A statement from members of the moderation team

Announcements
22

If the moderation team had been more transparent in what they’ve done (the complained-about request for “post-hoc” justification…also known to normal people as “showing your work”), it’d be easier to defend them.

If the moderation team had cited things like the Values or what-have-you in their moderation, it’d be easier to defend them.

If the moderation team was elected and subject to the same will of the people as the SC–instead of pulling this thin-blue-line act of investigating themselves and finding no wrongdoing (see @roberth’s comment)–it’d be easier to defend them.

As it is, well…I hope that we can all build on the learnings of their tenure.

22 Likes
23

This is very disappointing.

I felt like the moderation team took great care of the community and didn’t envy them the many difficult decisions they were faced with.

It’s a very difficult job and I worry that there has been a lack of respect for the skill necessary to do it. “Objective” decisions are impossible and while accountability is important, questioning and undermining whatever decisions were taken doesn’t seem like a step in the right direction.

I’m very sorry to see this team leave and wish the SC had prevented that. I had a lot of hope at the beginning of this first year of the SC and very little of that left, especially with the secrecy around conflicts of interests of one of its members.

20 Likes
25

Thank you for your great work and a glimpse of integrity. It has been a while in this project and its community.

3 Likes
26

Based on @roberth’s comments it would make sense if there were receipts to show that @arianvp isn’t just making things up?

I’m especially skeptical of an unelected body attempting to bully an elected body into any form of compliance.

This whole topic reeks of police force attempting to overthrow a government.

26 Likes
29

The SC has involved itself in matters of moderation since its inception, but has repeatedly failed to understand the issues in the community and the requirements of moderation.

I imagine it would be easier for the SC to understand your perspectives on moderation if you actually documented any part of your process.

This is the official NixOS moderation log: moderation/moderation-log.md at 940106d7d66f9ea34e1e37eed74d54459bffc865 · NixOS/moderation · GitHub

Observe all the things missing that would help the SC educate themselves about what it is that the moderation team does:

  • what the person said that led to the moderation action (i.e. a quotation)
  • who they said it to
  • where they said it (e.g. Matrix, discourse, GitHub)
  • when they said
  • which moderator made the ultimate call
  • how many moderators agree or disagree with the moderation decision
  • which section of the Code of Conduct was violated

Documenting these things would constitute something called “accountability”. But you dodge accountability at every opportunity. Why?

32 Likes
32

Given the fact that an independent moderation team was created by project leadership to make decisions about moderation, IMO it’s pretty hard to not take the side of that moderation team if/when they report that the steering committee is trying to exert influence over them.

At the same time, I’m disappointed that what is fundamentally supposed to be a boring package manager continues to be plagued by so much ridiculous community drama. I’m confused as to why Nix the package manager has a constitution and by-laws. I’d argue that if Nix the package manager didn’t have an official forum and an official Matrix server, it wouldn’t really need a moderation team either. I’d also argue that Nix the package manager would continue to live-on and grow, even if it didn’t have an official forum or official Matrix server.

Is it possible that Discourse (capital D, as in, this specific choice of online forum software) is part of the problem? I don’t think this is actually as ridiculous as it sounds; there are a few subtle properties of Discourse that conspire to make moderation much more challenging for Nix than it probably needs to be.

  1. The linear nature of a discourse thread allows (arguably, encourages) a commenter to use a single reply to respond to several comments at once. This prevents troll-heavy subthreads from reaching a natural dead-end. Contrast this with the threaded nature of mailing lists, where only the on-topic unresolved interesting lines of conversation are bumped and stay alive. If and when discussion becomes incendiary and attracts trolls, mailing lists make it much easier to just ignore that Message-ID and its replies. With discourse, because commenters can reply to multiple branches in a single comment, this natural filtration of bad comments does not happen; regardless of whether a reply is on-topic or not, it becomes the most-recent post and gets shown front-and-center when you open the thread.
  2. Discourse is too easy to join and access. There’s even a mobile app, which means that odds are good that there’s been at least one response on the Nix Discourse from someone stuck temporarily on the bus, or on the toilet, or on their phone before they go to bed. This is strictly a bad thing if you want to encourage long-form/in-depth discussion. One or two sentence replies are vanishingly rare with mailing lists, whereas I’d guess that the majority of posts on the Nix Discourse fall into that category.

