Eric S. Raymond to the OSS community on codes of conduct: Refuse to have one
twitter.comI’ve enjoyed https://github.com/WiseLibs/better-sqlite3/blob/master/docs/... as a code of conduct.
Simple, well-motivated, well-practiced (I think). Unlike more convoluted CoCs, it’s tightly scoped on work, not fun. Specific labor intended to produce contributions instead of enjoyable social experiences.
I feel a lot more OSS should avoid fun. Fun doesn’t seem to scale, especially when it gets in the way of real work.
What concerns me about that CoC is that the last point ("Attention to personal traits...") sounds like it could easily be interpreted to disallow members of minority groups from self-identifying. For an example of how this can play out, see:
https://discourse.ubuntu.com/t/on-discourse-rules-about-poli...
> I feel a lot more OSS should avoid fun.
This, so much. F/OSS os so much drama these days, and codes of conduct are the most effective tools to create more drama, usually.
Code of conducts are a waste of time. Trolls won’t read it, well-behaved people don’t need to be told how to behave well. It’s not like there are CoCs out there that grant you specific rights others don’t: there is no CoC that says, for example, it is fine to harass people or be a racist idiot. So they’re all a variation of:
Rule 1: don’t be a ****
Rule 2: basically rule 1 but with more words
CoCs simply exist because there is a large contingent of people that really do love making and enforcing rules.
The way I understand the necessity for CoCs is not to tell people how to behave, but to have a clear path to understanding why some people get removed or denied contributing to a project. It's easier to say "your behaviour contravenes our CoC" than "I don't like the way you're behaving".
I guess then it’s a cultural thing about avoiding confrontation, because to me there is nothing wrong to confront someone when they’re being a dick. No need to point at some rules, it’s self-evident and the project maintainers should be able to boot people that cannot behave.
I didn't mean it in that way, but people think having a codified set of rules makes it look less arbitrary and less of an abuse of privilege when they decide to boot someone.
I'm in your camp to be honest, as a maintainer I feel like I need no other justification than "I don't like you" for refusing someone access to my project.
Code of Conducts are misaligned, in telling people what is allowed and what isn't.
What is practical is a list of obligations the maintainers are willing to do for the community:
Maintainers can only directly control themselves, after all. Only through their actions does anything happen, either merging code, or removing people from accessing the project.
Knowing where the maintainers are going to draw the line is far more useful.
I think my problem with codes of conduct is that, in my experience, they're usually just accountability sinks[1].
e.g. "all violations must be reported to the committee" => "we will forbid anyone taking action without the committee's prior approval, and insist that disclosing any reports to us is a violation of policy, then refuse to take any action on reports".
So a thing that was supposed to explicitly communicate shared expectations and norms, instead becomes a way to minimize risk of bad PR for the project by silencing anyone who believed they were operating in good faith.
[1] - https://aworkinglibrary.com/writing/accountability-sinks
Ironically, "every bit of whatever authority I have" dwindles further and further every time he posts something like this.
ESRs gravity peaked around 2004.
CoC are redundant, weak, Cider House rules with the social contract. It's perfunctory, wishful thinking similar to digital government petitions meant to waste people's time with pseudo-participation.
Eric's suggestion for a CoC:
> "If you are more annoying to work with than your contributions justify, you'll be ejected."
is not unreasonable (whether written down or not) for an individual's hobby project, but for something run by a company or other organization I think more detail is required.
The OSI code of conduct seems too general, perhaps open for different interpretations or potentially selective enforcement: https://opensource.org/codeofconduct
The Linux one I prefer, since it's concise and clear, e.g. "sexualized language" and "harassment" are unacceptable: https://docs.kernel.org/process/code-of-conduct.html
My own employer's code of conduct (we have open source projects, discussion forums etc) looks similar to the Linux one. I think it is enforced about once a year, and I think the reason has been either sexual harassment or repeated personal criticism of individuals' work.
Everything can be selectively enforced and open to interpretation.
Good questions are: What purpose does a CoC serve? How does it help?
In most, if not all cases, they are not very helpful beyond, indeed, politics and drama because they turn into a weapon (on the assumption that this wasn't the aim all along).
When a CoC is pushed by a company I think it's probably because companies need company policies for employees for HR and legal reasons. I am not convinced they could explain why that must extend beyond that, and really this is an ass-covering exercise in case things get too "raucous" in the project and produces bad PR.
> What purpose does a CoC serve? How does it help?
At work I think it has helped since we now have a clear procedure to follow if there's a problem.
Having the Code of Conduct be a public document rather than only an internal one means there are fewer excuses available for not following it — for example someone can't claim they thought it would be OK to ask a developer for sex since we wrote down that that's not OK.
It's also easier to explain in case the problem individual is employed by a paying customer.
>someone can't claim they thought it would be OK to ask a developer for sex since we wrote down that that's not OK.
The person who thought this would be okay will not be stopped by a Code of Conduct.
The people who know it's not okay don't need a Code of Conduct.
The whole thing is useless. This is the same retarded shit as people stating their pronouns, trying to engineer every social human relation.
Don't be a jerk with people. If someone looks a certain way, it would be kind to assume they want to be treated in a certain way, etc, etc. You don't need terms of conduct and similar bullshit for that.
> > "If you are more annoying to work with than your contributions justify, you'll be ejected."
