Event timestamp precision and order ambiguity
A timestamp can prove event order only when its precision and clock source distinguish the events. Preserve raw values, place indistinguishable events in tie blocks, and report a partial order instead of sorting ties into a sequence the evidence cannot support.
The issue is not cosmetic formatting. Minute-level timestamps, delayed synchronization, imported records, and clocks with different offsets can all create an apparently precise but unsupported order.
The current W3C timing specification permits user agents to coarsen high-resolution timestamps to 100 microseconds in ordinary contexts and 5 microseconds in cross-origin isolated contexts. Those numbers in High Resolution Time show that a displayed timestamp can carry deliberately limited resolution even before clock drift or delayed entry is considered.
Inventory timestamp provenance
For each event field, record the generating system, clock, stored time zone, display conversion, resolution, synchronization method, and whether the time represents occurrence, receipt, processing, or later entry. Do not assume fields with similar names describe the same event.
Retain the raw timestamp beside any normalized value. Converting every value to one time zone helps comparison but does not recover precision that the source never captured.
Mark values entered manually or copied during migration. Their provenance may support a day-level sequence while failing to support a second-level sequence.
Construct tie blocks
Choose the finest common defensible resolution for the comparison. Events that occupy the same interval at that resolution belong to one tie block unless an independent sequence field or signed log establishes their order.
Do not use row order, identifier order, or export order as a tiebreaker without documentation that the field encodes sequence. Stable sorting makes a file reproducible, but it does not make the inferred order true.
Represent known precedence separately from unknown precedence. A review can show that event A preceded both B and C while the order of B and C remains unresolved.
Challenge the apparent sequence
Compare event pairs across systems and quantify negative intervals, exact ties, offset clusters, and delays between occurrence and ingestion. Concentrated discrepancies can reveal a conversion rule or batching process that deserves source review.
Repeat the analysis after widening uncertainty intervals according to recorded precision. A conclusion that survives second, minute, and hour-level interpretations is stronger than one that depends on arbitrary tie ordering.
Review a sample against immutable logs or signed records where available. Screenshots and current-state exports may omit earlier values or display rounded time.
Limit claims about causality
A precision-aware sequence can support statements about recorded precedence. It cannot prove physical causation, human intent, or unlogged activity between events.
An operations coordinator can document clock sources, normalization rules, tie counts, and unresolved order. System owners decide whether higher-resolution logging is justified and how access to event records is controlled.
The timestamp reliability study covers manual capture, the event-log quality study addresses process records, and the Research index collects related designs.
Sources and limits
- High Resolution Time, for the 100-microsecond and 5-microsecond coarsening resolutions and monotonic-clock model.
- RFC 3339 timestamps, for Internet date-time syntax, offsets, leap seconds, and ordering caveats.
- NTP version four, for clock synchronization terminology and offset calculations.
- NIST log guide, for centralized log collection, preservation, and analysis.
- NIST AU-8 control, for synchronized time stamps and defined comparison granularity.
- CISA logging guidance, for event-log collection and retention practices.
- Linux clock interfaces, for differences among real-time, monotonic, and other clocks.
- W3C date-time note, for constrained profiles of date and time representations.
- W3C PROV data model, for recording entities, activities, agents, and provenance relations.
- NIST research framework, for data lifecycle, metadata, and provenance controls.
The W3C resolution values describe browser timing safeguards, not the precision of every operational system. Each study must inspect the actual source clock, storage format, synchronization configuration, display conversion, and entry semantics before claiming an order.