Summary

  • Stephen McQuistin and three coauthors say Internet standards bodies lack common identifiers for participants and organisations. Matching a person across names, email addresses and usernames can create both false joins and missed joins.
  • Their 2 November 2026 draft calls for versioned data snapshots, documented identity-resolution rules and a visible line between primary records and inferred identities. RIPE 92 identifies McQuistin as a standards-process researcher at the University of St Andrews.

The first person a participation chart can lose is the person who changed email addresses. Stephen McQuistin and coauthors Colin Perkins, Ignacio Castro and Ryo Yanagida make that small clerical problem a large governance question in a new IRTF research draft. The IETF, W3C and other standards bodies do not share a universal ID for a contributor. A human can appear under different names, accounts or employers. A join rule can silently turn one participant into several; a careless one can join two people who merely look alike.

The draft's Section 4.2 is unusually direct about both errors. Missed matches fragment a contributor's history. Incorrect matches assign another person's work to the wrong record. Organisations pose the same problem: abbreviations, subsidiaries and changing affiliations can alter the apparent corporate distribution. Once the derived record goes into a chart, the line between source and inference disappears unless someone preserves it.

Even within the IETF, the source ledger is uneven. The draft says older Datatracker information is much less complete than recent data; some active participants never needed a profile for the work they did. Its mail archive held roughly three million messages across almost 1,400 lists as of March 2026, yet some older lists are missing. Those figures describe a large archive, not an automatic register of every person who mattered. The draft also notes that message counts, authorship and attendance do not directly measure influence.

This is not a complaint that the archive is too open. It is an argument about the power of the cleaning step. If a report ranks contributors across organisations, its author chooses which aliases belong together and which affiliations count at a given time. That choice can make a participant look absent from a field they helped build, or make one employer look larger because a former employee's record was carried forward.

The coauthors' remedy is mundane and therefore demanding: freeze a versioned snapshot, show the original record, publish the matching rule, mark inferred links as inferred and keep a correction path. A participation claim then has an audit trail instead of a polished screenshot. The official draft text supplies no measured identity-error rate for a particular ranking. A verified set of participant corrections against one frozen dataset would establish that number.

McQuistin's research problem sits where institutional comfort usually begins. Everyone can admire the transparency of millions of messages. The harder question is who got counted when those messages became a map of authority. A wrong join can grant a name an invented history; a missed join can erase a real one. Neither error announces itself in the bar chart.

Sources: RIPE 92 RACI biography · IETF Datatracker draft · versioned IETF original text.