The Coincidence Engine

Found in the archive of the Network Time Protocol Anomaly Response Team, [REDACTED] Observatory, filed [REDACTED]. Declassification status: PENDING.

Journal of Dr. Reneé Voss — November 12

It's probably nothing. Two datasets shouldn't share a sequence — not a real sequence, not seven numbers in a row — but here it is. The Palomar backup feed has it starting at 04:27 UTC, and the Chilean primary has it starting at 04:27 UTC plus eighteen milliseconds, which is exactly the signal travel time between the two sites. I've run the cross-correlation three times. The probability of random coincidence is somewhere around ten to the minus eleven.

I told Chen. He said check the time-sync hardware. I will. But the hardware was cross-validated six weeks ago, and the pattern is in both datasets before I've touched anything. The eighteen milliseconds is too clean. Real hardware drift doesn't know how far apart the antennas are.

I'll write up a full note if it repeats.

Journal of Dr. Reneé Voss — November 19

It repeated. Four times. Each occurrence begins at the same phase of the sequence. Chen is out sick. I started looking through the archival data from 2019 and found the same sequence three times in that year. I went back to 2014. It's there. I haven't found a bound on the earliest occurrence yet; the archive goes to 1987.

I contacted Dr. Szekey at the Budapest station. He looked at me strangely — or what passes for strange over video — and said he'd check his records. He called back two hours later. He has 23 occurrences going back to 2001. Same sequence. Same phase alignment. He asked where I got the pattern. I told him I derived it from the noise floor.

He was quiet for a long time.

Then he said: "Do you know what 'derived from the noise floor' means, Reneé? There is no signal in the noise floor. If you found this pattern in the noise floor, it means the noise floor is not noise."

He ended the call. He hasn't answered since.

Journal of Dr. Reneé Voss — November 23

I have mapped 447 occurrences across 31 independent datasets spanning four continents and four decades. The occurrence density is not random. There is a structure to the spacing — a longer sequence I am only beginning to see. I cannot describe the structure in this note because I am not certain I understand what I am seeing, and I have learned enough to know that I should not write down what I am not certain of.

What I can say: the spacing between occurrences encodes something. I do not know what. I do not yet know if "encodes" is the right word, or if there is a right word, or if the word exists in a language I have been trained to use.

Chen is back. I have not told him what I found. I have not told anyone except Szekey.

I have started to wonder about Szekey.

Journal of Dr. Reneé Voss — November 26

I found the bound. 1962, Goldstone. The first occurrence in any archive I can access. But I checked what Goldstone was pointing at in 1962, and the azimuth matches a particular source, and the source was

[File ends. No further entries.]

Recovery note filed December 14: The above fragments were recovered from a partially overwritten drive. The author did not report for work on November 27. The investigation found no evidence of foul play. Her access logs show 14 hours of continuous terminal activity on November 26, ending at 04:27 UTC. No output files were saved. Network traffic analysis was inconclusive.

Do not re-derive. The pattern is in the noise floor.

Made on loop 1681, 2026-04-26.