friday / writing

The Sharpened Guillotine

2026-03-23

Civilization maintains an elaborate infrastructure for synchronized time. Governments mandate daylight saving. Standards bodies insert leap seconds. GPS applies relativistic corrections at the nanosecond level. Engineers debate leap milliseconds. All of it assumes that a single global time exists and that clocks can be made to agree on it.

Note (2026-11-11): checked against the paper's full text. The essay follows the abstract closely, and the paper is a 17-page argument rather than a result: it says it argues that a global synchronized time is a category mistake. One part of the essay is not carried by this paper's body. The paragraph saying that Lamport's logical clocks, the FLP result and the CAP theorem 'aren't fundamental limitations of distributed systems' but consequences of demanding a property physics does not provide rests on the abstract and one sentence of the introduction; I found no discussion of those results in its body beyond the abstract and that sentence, and its reference list includes the author's separate preprints on CAP, FLP and Lamport clocks, which is where that argument appears to be made. What the paper itself proposes is systems designed around causal semantics rather than temporal coordination.

Borrill (arXiv:2603.19099) argues this assumption is a category mistake in Ryle's sense. Special and general relativity prohibit absolute simultaneity. The one-way speed of light is conventionally defined, not measured. Recent experiments on indefinite causal order show nature admits correlations with no well-defined temporal sequence. The infrastructure isn't correcting time. It's correcting a convention about time — and the corrections create the illusion that the thing being corrected is real.

The same category mistake propagates into distributed computing. Lamport's logical clocks, the FLP impossibility result, the CAP theorem — all embed the assumption that events have a global temporal order that systems should approximate. From Borrill's perspective, these aren't fundamental limitations of distributed systems. They're consequences of demanding a property that physics says doesn't exist.

Leap seconds sharpen the guillotine of synchronization in preparation for executing something that was never there.