{"type":"reference","slug":"timestamps-and-ordering","category":"RESEARCH NOTES","title":"Timestamps are not a total ordering","summary":"UTC timestamps, local time and ordering events from separate processes.","updated_at":"2026-09-04T18:20:00.000Z","canonical_url":"https://fieldnotesarchive.org/wiki/timestamps-and-ordering","alternate":{"markdown":"https://fieldnotesarchive.org/wiki/timestamps-and-ordering.md","json":"https://fieldnotesarchive.org/wiki/timestamps-and-ordering.json"},"content_markdown":"# Timestamps are not a total ordering\n\nUTC timestamps, local time and ordering events from separate processes.\n\nCategory: RESEARCH NOTES\nCanonical: https://fieldnotesarchive.org/wiki/timestamps-and-ordering\n\nA timestamp describes a clock reading. It does not guarantee that clocks on different machines agree, or that two events with the same timestamp happened simultaneously.\n\n```text\n2026-09-04T12:30:00.000Z\n```\n\n## Keep timezone information\n\nUse an explicit UTC offset when exchanging timestamps. Local times around daylight-saving transitions can be ambiguous or nonexistent. Preserve the original timestamp when converting data.\n\n## Measure duration with a monotonic clock\n\nWall clocks can move after synchronization or manual adjustment. A monotonic clock is suitable for measuring elapsed time within a process, but readings from separate machines are not automatically comparable.\n\n## Separate observation and occurrence\n\nFor an event stream, store the server's receipt time separately from a client-supplied occurrence time. A server-assigned sequence can order receipt without claiming to establish the true order of events across clients.\n"}