Discourse is a relatively recent invention. Only a handful of longstanding open source communities have shut down their Mailman mailing lists and transitioned them to Discourse. Fedora is one example that comes to mind… But the overwhelming majority of established open source projects – eg: linux, yocto, gcc, glibc, systemd, debian, arch linux… all of which were built on top of mailing lists, continue to prefer listservs as their primary support forum. I think it’s fair to say that the jury is still out on whether Discourse is an improvement on what came before it.

19 Likes
33

I’ve been vaguely involved with the NixOS community for about year, so much less than you certainly. In that short time I’ve seen far more drama than I’ve seen in any other tech community I’ve ever engaged with. It seems like it spawns(at least partially) from a hellish combination of these factors:

  1. Having an official forum at all. Especially an official forum in which users can customize a profile and talk about arbitrary topics, that provoked at least some of the drama.

  2. Having the somewhat nebulous role of a moderator, on an already somewhat nebulous forum. It’s very easy for a moderators role to scope creep up when there are not explicitly defined rules for what is and isn’t allowed(as was historically resisted)- especially when there is not a bias towards significant restraint on the side of moderators.

  3. A culture of moralism. This by itself isn’t bad, I do think some rules in a forum are necessary, and I think stigmatizing and ostracizing some (in my opinion harmful) social behavior can be good. The issue is that feels like bullying when it is not equally applied, clearly defined, or equitable in how its applied.

IMO this ship has probably sailed, but it seems like it would be healthier to me to divorce any “communities” from the Nix project itself at this point. My own activity on here, despite being deeply interested in reproducible software, has been very sparse(except for on issues like this that I think are extremely important) largely because of the amount of vitriol I see in Nix discussions.

11 Likes
34

I agree with you, it seems to me that the goal is to make nix better and spread usage, I don’t believe the moderation team, SC, discourse, is actually fulfilling either of those objectives.

3 Likes
39

Sounds like you’ve not engaged with them very closely.

I’d argue the reason you don’t see this as much on the linux/debian/fedora/etc. spaces is that it’s much more hidden to the average layperson. Those projects are also largely consumed dispassionately, since they’re old and “stable” - there’s not much excitement, or new “social genesis”, unlike our relatively niche community which has been seeing a ridiculous influx of new users for about a year or two now.

That doesn’t mean similar drama doesn’t happen there. To give some examples; The amount of vitriol around e.g. some of Linus Torvald’s emails is legendary, bcachefs was practically removed from the kernel just a month or two ago seemingly just because a prominent kernel contributor clique including him didn’t like the module maintainer (likely for valid reasons, but it’s still the result of purely “social/political” human behavior). Debian also just recently had its own moderation shitstorm when a prolific contributor was banned by their steering committee for no apparent reason (iirc, the details are hazy). The debates around systemd and its impact on distros (including Arch) was unending in the early 2010s, causing forks much like nix/lix/dix/etc., and still casts waves everywhere to this day, including here - which sounds technical, but it was clearly because a handful of people personally dislike Lennart Poettering for some reason, and their initial half-reasoned arguments finding hold in the minds of people who didn’t understand them. The arguments usually carry on on reddit/discord/stackoverflow even if the initial mailing list threads die down anyway; the communities are typically much larger than the mailing list folks.

This is not a call to re-litigate those things, I’m sure there are people here much more familiar with them than I am. I just want to point out that this is perfectly normal for a software community. IMO, if you don’t think so, it’s mostly because you’re not actually paying attention elsewhere; I don’t blame you, I’d guesstimate mailing lists are utterly impractical for 80% of even software engineers, let alone laypeople, including yours truly. I’m glad I’m not in a space where I’m forced to use them anymore.

