Possible? Maybe. Practicable? I’m dubious; it would require some level of human interaction in order to be any good, and “20 posts” could be less than an hour in the hottest of discussions. But if someone wants to make a HUGELY beneficial contribution to a thread like this, maintaining that sort of wiki summary would be it. Even if the wiki lags behind the thread by a day, that would allow people to “catch up” by reading the wiki page and then the new posts since it was last updated.
But in order to be useful, that would definitely require more than just an AI summary (which people can already request just by clicking a button). Which means it’s a lot of work. Who knows though, maybe someone will step forward, even if only for a single PEP that they are personally passionate about.
Doesn’t even need to be a wiki post (which involves finding a wiki somewhere to host it). It could just be a regular “the story so far” posting to the thread, summarising where we are up to.
As you say, it’s a lot of work to do, but it would be immensely useful, especially on longer threads, and even more so once the person (or people) doing it have gained a certain amount of community trust.
I almost talked myself into doing summaries for the PEP 505 revival before it tapered off itself. I was relatively neutral on the proposal itself but after having seen it stuck in a loop for so long I was keen to see it finished – with acceptance or a permanent death – I didn’t care. Anything to get away from the ~10-20 message FIFO that’s what any reasonable lengthed discussion devolves to…
If not a dedicated discussion summary, possibly the PEP itself should more proactively absorb discussion points and their resolutions. The PEP will get long but a long PEP is still a lot shorter than a long DPO thread.
That’s true, but that would create a tension between wanting the summary to be reasonably current (Carol suggested “every 20 posts or so”) but not wanting the summaries themselves to be a significant amount of noise. Having it be a separate page, always accessible, would remove that need. Ideally, it would be part of Discourse itself. The term “wiki” usually implies that multiple authors can collaborate, but if it has only a single author, this post could simply be a thread of its own that keeps getting edited. (At least, that’s assuming thread authors aren’t time limited in how long they can keep making edits; I think that might be true for higher trust levels but not for everyone. So it might require that our courageous summarizer be granted a high enough TL.)
Agreed; and I will also say, doing a good (or even just decent) job of this would be a great way to gain that trust. For someone who has an interest in Python language discussions, this would be a huge step up, without requiring any kind of permissions beyond having a place to edit.
Discourse has a wiki feature. I used it years ago when I wrote the how to use Discourse post here. But I have forgotten what I did. You can change an existing post to a Discourse wiki post.
Apart from technical details, could someone provide a summary here so we can assess whether it suffers from common summarization issues (for example, decontextualization) and whether any information is lost? In other words, will meta-discourse lead to longer discussions than the original discourse itself?