I'd argue that this is unreasonable for any project large enough to need a CoC at all. What it amounts to is excusing misconduct by project insiders, while allowing them to drive off new users by declaring their behavior 'annoying'. This isn't a code of conduct at all; it's a formalized "old boys club" dynamic.
You are correct, thanks for the explanation.
An OSS project has de facto owners and an established group of contributors. If they want to "drive off" new users that's their prerogative.
In fact, a CoC is no more than them laying down the law explicitly on how they want people to behave in the project.
"If you are more annoying to work with than your contributions justify, you'll be ejected" is spot-on on how things work in general.
That might lead the project to fail or fork (beauty of OSS) but that's life.
It also establishes that the project is a meritocracy, which is not necessarily something that every project wants. (And from a personal point of view, many meritocracies seem pretty toxic to me.)
> it's a formalized "old boys club" dynamic
There's a risk that this happens, sure, but it can be more charitably read as allowing/retaining people like Steve Jobs or Linus Torvalds, who, despite being assholes, are effective assholes and usually spew vitriol only in the interests of the product itself.
Weighing annoyance against contributions is a valid methodology. Your concern is valid, as well, so governance should always be introspective enough to ensure that the metrics of "annoying" and "useful" are objective ones and don't grandfather in the insiders.
A man suffused by his own myth. I stopped reading when he said he codified the rules of open source. Eric, you don't get to define what a DECUS tape is.
People who run projects get to decide if they have a CoC or not, Raymond isn't the boss of the bazaar or the cathedral.
> But we should be ruthless and merciless towards people who try to turn "Be kind!" into a weapon. Indulging them never ends well.
Indeed, this is how cancel culture started (not just in OSS), and look where it got us.
"If you are more annoying to work with than your contributions justify, you'll be ejected." should be all that is required, if it really needs to be said at all.
CoC should facilitate free speech and thought without the worry of repurcussions. But in reality it is now a tool to say "I don't like you/your thoughts, so here comes the hammer". I have seen some of the flimsiest excuses used to ban someone. But the same reason doesn't warrant ban on the people the admins like.
If you are actually right, then talk about in the open. Show restraint from name calling. Not everybody is out there with a hidden agenda. Lobste.rs is a great example of this. No shadow banning. Everything has a reason and accountability.
Shoutout to people like JT from the BSD Now podcast as well. I am a fan of his moderation on the Telegram channel. Atleast as far as I have witnessed, I have seen some of the controversial topics going on and on but not banned for difference of opinions. A lot of de-escalation and patience. It would've been so easy for him to just do some Hammering. But that's not how it goes most of the time.
We need more spaces like this.
Completely bogus argument. The core of the issue is that non-technical topics were brought into technical environments, creating heaps of drama, and then the concept of code of conduct was created to try and stir that drama.
It’s basically addressing the symptoms rather than curing the disease: if you want to do foss, you have to accept you might be collaborating with somebody whose ideas are completely antithetical to yours. Any other approach is not free, and is not open.
> Completely bogus argument. The core of the issue is that non-technical topics were brought into technical environments, creating heaps of drama, and then the concept of code of conduct was created to try and stir that drama.
Yes. Looks like I didn't clarify my point in the main post. I don't think CoC is helping anyone. It is better to be without it.
I was talking about how CoC's was being applied at present and how it was first introduced. If you've been around since at least early 2000s with flamewars, then you know why CoC came about. It was not a pretty sight. A lot of communities were like 4chan. Unfiltered and with people getting banned when nobody is backing down from an argument. I wasn't rallying for CoC to be made better. But I understand where it came from at the same time.
> then you know why CoC came about
I've been on the Internet when flamewars were common, and CoC weren't a thing yet.
CoC (along with committees) are a very recent thing. They weren't introduced until certain people started exercising moral blackmail onto projects, people and organizations by pushing political topics into technical discussions.
Back in the days of flamewars there were a few rules and pretty much just moderators to enforce them.
Rules were simple, you were usually banned if your messages were either not civil (and here again, the bar was high), if your contributions were comically off-topic or if you spammed the forum/newsgroup/mailing list.
What free speech do you expect in a free software project? There is a big chance that discussion is totally off topic and anyone managing a community knows that, for you to not be allowed to destroy it, you should be booted.
Free speech as in right to opinion without worrying about being silenced. Dead simple. On merit. It's not off-topic all the time. Right now, moderation is unfair and used to silence people.
And it's not just OSS projects you know? There are OSS related communities like Linux / BSD and many many other projects/communities. CoC unfairness is all over there too.
I don't care about your opinion when I'm reporting or fixing a bug in a open source project, go preach somewhere else.
Yikes, I didn't realize Raymond was the kind of guy that posts about "low IQ black people" and sees "Marxists" everywhere.
I don't like CoCs either, but ironically guys like him make me think they might be needed after all...
People of ESR's ilk are why you can't have a serious project without a CoC these days. It's just tragic. Else before too long you end up losing your black, trans, etc, contributors. Because somehow it's not "drama" when the ilk does racism/misogyny/transphobia but it is when they're being called out on it.
Having a clear CoC helps keep the good actors safe while having clear red lines to keep the bad actors either in line, or out. But it's so sad that we got to that point at all...
He's been this way for a long time, unfortunately. I remember reading a blog post of his in the '00s that was openly racist. Come to think of it, that may have been the last thing of his I read.
It feels to me like a lot of people who don't like codes of conduct have similar views.