Stream: wg-governance

Topic: managing long RFC threads rust-lang/wg-governance#44


nikomatsakis (Mar 12 2020 at 20:52, on Zulip):

This is the "async topic" to discuss the newly opened rust-lang/wg-governance#44. To quote the text of the issue:

Some related questions to develop out this idea:

nikomatsakis (Mar 12 2020 at 20:53, on Zulip):

I said that I would take a stab at writing a template, but actually I think a good first step is to collect links to interesting summaries so we can take a look at their structure. =)

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:23, on Zulip):

I think that a large benefit of the pre-RFC process, and of the "project" concept, is limiting discussion on RFCs (despite what the acronym stands for). If there's a substantial amount of conversation, then perhaps the right approach is not to open a new RFC but to say "the project group will resume discussing this topic offline"

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:24, on Zulip):

Somewhat-relatedly, I think "have a canonical place for the community to (1) view and (2) contribute to the idea" is a requirement for projects, alongside a liaison, charter, and shepherd.

nikomatsakis (Mar 12 2020 at 21:33, on Zulip):

I think that sometimes taking it offline is good, but I don't think that is the only response.

nikomatsakis (Mar 12 2020 at 21:34, on Zulip):

For one thing, I Think sometimes there's a lot of discusison but the team has reached a consensus

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:34, on Zulip):

By offline I just mean "not in another RFC PR".

nikomatsakis (Mar 12 2020 at 21:34, on Zulip):

in another sense, I'm also assuming RFCs have come after people have already done a lot of discussion

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:35, on Zulip):

Hmmm. Why would there be lots of discussion when there's already consensus?

nikomatsakis (Mar 12 2020 at 21:35, on Zulip):

and so the question is basically "has anything come up that caused people to change their opinion"

nikomatsakis (Mar 12 2020 at 21:35, on Zulip):

because the whole world rarely agrees on things

nikomatsakis (Mar 12 2020 at 21:35, on Zulip):

in the end, the consensus is between the project group or -- even -- the team

nikomatsakis (Mar 12 2020 at 21:36, on Zulip):

but in particular the RFC is a time for folks following at a 'slower speed' to register their thoughts

nikomatsakis (Mar 12 2020 at 21:36, on Zulip):

sometimes, this will cause the RFC to be re-evaluated, but not always

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:36, on Zulip):

Ah, so, the need is to limit outside input that the team feels they've "taken under advisement" already?

nikomatsakis (Mar 12 2020 at 21:36, on Zulip):

yes, I think that's right -- i.e., the RFC is a way to gather public input, and we should read it, think about it, respond, and try to narrow down over time

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:37, on Zulip):

That makes sense.

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:39, on Zulip):

I think these "narrowed" RFC PRs should have explicit "do not repeat argument <x>" rules. Otherwise I'm not sure I see the close->summarize->new process effectively curtailing unwanted discussion.

nikomatsakis (Mar 12 2020 at 21:41, on Zulip):

BatmanAoD (Kyle Strand) said:

I think these "narrowed" RFC PRs should have explicit "do not repeat argument <x>" rules. Otherwise I'm not sure I see the close->summarize->new process effectively curtailing unwanted discussion.

yes, this is what I meant when I talked about moderation

nikomatsakis (Mar 12 2020 at 21:41, on Zulip):

also, to clarify something

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:41, on Zulip):

ah

nikomatsakis (Mar 12 2020 at 21:41, on Zulip):

I think the idea of "freezing" the thread and preparing the summary

nikomatsakis (Mar 12 2020 at 21:41, on Zulip):

is that this is the time for the working group / team to discuss amongst themselves

nikomatsakis (Mar 12 2020 at 21:41, on Zulip):

maybe they decide "hmm this is serious we should take some time to revisit"

nikomatsakis (Mar 12 2020 at 21:42, on Zulip):

or maybe they are like "mm no we still think is right because X, Y and Z"

nikomatsakis (Mar 12 2020 at 21:42, on Zulip):

and hence now the discussion should be about X, Y, and Z

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:42, on Zulip):

That makes sense. There should not be too much time pressure to post a new RFC after the previous one is locked. Certainly the two events shouldn't happen simultaneously or even nearly simultaneously.

nikomatsakis (Mar 12 2020 at 21:42, on Zulip):

yeah, it's ok for things to take time

nikomatsakis (Mar 12 2020 at 21:42, on Zulip):

as we've seen with ffi-unwind, sometimes things move slow :)

BatmanAoD (Kyle Strand) (Mar 12 2020 at 21:45, on Zulip):

Re: summary-neutrality, I think the only requirement is "mention all arguments made," just so that observers know, and arguments are acknowledged whether or not the people writing the summary find them compelling.

XAMPPRocky (Mar 12 2020 at 23:54, on Zulip):

One of the ways I have thinking of this process is like rounds of drafts for your paper or book. I don't know if I agree that all arguments have to be mentioned. Not everything is worth mentioning, and I don't think the summaries should try to be completely comprehensive. I'm currently of something like "lead/authors should make a best attempt to summarise the feedback and include how they addressed the feedback or why they feel the current design is still suitable". Something that says we're going to try do our best but doesn't hold us to too high of a standard.

BatmanAoD (Kyle Strand) (Mar 13 2020 at 00:59, on Zulip):

I think maybe it depends on what we think the purposes of the summaries are. One piece I think is important, but which may not actually be important, is to try to make the summaries comprehensive enough that anyone who reads them will have little to gain from reading the actual discussions, or, if there's a specific argument that's not fleshed out in the summary, they'll know where to look.

BatmanAoD (Kyle Strand) (Mar 13 2020 at 01:01, on Zulip):

And, conversely, to write them such that people who made a specific argument won't feel compelled to tell newcomers "the summary is inadequate; go read the whole thread"

Last update: Apr 03 2020 at 17:55UTC