Colin Perkins

Exploring IETF Email Interactions

5 November 2021 / protocol-standards

The development of IETF standards takes place by a mix of plenary and interim meetings, and continuous, ongoing, discussion on mailing lists. We analysed the public IETF mailing list archives to understand the role email plays in the development of Internet standards, collecting 2.4 million emails, sent by 75 thousand participants to 1153 mailing lists during the period 1995-2000.

The figure below breaks down the number of emails sent per year (red dashed line), showing that email volume has grown significantly with time, plateauing at around 130,000 messages per year since around 2010. The figure also shows how the number of people sending email in each year (solid blue line), showing a decline in the number of people participating since 2007.

These trends align with the trends in RFC publication rates, where we previously observed the peak in number of standards documents published in the mid-2000s, and with the observations that Internet standards are becoming more complex since the people involved are sending more emails to develop fewer documents.

The next plot (below) categorises the emails by the type of sender: “datatracker-mapped” emails are those sent by people with an account in the IETF’s systems, “automated” emails generated by various systems, “role-based” emails comprising announcements from the IETF chair and similar addresses, and “unmapped” addresses that can’t be characterised.

We see that an increasing fraction of messages originate from automated addresses. This is largely as a result of the growing use of GitHub, and other version control systems, for managing drafts, with email notifications from those systems being logged to a mailing list for archival purposes. There were 122 active IETF working groups at the time of writing: of these, 17 listed a GitHub repository in their metadata. The QUIC working group, as one example, has replaced the typical email list discussions with GitHub issues: indeed, this is a significant part of the surge in automated emails in the last few years.

To measure how much of the email discussion relates to particular documents under development, we next identify mentions of working drafts and RFCs in the mailing list messages. As next plot shows, we see a strong increase in the number of mentions over time, correlating strongly with the increasing number of documents being published over time. This speaks to the influence of email in driving document production.

Finally, we look into the longevity of contributors, in terms of the number of years in which they have actively participated in IETF mailing lists, and categorise the authors of each document by the length of their participation. We then look at the experience of the junior-most and senior-most authors of each document.

Our results, below, show that the majority of junior-most authors have participated for less than 5 years in the IETF, whereas the majority of senior-most authors have participated in the IETF for more than 10 years. In fact, 35% of authors exceed 15 years of IETF participation, showing that RFCs tend to be authored by a mix of seniority levels, showing some evidence of mentorship and passing on of experience between generations.