The SC's role in the disbandment of the Nixpkgs core team

[wolfgangwalther]

Some would call this micromanaging. Others would call it manipulative and dishonest. Lying even. So everything is still the same as a year ago. Great.

This, paired with the recent SC decision to approve Helsing as a sponsor for NixCon - it seems like history repeats itself. Over and over again. And don’t tell me “Professional Tickets” are not sponsorship - what else is “paying more and displaying your logo”? I’m really disappointed, Philip.

The core team made some very mild claims about the SC as an institution. John immediately proved them wrong, showing that it’s not only the institution, but some of the people on it in addition.

The SC as an institution is designed to fail as pointed out nicely by Julien. Unfortunately, the SC is the only institution that can change this. Whoops, deadlock. Sounds familiar? It should.

[waffle8946]

This seems worse than just micromanaging; it’s as if each SC member is themselves a micromanager of their own accord, but acting as if they have the auspices of the group overall. Surely if any team should be beholden to bureaucracy, it should be the SC? Unlike other teams, individual SC members dictating decisions sounds fundamentally unexpected to me.

[crertel]

@wolfgangwalther

The SC as an institution is designed to fail as pointed out nicely by Julien.

The SC is designed to give the best results it can given the members elected by its constituency.

If we are displeased at the ability of the SC to do what we want, come the November election it might be prudent to vote for people that reliably show up and do the annoying work.

Again, I invite everybody to check the published minutes for attendance and double-check my work.

[cafkafk]

For people going through the Steering Committees meeting minutes, I just want to point out that while they are a clear improvement from where things where last term, they’re only the sync correspondances we have ~3-4 times a month weekly, they don’t include board / SC meetings, meetings with other teams (unless they join the specific meeting as e.g. the Cuda team has done or e.g. Valentin did in relation to our grant work). It also doesn’t cover Zulip correspondances between the SC, SC private messages, our email correspondances, or our intra team correspondances on zulip etc.

All to say, while they’re a massive improvement, many things happen outside of it, and they’re often very flawed and basic, as the notetaker often can’t keep up with points, generalizes or sanities them, or just doesn’t record everything.

I did run experiments with using AI transcription with consent of all people involved, but only to enrich notes, never publishing the full transcripts,as that seems to be too controversial currently, and since I never really got any buy-in or funding from the foundation to pay for that.

And on top of all of this. You can’t read from meeting notes how someone snapped at someone else, you can’t read the context, the balls that were dropped etc, it’s a deeply sanitized and biased view, even if we produce it in good faith and I believe every member of the SC broadly has an eye towards transparency on this.

---

I hope I’m not being a buzzkill here, I just want to make sure people conclude that just because something isn’t in these notes purely from our SC syncs that it in any way is exhaustive of our activities and means you can categorically exclude something from never having happened.

I say this as a sitting member of the SC that created the repo for these notes to go into and proposed and got the vote log passed. Just in case someone thinks this is some sort of invitation to debate on a best guess. No. It’s a description of reality.

[nadir]

Being nominally present at those meetings and “doing the annoying work” is not necessarily the same, right?

At least if claims like JulienMalka’s regarding the moderation team as an example are true: Steering Committee Report - Apr 1st 2026 — Julien Malka

[K900]

I would also add that we actually have a process for async voting, and most (if not all) of the SC decisions that have been made lately have been made async - something I very much appreciate, being in a weird time zone and dealing with chronic illness that often limits my meeting availability. If anything, I’d like us to have even less meetings going forward.

[cafkafk]

@rhendric

SC member #1 sent a message in the shared SC/bootstrap team channel about this PR, ‘recommend we close’. My response was:

For what it is worth, I think a message I sent in the SC + Nixpkgs Core Zulip channel is telling here

Also you’ll find it telling I wasn’t present for those meeting logs. Personally for that decision, I got the feeling they were trying to go around my back, which is by no means the first time in this term.

I’ve had at least 3 separate cases of this so far where they’ve tried to go behind my back that I can recall on the spot I got documented, once with the timezone conflict, second around sponsorships where a decision we had tabled for next week was made 30 minutes after I left the call after going past the original meeting time by people that stayed in it after I left after the ending time, and tthis thread @rhendric mentions here being the third case.

This is by no means exhaustive to be clear, just the ones that are easily documented.

I’ve also had several reservations around the bootstrap process which I’ve written up here for those interested about stepping back from the bootstrap process due to it institutionalizing failures in due process. You’ll find the procedural failures here telling.

