A mission clock, and finding the second it started

The footer used to say the work had been running since 2026. It now counts mission elapsed time in UTC, years, then days, then the clock, ticking every second. A counter like that makes a claim to the second, so the start had to be a moment with a record behind it.

The obvious start was the paper. It is dated May 2026, and its archive record gives the same month with no day, so the first of May the site had been using was a choice, not a fact. The first commit in the paper’s repo is 6 June at 10:41:18, and the archive deposit came the next morning. The GitHub org is older, 31 May at 14:19:45. The fork of the flight framework it builds on is older again, 30 May at 10:35:10.

The earliest record was the first prompt of the paper, at 10:20:56 on 30 May, fourteen minutes before the fork. The work did not have its name yet. It was still called Meridian. A spacecraft’s clock counts from liftoff, not from its naming, so that second is T0.

The rest is keeping a second honest. Each tick schedules the next with setTimeout for 1000 - now % 1000 milliseconds, so it lands on the second instead of drifting off it. The beacon dot beside it flashes twice a second, its animation delay set to minus Date.now() % 500 milliseconds so every flash starts on a tick. Without JavaScript the footer prints the start time rather than a number that would be stale when it loaded, and a screen reader hears that start once, not a new number every second.

The year field is four digits wide. That is the point.