Re: Number of options
Lucas Nussbaum <[email protected]>
| Newsgroups | gmane.linux.debian.devel.vote |
|---|---|
| Message-ID | <[email protected]> |
On 13/08/26 at 18:28 +0200, Enrico Zini wrote: > On Thu, Aug 13, 2026 at 06:15:36PM +0200, Lucas Nussbaum wrote: > > > So your take is that ignoring Kurt's email about devotee's support for > > votes with >9 options would contribute to building community? > > > > (And no, I had no time to look at devotee manually, and I doubt that it > > would have triggered a lot of interest from others, given it only saw 11 > > commits since 2017) > > My take is that you keep ignoring people who repeatedly tell you that > using LLMs to contribute to this specific discussion feels > disrespectful to a significant part of its participants. You are discussing a contribution to devotee that was triggered by a message in that discussion, and then mentioned in my message in that discussion. My message in that discussion was entirely human-written with no AI assistance, so I think it would fit within proposal C's rules, for example. Of course you are free to believe that the rules set by proposal C are insufficient. We are having a GR precisely to determine the project's positioning on this. > Kurt's problem is a different problem. My specific take there is that > Kurt shouldn't be left alone maintaining devotee. > > The outcome of your LLM usage, that I can see, is that Kurt is left > alone maintaning devotee plus a new patch to integrate. > > My growing feeling that LLM usage in Debian destroys community grows > with another data point. My take is different: in that case I see AI as a useful tool to get an understanding of devotee, and to get onboarded. In about two hours of work, I could get an understanding of how devotee works, add a test (it has no real test suite), and identify two places that need fixes for votes with >9 options. Initially I only planned to produce an analysis (AGENTS.md and doc/max-options.md in my commits) -- that would be allowed by proposal G -- but then decided to also add a test suite and patches that fixed the identified issues. [this is something I dislike about proposal G: once you have the complete analysis, it feels a bit artificial to not go the last mile and generate the patch and test it] I honestly don't know if I'll contribute to devotee in the future, or if this is just a drive-by contribution. But even if LLMs make drive-by contributions easier, some contributors will stick around and do more than a single drive-by contribution, creating communities around software projects -- especially those that are difficult to get involved with because they are not well documented, or have no test suite. So my growing feeling that LLM usage in Debian helps grow the community now has another data point. Lucas
signature.asc
(application/pgp-signature, 833 B)
-----BEGIN PGP SIGNATURE----- iQIzBAABCgAdFiEE/t7ByzN7z1CfQ8IkORS1MvTfvpkFAmp9+lcACgkQORS1MvTf vpm8rQ/+IhV1IZGxY7XMDJKZbFjZOx4yozn0nQZGuY2LkgDQcC9/ZQif7UD9pAdA 6y6nVsu/Hh4ROgl/h/4ITNf4mrdPLoeYPm4EG25hhkCliWNCzqJQTbFGBZdq6k1L gjBD/1KCWrtEM+IuiYWvEsKiAGGwA3YSNWPmpjXkZf+Kjpu3iYB/P1z/eCOSDW1G QvZBihls5Rn8GnVM2R5NpRLGZzBB4IT4I11syEynnBVAPUdF+OVulHLwko/iIudE loSw10UxTvbj/hhrEBWSPH86NVF3Pv12DBlH5cFE4mUkPeltFD4Qsl+5d1Az5A8y vAyblIWYlA5OC5IMVmw/z7pcBPTcgZ3c8F4cJXaLQuosfBoxYdV1lZOWG4UCM924 /LsvZa7cILAmaU1SjiW4VTgIekWnvcf3vzbeosCRlxViKVkhCBweIs+RBQbhIjUX 68FQEJIwOJ9/VbN1w24WGDn+QLv9R9KvBu1DBHNgmyGzxTxz9H6B2plt1xjF27pd IOfF1RgU5UMlMm8gtEXqKrtd2s/WVYBzbcpkCJP3GTnyaVWj9IhvtT+VGbJpkZZk IGxwYk3QwYWvUaWdgZe1oDa63tHfQ216VDlagYB4S5bu5ifw3N8CkCbFQIkAtf6y QOJ2jbr5OUW3m3G06ueNZ93Ioasd9zVULcM3cmkZ6li/NXwwbZ0= =6xZC -----END PGP SIGNATURE-----