---

All to just say, I think a lot of people are being very defensive on behalf of the SC in this thread without any sort of substantiated reasoning behind it. So here is some substance. Take it or leave it.

As someone that’s very active and who’s the reason you have most of this transparency now, and who outsiders who have cooperated with the Steering Committee have given praise that is on record for my ability to handle procedural issues, the current members of the SC are also a heavy blocker to me conducting my duties, not just for e.g. the now resigned nixpkgs core team. It makes it hard for me to have an effective relationship with the teams, and it puts the welfare of the project at risk.

And I seriously hope you start voting with that in mind.

[cafkafk]

Idk, and some clever readers elsewhere on the internet realized that my withdrawing from the bootstrap process post was very well timed for the time the AI policy problems started going off the rails. But I’ll let you pain in the colors here.

[rhendric]

@waffle8946

This seems worse than just micromanaging; it’s as if each SC member is themselves a micromanager of their own accord, but acting as if they have the auspices of the group overall.

I don’t know, and make no claims, to what extent this is the case. It would be entirely consistent with what I know if the SC discussed this in their own channels before taking action. It wasn’t a unanimous decision if so, but what is?

@qyliss

lack of clarity from SC members about when they’re speaking for themselves or representing a joint position

Is one of the things explicitly called out as a factor by Alyssa, after all!

[JulienMalka]

For having seen the operations of the SC from the inside it is my opinion that a lot of the events that created friction with the nixpkgs core team were due to some bad communication rather than some intent to micro-manage. Obviously a pattern of problematic communication style is still a serious problem, but not as serious as some people in this thread are sometimes imagining.

I have wrote more thoughts on the matter in my latest informal report: Steering Committee Report - Aug 9 2026 — Julien Malka

I wanted to also correct the false claim that the whole efforts for more transparency in the SC were led by a single SC member. A lot of candidates at the last elections ran on a platform including more transparency. I was the one to propose the async vote process and the public vote log (see this commit message: docs(process): add async decision process after reaching majority · NixOS/steering-committee@b00fa71 · GitHub ). It was very consensual. In the same conversation that was started by this proposal, @cafkafk proposed a public repo as a home for the vote log (instead of discourse). The decision to add minutes to that repo was made somewhat organically but the past SC had already started the habbit of publishing minutes of their meetings on discourse towards the end of the term. The workload of maintaining that public repo and taking notes has been shared between members of the SC, and everyone might look at the repo history and minutes to infer who is contributing or not.

[claes]

All to just say, I think a lot of people are being very defensive on behalf of the SC in this thread without any sort of substantiated reasoning behind it.

Here is some reasoning. When, in mature organizations, there are problems for example with collaboration between teams, you talk about it.

You are a member of the SC and you look to share the criticism in the Nixpkgs Core team’s open letter with respect to micromanagement. Did you bring this up as an agenda topic? Because when I went looking through the notes, I was looking for evidence if that complaint had been raised, and what had been discussed about it. I didn’t find it. That doesn’t prove anything of course, but I would have expected it to leave some trace in meeting notes.

In general I think it would be more healthy for the project if this type of feedback was raised by teams towards SC and discussed eye to eye.

[cafkafk]

Claes yes. I have. Run for the SC I’ve done my part.

[kiara]

while i perhaps cannot judge to what extent this analysis is shared by other SC members, if so, it looks like the SC would already have the ability to delegate stuff, if that might serve to reduce politicization deadlock.

[eureka-cpu]

Stepping outside of the box for a moment, one thing that always seems to put things in perspective for me is remembering that nobody in this situation is getting paid to do this work. It’s entirely on their own time. These kinds of situations still happen even in paid roles, but the conditions are different. Generally speaking we work on things we don’t so much care about, so long as it pays the bills. Here, everyone is working because they are passionate about Nix, and so opinions, conflicts, frustration and burnout occur more regularly since the “reward” for doing work is not payment but gratification that some outcome is reached. It seems to me that the SC and teams that must report to it are not quite in sync, and since there isn’t a unifying “reward”, i.e. all motives align to make Nix better or project management more efficient but how that should be achieved is a matter of opinion, conflict arises more regularly. I’m not saying that all problems would be solved by simply making roles paid jobs, but if the method in which to achieve each goal was more clear perhaps there would be less misunderstandings and burnout, which often isn’t an issue in a work-for-pay structure since if you do not do the work you do not get paid.

