Summary

  • In 1999, William B. Norton turned business cards from a NANOG peering session into a contact spreadsheet shared with contributors, addressing a practical problem: finding the right person at another ISP.
  • By 2001, “Peering Personals” added brief, self-described network and policy profiles to in-person introductions. The format could help coordinators identify whom to approach; it did not establish that an ISP would peer.
  • The surviving record documents design and intent, not a measured conversion rate from introduction to agreement or an enduring public directory.

Start with the person who can answer

A peering discussion can stall before anyone compares routes, locations or operating terms. One network sends a message to an address that no longer reaches the relevant team; another cannot tell whether the recipient can discuss policy, implement a change or only forward a request. William B. Norton’s late-1990s notes identify this first-contact problem in unusually direct terms: peering coordinators found it difficult to identify the right staff at another ISP.

Norton proposed a modest response. In a 22 September 1999 message, he suggested that participants at a Peering BOF put their business cards into a hat and that he email a spreadsheet to the people who contributed. The NANOG 17 notes later described the first database session. Roughly 50 ISPs had submitted cards and received the contact file. Participants could cross out information they did not want included; the spreadsheet was circulated to contributors, not posted as a public directory. Norton asked new participants to supply their own details and said he would keep a person in the loop to review additions.

That boundary mattered because the file contained information tied to real people and organisations. The notes name fields such as contact name and title, company, address, autonomous-system number, peering email and telephone numbers. A card was not a licence to publish everything printed on it. Redaction and limited distribution were part of the initial design, not incidental housekeeping.

A directory could still misidentify the decision-maker

The mailing-list exchange around the launch also records a problem that a spreadsheet could not settle by itself. Jessica Yu asked how Norton would verify someone who claimed to be an ISP’s peering coordinator. Norton agreed that a human reviewer was one first safeguard, while acknowledging that several people at one company might all describe themselves as “the peering person.” David Diaz added a distinction between people who set policy and staff who implement it on routers.

These comments make the database more interesting than a simple address book. “Who should I contact?” has at least two answers: who can explain the network’s policy and who can carry out the technical work if both sides proceed. A name on a list may help begin a conversation, but it does not prove authority, current employment, an up-to-date policy or permission to commit the network. The archive does not describe a formal identity-proofing process, error rate or appeals procedure. Human review was a practical check, not evidence of a complete verification system.

Norton’s contemporaneous slides sharpen the same limit: the contact process did not mean that ISPs would peer. This distinction protects both sides. A contact entry can lower the cost of reaching a counterpart without turning availability into an obligation or a profile into an acceptance test.

From names to short introductions

By September 2001, Norton described the Peering Contact Database as an Excel file sent about every six weeks to participating coordinators. The Peering BOF IV at NANOG 23 added a face-to-face layer he called “Peering Personals.” Coordinators could submit broad descriptions of their networks and policies: for example, whether they were content-heavy or access-heavy, required multiple locations, would consider a single location, or were open to peering with content networks. A selected group would make two- or three-minute introductions; icons behind them would make the self-descriptions easier to scan.

Participants could then approach coordinators with whom they wanted to discuss interconnection.

This moved the task from “find an address” toward “find a plausible conversation.” The format assembled three things: a name, a short account of what the network sought, and a chance to speak directly. Its practical promise was to reduce wasted approaches and help participants decide whom to meet. But those profile labels were supplied in advance by participants. The public record does not show that an independent party measured, certified or kept each attribute current.

The NANOG 25 program in June 2002 repeated the format and said the expected next step was peering negotiation. “Expected” describes the organisers’ theory of change, not the outcome. The available archive does not count resulting negotiations, completed agreements, cost savings, reliability gains or participants who decided against proceeding. Norton also said he maintained the contact effort as a community service separate from his Equinix role. That is a useful statement about how he described the work; it does not by itself establish every participant’s view of the arrangement.

The useful boundary was the one it could not cross

The contact database was neither a peering exchange nor a registry of entitlement. It did not route packets, choose policy, certify that two networks were compatible or require either side to connect. Its contribution was earlier and narrower: make a relevant human easier to find, and give that person a short, participant-supplied context for a first meeting.

That is why the distinction in the 1999 record still matters. A successful introduction is not a successful negotiation; a negotiation is not an agreement; and an agreement is not proof of a live interconnection. The surviving sources describe the first links in that chain, not the whole chain. Norton’s Peering Personals are best understood as a low-cost contact and discovery practice whose value depended on consent, useful details and the freedom of both networks to say no.

Sources

Heng Lu’s Note 64 is an editorial lens only, not historical evidence about Norton or NANOG. Image identity references only: the AfPIF public speaker profile and its portrait photograph ground the generated likeness; they are not evidence for the peering-history claims.