Either way, my point is, regardless of whether you’re engaged with them or not, there will always be communities around larger software projects. You simply cannot remove the people from the software, and where people meet, cultures and ideals clash. This causes rise to the emergent property that we refer to as “politics”. The result is threads like this one. You can choose to engage or not, but they will exist, regardless of how you try to constrain the medium on which they happen.

49 Likes
41

Moderation is a complex topic. You usually don’t make friends, only enemies, while doing it. You never appear fair to most people when you punish them or their friends, and our brains expect a kind of tit for tat when we punish someone. So if we punish someone speaking out against the green team, people also expect us to punish the red team, to be fair and balanced.

Most people in the project are used to interacting with democracies. NixOS is not a democracy.

Democracies are built to protect the livelihoods and well-being of people who don’t have any other choice than to be part of it. NixOS is a community built of people who choose to be part of it. Respecting and understanding that choice is what allows us to be great.

We want more people to choose our community because we have a lot of work to do. We want NixOS to rule the (digital) world someday. NixOS itself is a meritocracy. People with a bigger impact should have a bigger say in the direction. Elections cannot fully grasp the complexity of that contribution investment, but they are a good enough approximation.

I’m deeply saddened by the SC’s choice to directly interfere with the Moderation team’s composition. Keeping a motivated Moderation team alive is a complicated and fragile endeavour for many reasons. This is what a significant part of my time over the last year has been spent on, and seeing it shattered makes me very sad. I also helped establish the SC and gave them the authority to oversee the moderation team, in the hope that I could communicate the need for autonomy.

Rerolling the SC completely is not a good idea. Knowledge outside moderation topics has to be passed on, and roles/jobs have to be done. Conducting a 5/7 election is already 2 positions more than the NCA planned, and I think it provides ample room for reforms. While I disagree with the remaining members on the moderation topic, I still think they are a valuable steering force.

We, as the NCA, gave the SC a lower public communication requirement to help speed them up. Writing public statements and redacting private information from meant-to-be-public logs is work that slows down the decision-making. Even without this requirement, we have a higher burn rate of SC members than I would have hoped for.

The post before mentions that we did not enforce the CoC consistently. I don’t feel that is the case, and I expressed my opinion that not all violations of the CoC have to be enforced the same. Speaking with violators first usually yields a better outcome than simply punishing them directly. Moderators are not machines but humans with context, understanding and friends. More often than not, the person you need to moderate is someone you know.

I understand why the moderation team resigned. I would have done the same if not in this special position. But I also understand why the SC wants to show actionability. I just think in this case the SC is misguided, and it is my fault for not communicating things earlier.

80 Likes
42

The moderation team’s processes emphasise keeping details of moderation discussions private with the individuals concerned (whether reporters or reportees), including where relevant between moderation members. Moderation gets involved when things are most heated, and it’s exceedingly rare that everyone (or anyone) leaves happy. The idea is that this helps avoid shame, minimises fear of reprisals (again against any party), facilitates frank discussions, and generally is the most constructive way to support a healthy community.

That’s a choice of policy and values, made deliberately (and, fwiw, before the formation of the SC). It’s one reason why adding other governance groups for topics without this focus is welcome. Other policy choices can be discussed and much background has been written elsewhere. That’s not the discussion we had with the SC (or less and less so as time went on).

There’s a policy and mechanism for transparent discussion of decisions with third parties. SC knows about it; SC explicitly asked us to enforce it more stringently. SC never once used it.

26 Likes
43

Random thought: would there be something to be said for delayed release of (probably somewhat redacted) discussion details after say 12 months? It might allow for a more dispassionate review of whether things are working as intended, while still hopefully not fuelling an immediate vicious circle at the time the decisions are implemented?

1 Like
44

FWIW, it’s not a priori obvious to me that this should be the goal of moderation team. For example, in the Rust community there’s a pretty strict rule that the stuff which is simultaneously non-technical and controversial is not discussed in the official spaces.

16 Likes