Re: Dealing with LLMs in IETF discussions draft

Robert Sparks <[email protected]> Fri, 31 Jul 2026 11:16:32 -0500
Newsgroups gmane.ietf.general
Message-ID <[email protected]>
This is a multi-part message in MIME format.
--------------2rbpHr7h1JfEvnQRS60kytLG
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit

inline

On 7/31/26 10:13 AM, Stephen Farrell wrote:
> On 31/07/2026 13:43, Orie wrote:
>> What do you mean by "know how to interpret"?
>
> Isn't that described in the draft? If not, be interested
> in fixing that.

A stray thought that I wanted to capture and here seems maybe a good 
place to insert it.

"Voice" in IETF email correspondence is a thing. We learn to read each 
other's messages (like the very terse two above) in the "voice" of the 
sender. It's part of why AI generated or restated content sticks out. 
That idea is scattered through the draft, but not explicitly mentioned.

Perhaps at least the risks discussion should more deeply explore the 
consequences in the quality of interaction when the original 
contributor's  "voice" is being lost.

(I asked Claude to rewrite Stephen's message above. The result, while 
preserving "fidelity", was _definitely_ not something Stephen would write).

>
> S. 
--------------2rbpHr7h1JfEvnQRS60kytLG
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: 8bit

<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <p>inline</p>
    <div class="moz-cite-prefix">On 7/31/26 10:13 AM, Stephen Farrell
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:[email protected]">On
      31/07/2026 13:43, Orie wrote:
      <br>
      <blockquote type="cite" style="color: #007cff;">What do you mean
        by "know how to interpret"?
        <br>
      </blockquote>
      <br>
      Isn't that described in the draft? If not, be interested
      <br>
      in fixing that. <br>
    </blockquote>
    <p>A stray thought that I wanted to capture and here seems maybe a
      good place to insert it.</p>
    <p>"Voice" in IETF email correspondence is a thing. We learn to read
      each other's messages (like the very terse two above) in the
      "voice" of the sender. It's part of why AI generated or restated
      content sticks out. That idea is scattered through the draft, but
      not explicitly mentioned.</p>
    <p>Perhaps at least the risks discussion should more deeply explore
      the consequences in the quality of interaction when the original
      contributor's  "voice" is being lost.</p>
    <p>(I asked Claude to rewrite Stephen's message above. The result,
      while preserving "fidelity", was _definitely_ not something
      Stephen would write).</p>
    <blockquote type="cite"
      cite="mid:[email protected]"><br>
      S.
    </blockquote>
  </body>
</html>

--------------2rbpHr7h1JfEvnQRS60kytLG--