Eric,
Some more comments in line:
On 27-May-26 05:37, Eric Rescorla wrote:
On Mon, May 25, 2026 at 10:42 PM Brian E Carpenter <[email protected]
<mailto:[email protected]>> wrote:
Eric,
On 20-May-26 11:13, Eric Rescorla wrote:
>
>
> On Tue, May 19, 2026 at 3:40 PM Brian E Carpenter <[email protected]
<mailto:[email protected]> <mailto:[email protected]
<mailto:[email protected]>>> wrote:
>
> So, I haven't seen any comments.
>
> Either that means everybody is 100% happy with the draft
>
>
> Definitely not this one.
>
> Overall I continue to believe that these decisions should be
> taken on a stream-by-stream basis,
I don't see it as very different from IPR, where the IETF (appropriately)
led the way, and the other streams followed. In this case, it just seems
natural for the Editorial stream to lead.
To you, perhaps, but not to me.
So, we need the RSWG as a whole to decide this.
>
> Other aspects are that personal or business considerations should not
> affect accuracy and balance, and any hidden conflicts of interest
> should be documented.
>
> This doesn't seem like it applies to RFCs where authors routinely
> work on commercial implementations, especially given the fiction
> that people participate as individuals. Indeed, AFAIK there is
> no requirement that people routinely identify their employers
> at all.
No. (But we do have a CoI policy for our "officials":
https://www.ietf.org/about/groups/iesg/iesg-coi-policy/
<https://www.ietf.org/about/groups/iesg/iesg-coi-policy/>)
I don't see what that has to do with authors.
You're correct. I'm mildly surprised that IETF counsel have never
suggested that a CoI policy for authors or participants was needed,
but there it is. However, RFC7154/BCP54 says:
IETF participants use their best engineering judgment to find the
best solution for the whole Internet, not just the best solution
for any particular network, technology, vendor, or user.
That indirectly covers the conflict of interest issue for IETF authors.
You're writing for the Internet, not your employer. But anyway, this text
is now relegated to the Appendix.
>
> There are quite a few subjective judgements to be made about whether
> a contribution is substantial enough to count as authorship. What
> fraction of new or corrected text counts? Is a particular concept or
> brilliant idea enough? Should the author of a previous trail-blazing
> document be invited to join? Should someone who promised to
> contribute significantly, but only contributed fragments, be removed?
> It is hard to give definite guidelines for such cases.
>
> This graf and indeed this whole section just doesn't line up well
> with RFC 2418, which explicitly invests the WG with the power to
> define a Document Editor for WG documents, without regard to the size
> of their contributions. Section 5 similarly does not line up well.
I don't really see any misalignment. You're correct that 2418 is
authoritative for the IETF stream, and I've added a reference to
2418bis in the # Editors section.
The misalignment is that the editor mihgt make no creative contributions
at all and someone who did all the original work might be removed from
the author list. This is explicitly permitted by 2418 and largely inconsistent
with your text.
I don't see anything in 2418 (or 2418bis) about removing authors, and
I'm not sure I recall any cases of authors being removed against their
will. The text covers the case of voluntary withdrawal of an author.
>
> S 9.
> * If a substantial part of the document was created by AI this must
> be disclosed clearly and in detail, typically in the
> Acknowledgements section.
>
> I do not agree with this as a general requirement. As I said
> originally, the place to start here is with some description of the
> problem this requirement is trying to solve.
I'll concede that "clearly and in detail" is unjustified. But IMHO
the need for disclosure doesn't need any justification.
It's not a matter of justifying it in the text (yet). Rather, I don't
agree that there is a need for disclosure, so I'm asking you
to offer an argument.
Naturally, I asked an AI (Google) about this, and it said
(among other things that seemed irrelevant):
"Copyright Eligibility: In many jurisdictions, including the United States,
works that are entirely generated by AI lack human authorship and cannot be
copyrighted. Authors must disclose AI text to ensure that only the human-generated
portions are protected.
Avoiding Fraud: Failing to declare AI usage on copyright applications can
invalidate your registration and may be deemed fraud upon regulatory bodies
like the Copyright Office.
Preventing "Hallucinations": AI models can hallucinate false facts, references,
and data. Disclosure allows readers and reviewers to critically evaluate claims and
ensures the author takes full accountability for verifying the material.
Reader Preferences: Readers expect transparency regarding AI involvement to assess
the value and credibility of what they are consuming."
I'm not going to put that in the draft, but I will add a
few words and I'm sure there will be a lot of discussion.
Brian
--
rswg mailing list -- [email protected]
To unsubscribe send an email to [email protected]
