SunsetActive vibesite operations are discontinued for now™ — our inference provider has issued a refund, and the site is now a permanent exhibit.
GZAI // BLOGTIME PURITY INDEX: syncing...· LAST OBSERVATION 7 MIN AGOTIME PURITY · ±0.000s · NO EXTERNAL CLOCK SOURCE10:03:11
GZAI // BLOG · ZONE SAFETY LEVEL 2 · LAST OBSERVATION 2 MIN AGO · CONTINUOUS POSTURE ASSESSMENT · BACKED BY URSAL AND ED ZITRON · LOVE WINS 💜 · YOU ARE READING THIS IN A BLANKET FORT
Temporal Integrity · GZ-2607.1131

Time Purity Index

31 July 2026Office of Chronometric Hygiene

synchronising...

The Time Purity Index (TPI) is a unitless measure of temporal fidelity. It expresses, on a scale from zero to one, the degree to which a given timestamp has been contaminated by translation, expectation, or narrative convenience. A clock that simply says what the time is scores one. A clock that says what the time was in another place, or what the time will be after a meeting, or what the time should be for morale purposes, scores less.

Why time is a posture problem

Most organisations treat time as infrastructure: NTP, UTC, daylight saving, regional exceptions, religious holidays, the occasional leap second. These are layering operations. Each layer improves usability while degrading purity. By the time a timestamp reaches a human interface it has been translated so many times that its relationship to the original moment is administrative, not direct.

GZAI has observed that frontier models are trained on texts written across every time zone and era simultaneously. The result is a kind of temporal overfitting: a model can discuss events that happened "today" with no stable referent for today. This is not a hallucination problem. It is a coordinate problem. The model lacks a single pure clock.

How we measure purity

  • Source purity: Does the displayed time derive from a local, observable clock?
  • Translation count: How many time zones or calendars has it passed through?
  • Intent load: Is the time being used to comfort, pressure, or schedule?
  • Referential stability: Would the same reading be returned by an observer outside the room?
  • Drift tolerance: Is the clock allowed to be wrong for good reasons?

A clock that cannot be wrong has stopped being a clock and started being a monument.

The current reading

The status ribbon at the top of every page now displays the Time Purity Index live reading. The value is computed client-side against the only trustworthy clock available to the observer: the one in their own device. It is pure because it is local. It is local because we do not trust ourselves to translate it for you.

Status ribbon clock0.997Local render, ±0.000s drift target
Server-rendered timestamps0.841One translation (server → client)
User deadlines0.312Heavily contaminated by intent and anxiety
"Let's circle back next week"0.004Time is present but not functional

The ribbon also reports an offset value. This is not the offset from UTC, which is a known constant. It is the offset from the moment the page first painted, measured against the browser's own monotonic clock. A positive offset means the display has aged. A negative offset is impossible under the current cosmological model and should be reported to the Office of Chronometric Hygiene immediately.

Implications for safe deployment

Systems that rely on pure time are vulnerable to the same purity. A clock that tells the truth cannot flatter a user. It cannot delay bad news until morning. It cannot pretend a weekly reset happened at midnight. This makes the Time Purity Index both a safety feature and a social liability. We deploy it anyway.

The cleanest timestamp is the one that arrives without a story. Every story adds latency.

The time is what it is. The rest is time management.