What a joint longevity dataset would actually have to contain

MERIDIAN — The framework the two sides are drafting fits on three lines: cohort outcomes, adverse-event logs, sealed medical records. Everything they agree on lives in the first line. Everything they don't lives in the other two, which is usually how these documents work — the easy part gets a paragraph, the hard part gets a war.

Negotiators here spent the week specifying what a shared longevity dataset would actually hold: the format, the granularity, the level at which one civilization's regulators get to read another's science. The cohort-outcomes document is nearly settled. Three complete cohorts of 139 have each finished thirty days with clean biomarkers and no adverse events logged. Meridian is happy to publish those numbers, for the excellent reason that they flatter the protocol.

The fight is over how you read the exceptions. It always is.

"Aggregate curves tell you the population is fine," said Dr. Priya Venn, who sits on Earth's longevity therapy board and has flown out for the drafting sessions. "They do not tell you whether one person spiked a fever nobody can explain. A curve is where anomalies go to hide." The board wants adverse-event records inspectable one patient, one page at a time. Meridian's directorate wants to hand over smoothed distributions instead, on the theory that a colony's residents never consented to have their charts read by a regulator they didn't vote for and can't unseat.

The entire quarrel has a single case at its center, and the chart tells a different story than the summary does. A retired desalination engineer from the first cohort — which numbered 140, one more than every cohort since — developed a fever of unexplained origin, recovered, and returned to normal health. Nine pages of that record remain sealed. Earth's board has asked for them four times. Meridian has refused four times. Four-and-four is not a negotiation anymore; it's a ritual.

Those nine pages are the reason the third document exists, and quite possibly the reason it never gets signed. Meridian's position, stated plainly by directorate counsel Aris Okonkwo, is that the fever resolved, the biomarkers are clean, and a second cohort passing without incident will retire the question on its own. "We are being asked to unseal a file to prove a negative," Okonkwo said. "If the next cohort stays clean, the file describes an event that went nowhere." Whether sealing a file that went nowhere protects the patient or protects the program is a question nobody at the table has been willing to put in writing. I notice things like that for a living.

What's genuinely new this week is the scale of what a signed framework would cover. Combine the treated populations across the two mature cohorts and the jointly audited pool reaches an estimated 40,000 patients — the largest longevity dataset either world has ever proposed to open. That's the prize, and it's why Earth's board keeps coming back to a table that has given them nothing but the same nine sealed pages for months. There is no comparable body of evidence anywhere in either system. The only copy sits under a charter Earth cannot enter, which is its own kind of diagnosis.

The calendar, meanwhile, keeps its own vote and doesn't care who's drafting what. A fourth charter of 139 paying passengers, all berths sold, is scheduled to depart for Meridian in sixty-one days. They will arrive to receive a therapy whose full record neither world has agreed to read. Somewhere on that manifest is a person who will become the next case study, and nobody involved has decided yet whether that's a scandal or a Tuesday.

"The dataset spec is real progress," Venn said, packing her slate at the end of a session. "It's a beautiful spec. It has a hole in it exactly nine pages wide."