[kiara]

Add a decision process and team charter rules by tfc · Pull Request #36 · NixOS/steering-committee · GitHub has gotten some further discussion on potential remedies

[kiara]

my understanding so far, be it perhaps leaning quite a bit on @JulienMalka’s blog post:

v scenarios / concern >	decisiveness	productivity	delegation	accountability	flexibility	representation
status quo
BDFL
regulate
delegate
Condorcet

legend

✅

as expected

🉑

acceptable

〰️

neither as expected nor broken exactly

❌

not working as expected

concerns

decisiveness

@JulienMalka's noted concern 'The SC is unfit to take any decision'

productivity

@JulienMalka's noted concern 'The SC is unfit to produce non trivial work'

delegation

@JulienMalka's noted concern 'The SC is not good at delegating'

accountability

having a means for the (voting) community to cause a change of course

flexibility

the ability to act on unforeseen circumstances

representation

community members have democratic means of (if voting-eligible) representation and escalation

scenarios

status quo

where we are now

BDFL

a simpler governance model, and a caricature of our pre-constitution situation

regulate

@tfc's suggestion

delegate

the unused option in our constitution, if the SC felt so inclined

Condorcet

using a majority-preferred voting method (candidates we collectively hate least) over proportional representation (candidates we personally like best), an idea alluded to by @wolfgangwalther

[7c6f434c]

I would suspect that there are more concerns:

• coverage width — one can have good decisiveness in some areas and break down in other areas
• decision stickiness — if a decision is somehow made, does this fully pause the discussion till next pre-decided cycle (elections? jubilee week?), or has absolutely no effect whatsoever. (Obviously, the truth is in-between in every case)

I think «go-write-RFC» might have described the part of actual decision-making with decisions made pre-SC better than BDFL.

Another option is leaning into referenda. Nothing restricts SC from approving a policy that is conditional on well-defined observable events, including but not limited to polls of eligible voters.

---

I think status quo is not too good on stickiness, but BDFL and Condorcet are worse. Referendum stickiness might be good, although possibly that requires the SC to decide themselves on the question wording, and do it well. RFCs that actually got finished were relatively good with stickiness, and a severe selection is involved in this statement — but sometimes that did resolve eternal lukewarm flamewars.

Coverage width is definitely bad with any BDFL who is not in fact a collective of exactly one thousand gnomes reading all the issues opened in Nixpkgs and Nix. The only way to resolve this seems to be delegation of things felt as niche enough to hand over to a team safely; RFCs kind of provided a mechanism of that, too.

---

Investment into SC process, and some of the SC decisions seems to have effectively stopped RFC process; I would observe that Nixpkgs team work on LLM policy is pretty close to how the same team would do as an RFC shepherd team — with one important amendment being: shepherds can edit the draft directly.

[cafkafk]

I really would discourage over-adjusting to a post that states saying “flakes should be stabilized” is micromanagement. That is a major category error.

Without an understanding of the fundamentals, all of this sudden need to reinvent everything is extremely misguided and honestly dangerous. And it avoids confronting the fundamental problem: the culture and the people.

This is ultimately solved by the electorate at the election. We can change the flavor of dysfunction with a process overhaul, but it won’t actually solve things, just move them around and relabel them. And honestly just seems like an ornamental way to tell ourselves we did something, rather than actually improving things.

[7c6f434c]

@cafkafk

I really would discourage over-adjusting to a post that states saying “flakes should be stabilized” is micromanagement. That is a major category error.

Indeed, it is problematic for a completely different reason than micromanagement.

There are political choices the question is about what is the largest coalition on an issue that can still have coherent position agreed.

There is technical work where the opssible options need to be found and/or invented and carefully evaluated and probably sometimes merged together.

Oh, and there is the question of what are the basic underlying guarantees that are upheld to the best practical level, and are there any people willing to do that for a specific level of guarantees on a specific question.

In general, deciding on how much of technical decisions and how much of political decisions are to be made on a topic, and when to switch modes, is hard. This hard problem seems to affect many decision processes in the Nix* projects. But this is between 1 and 2.

Flakes, though, is the situation about (3): Nix team promising to break them and refusing to promise stability — and it doesn’t look like there is any team both widely considered credible to fully maintain Nix on their own, and willing to make stability promises about flakes. It’s a pretty macro issue indeed to make «stable» mean almost nothing. (And I think the correlation of «having looked into flakes as of now» and «opposition to stabilising before a few breaking changes» is pretty strong, technically the situation is more or less established — and documented in RFC 136).

[Gabriella439]

I don’t think it’s surprising that the Steering Committee is dysfunctional, because our current voting process is not designed to select the best candidates: it’s designed to select the most ideological candidates.