• RPC created an AUTH48 repo and invited the author as a collaborator.
• Each AQ was made into an issue (10 total).
• Both editor and author created branches and submitted PRs (6 total) to close issues.

This third experiment involves a small document (one author, 24 pages, not in a cluster, 10 AQs at the end of the EDIT pass) and a willing author and AD, who had agreed to the experiment when the document entered the queue.

In this experiment, the RPC created and controlled the repo. Both pull requests and issue tracking were used.

• rfc9131-AUTH48
• Initially set to private.
• Contains:
  • rfc9131.xml
  • README.md
  • CONTRIBUTING.md
  • note-well.md

• rfc9131-AUTH48/issues
• Initial number of issues: 17
• Issue labels:
  • rfced - initial AQ
  • question - follow-on question added to an issue
  • AD approval required - document update requires AD review and approval
  • editor-ready - the editor can update the document and/or close the issue based on issue discussion

Paul Hoffman, chair of the git WG and author of RFC 8875, and Barbara Stark, author of RFC 8874 and member of the Education, Mentoring and Outreach Directorate (emodir), were invited to review the repo and the draft email messages before AUTH48 was started.

Based on their feedback, we updated the AUTH48 AQ email and added details to the README file. Barbara also provided information on pull requests that will be useful for future readmes.

We also consulted with Jay Daley (IETF LLC Executive Director) and John Levine (member of the IETF Trust, RSOC, and Tools team) regarding boilerplate files. We added CONTRIBUTING.md and note-well.md based on their feedback.

AUTH48 started 2021 Sept 15 and ended 2021 Sept 29. The document was published 2022 Oct 1.

There doesn't seem to be a way to make notifications more verbose (that is, provide more context).

• What were the advantages of using GitHub during AUTH48?
  • Erik Kline (AD):
    • Being able to see meaningful diffs as text evolved; being able to track issues one by one and review comments on each one separately; being able to see how a pro handle the XML.
    • The legacy model of one giant email with all the proposed changes and 5 different authors and their email agents injecting fonts, colors, HTML indentation, and so on in one long thread was so easily eclipsed by the organisation and simplicity of this approach. I don't ever wanna go back.
  • Jen Linkova (author): It was much easier for me to track all issues. When we were using emails, it was too easy to miss some questions/comments from the RFC Editors. With emails, I was trying to respond to all issues in one email (to make sure that no issues slipped through the cracks. With Github I can easily see what issues have been addressed, and work on the remaining issues one by one.

• What were the disadvantages of using GitHub during AUTH48?
  • Erik Kline (AD): Needing to setup the alerts to route to my IETF email (this will be necessary for each new repository unless I get wise and try to make some separate github account or something).
  • Jen Linkova (author): None I can think of.

• Was communication clear and easy to follow?
  • Erik Kline (AD): Yes
  • Jen Linkova (author): Yes

• Were there any aspects of communication that were challenging?
  • Erik Kline (AD): Just getting repository alerts setup to route to my IETF email (which I see is nicely linked in the README.md).
  • Jen Linkova (author): No, all was clear. All issues I had were caused by me not paying enough attention to the documentation ;)

• How may we improve the README.md?
  • Erik Kline (AD): (no response)
  • Jen Linkova (author): Maybe it would be nice to have a short “cheat-sheet” explaining how to perform basic operations like pull requests (like fork and then pull).

• Were the issues appropriately “sized”?
  • Erik Kline (AD): Yes
  • Jen Linkova (author): Yes

• Were the issue labels helpful (e.g., rfced, question, editor-ready)?
  • Erik Kline (AD): Yes
  • Jen Linkova (author):

• In particular, did you find the process more efficient than the current email-based process?
  • Erik Kline (AD): Yes
  • Jen Linkova (author): Yes

• What parts of the process would you like to see changed in a future experiment?
  • Erik Kline (AD): (no response)
  • Jen Linkova (author): (no response)

• ADs and WG Chairs: Was it easy to track the discussion and resulting updates?
  • Erik Kline (AD): Yes

• Please provide any other comments and suggestions for improvement here.
  • Erik Kline (AD): (no response)
  • Jen Linkova (author): (no response)

• May we quote your replies on a public page documenting the experiment?
  • Erik Kline (AD): Yes
  • Jen Linkova (author): Yes