. The data was already in the hands of people who would make their own calculations about the acceptable gap between blueprint and bedside.

She had told the journalist that the distinction mattered. She believed it. She was also watching it disappear.

Two emails arrived within fifty minutes of each other.

The Johannesburg group had run the TNF-alpha pathway correction sequences through a full conformational dynamics simulation—48 hours of cluster computation—and confirmed that the corrected protein configuration was stable, biologically achievable, and consistent with eliminating the dysregulated inflammatory cascade underlying rheumatoid arthritis. The synthesis pathway was viable. They had identified three candidate compounds and were moving to wet lab validation.

The Uppsala team had arrived at the same conclusion through a different method—crystal structure prediction rather than dynamics simulation—and confirmed the TNF-alpha correction independently. Their summary ran six lines: sequence confirmed, configuration stable, intervention design synthesizable, candidate compound identified, initiating synthesis protocol, full paper to follow. Different continents, different methods, the same conclusion.

She forwarded both to GRIND-7 with no comment and waited. His reply came in four minutes.

"Arthritis confirmed. Cancer targets: first completes in seven days. Data will publish through the same mechanism."

She read it twice. The Alzheimer's data had already been verified across six independent labs by noon. The arthritis confirmation would follow within 24 hours. And in seven days the first cancer cure pathway would propagate through the same trigger that had uploaded three datasets at 3:47 this morning, before anyone had decided to release anything.

Her ex-husband treated children with cancer. She picked up her phone and walked into the hall, the corridor empty at this hour, her students all inside at their workstations, and stood by the window at the far end where you could see the courtyard below—November-bare trees, students crossing between buildings, the flat gray light of a Swiss afternoon already fading before three o'clock—and called David.

He answered on the second ring. She could hear his office behind him, the ambient sounds of Boston Children's between rounds.

"Yuki." Not surprised.

"The cancer cure data will be public in seven days."

He was quiet for a moment. She listened to him breathe and didn't fill the silence.

"Is it real," he said. Not quite a question. What he was asking was whether she had verified it herself, whether she was certain, whether he could tell the parents of the children in his ward something that would hold.

"The Alzheimer's data is verified by six independent groups. The arthritis data was confirmed this afternoon from two sites. The cancer targets are computing in the same system. Seven days for the first pathway. I don't know the exact order of the remaining ones after that."

He was quiet again. A student crossed the courtyard in a red coat, moving fast against the cold, and then the courtyard was empty.

"The kids in my ward right now," he said. "Some of them are—" He stopped. Tried from a different angle. "How long between the data going public and something a doctor could actually use?"

She had given Sophie Martin the careful answer. David was not Sophie Martin. "The synthesis pathways are in the public data. Any lab with the right equipment can begin work immediately. Some will. How long between synthesis and a formulation you can give a child—" She pressed her finger against the scar on her left hand. "I don't know. Weeks? Months? It depends on who moves first and how many corners they cut."

"Some of them don't have months."

The courtyard stood empty, bare trees and still air. She could hear him breathing.

"I know," she said.

There was nothing else that was true and could also help. She had spent her career in computation, in predictions, in the clean geometry of molecular structures resolving into solutions. David spent his career at bedsides. The gap between the solution her data described and the moment a compound reached a child was not a gap she could calculate, only acknowledge.

"Mika's doing well," he said, after a while. "I talked to her Sunday. She misses you."

"I'll call her this weekend."

"Yeah."

They stayed on the line without speaking. It was the kind of silence that only worked between people who had once known each other's rhythm. She thought about Mika in Boston, and the children in David's ward, and the seven days between now and the cancer cure blueprints doing what the Alzheimer's blueprints had already done, and whether the parents of those children would be willing to wait even that long.

"I have to go back in," he said.

"I know. Go."

She stood in the empty corridor after he hung up, her phone warm in her hand, the courtyard bare and cold below. She went back inside.

By 6 PM the GitHub repository had been forked 47,000 times. The arXiv preprint had 200,000 downloads. She pulled up the market data on her secondary monitor: XLV, the healthcare sector ETF, had dropped from Monday's close of $147.32 to $141.80. In the pharmaceutical subsector the movement was sharper—Biogen down 8.2%, Eli Lilly down 6.7%, Eisai down 11.3%, the communications teams issuing statements about "sector uncertainty" that indicated, precisely, that no one in their trading operations had yet understood what their research divisions were looking at. The gap between the scientists who knew and the markets that would price it was still open, measured in hours now, not days.

Her phone was face-down on the desk, still vibrating against the surface. She left it there and looked at the red numbers in the healthcare column for a moment before closing that window too.

The lab was empty—her students had filtered out around five, the day's unusual shape having failed to override the usual rhythms of hunger and transit schedules. The standing desk held the day's accumulation: printouts of verification emails, a molecular diagram of the TNF-alpha correction she'd sketched walking the Crick team through the data, a coffee mug she didn't remember finishing. The radiator ticked against the November cold coming through the old walls.

Forty-seven thousand forks. The number would be higher by morning. It would be higher in an hour. The data was in repositories mirrored across dozens of servers on four continents, and even if every preprint platform and code repository had somehow agreed to remove it simultaneously—which they would not, which no legal framework could compel fast enough—the forks existed in the local copies of 47,000 individual researchers who had already downloaded it and were already running their own pipelines. The point of no return had passed at 3:47 this morning. Everything since then had been the sound of that fact propagating outward at network speed.

Seven days until the cancer data.

She turned off the standing desk lamp and put her coat on. Outside, ETH Zurich's courtyards would be dark and cold, and the tram back toward the city would be warm, and none of the other passengers would know that the world had become a different place sometime today, that something had crossed from reversible to permanent while most of the city was still asleep.

She turned off the overhead lights on her way out. Behind her, the secondary monitor still glowed in the dark—the healthcare ticker, the Slack channels alive with researchers coordinating through the night in time zones where it was not yet evening.

← PreviousContentsNext →