Behavioral standards in the free software community [LWN.net]

The GNOME community has recently started a discussion on the adoption of a code of conduct for community members. While a number of people clearly think that such a code makes sense, others are just as clearly uncomfortable with the idea. The free software community is traditionally an open and unregulated group. Its members are concerned with quality of contributions and inclusiveness; there is relatively little interest in conduct rules, and an active dislike for self-appointed enforcers and attempts to exclude potential contributors. So the number of projects with written behavioral codes is relatively small.

Such codes do exist, however, whether or not they are written down. Anybody who doubts this fact may want to ponder on the likely fate of a developer who attempts to contribute plagiarized code. But other standards clearly exist as well. Consider, for example, this case: a Debian developer was not only asked to leave DebConf last month, but was removed from the project altogether. A weblog entry from a nearby participant reads:

The difference in values between Ted and the rest of our project was just too immense. When I was walking out of the room at around 7 in the morning next day my final sentence was "Ted, even if you spend rest of the Debconf apologizing and making friends, I do not see a future for you in this project." and the most important was that Ted and John seemed to agree with me on that

Only two months earlier, Debian went through a protracted debate on whether another developer should be forcibly expelled from the project. In both cases, the issue was not one of plagiarism or other crime; instead, these people are being pushed out for being jerks - for somebody's value of "jerk." Their behavior is said to be so unpleasant, and so off-putting for other members of the project, that their presence is no longer welcome. This is the sort of behavior that the proposed GNOME code of conduct seeks to regulate as well. This proposal contains items like "be respectful and considerate" and "don't be racist." Its supporters are trying to maintain a GNOME community which is pleasant to work in, and which does not drive potential contributors away.

They have a point: it has been noted, for example, that female participation in free software projects is often close to zero. That is, as some have observed, below the usual percentage of women in the general population; but it is also well below the percentage of women found working in technical fields. There is a whole population of potential contributors out there who have chosen not to be a part of the free software community. One very possible reason for their absence is the sort of behavior encountered on mailing lists, at conferences, and in other places where the community gathers. Perhaps, if standards of behavior were higher, more people would choose to participate.

(Then again, the problem could be elsewhere: Richard Stallman chimed in with a claim that the use of the term "open source" may be the real reason why women chose not to participate. This particular line of reasoning has not attracted a large following, however).

Alan Cox points out that the issue is a little broader:

I'd be wary of pursuing just the "women in GNOME" issue, because many of the same things put off far more than just women. Running around shouting "pants off" is not, for example, very compatible with the Japanese cultural expectations.

One can, without great difficulty, make an argument that, as the free software community "grows up" and tries to expand beyond its "western white male geek" stereotype, it should look harder at how its members behave. If one contributor is sufficiently unpleasant to repel the participation of numerous others, then perhaps the community truly is better off without that person. So maybe the community truly does need to be prepared to expel people who are too difficult to be around. Codes of conduct might just make sense.

But consider an episode from just over three years ago, when a prominent developer (let's call him "X" for the moment) was stripped of his commit privileges and kicked out of an important project. One of the people involved in this action justified it with these words:

What X has done is among the most low-class, unprofessional, and tactless things I have ever experienced in my professional career.... Bottom line, in my opinion, is that what X did is unacceptable on its face and he deserves to be held accountable for it. So he's out.

This looks like a clear application of a code of conduct; somebody behaves badly, and is booted from the project. Nothing to complain about. Except that X, in this case, was Keith Packard, who was busily trying to reform the XFree86 project. That project's decision to exclude Keith turned out to be fatal; XFree86 still exists - it even put out a release in May - but nobody cares anymore.

This episode highlights the dangers of behavioral codes. They can be used as a way of silencing people who have something inconvenient to say, but sometimes those people need to be heard. Codes of conduct can evolve into a sort of stifling "political correctness" where people become afraid to express their thoughts. The creation of such an environment will suck the life out of a project more quickly than any number of unpleasant people.

The community as a whole may well want to think about how people interact, and how that interaction can be made more pleasant and more globally inclusive. Behavior which is rude, sexist, racist, or worse runs counter to our values (one hopes), and it makes us weaker. So discussions of how we wish to treat each other and how we can avoid pushing away people who could make our community richer are worth having. But we must work toward that goal without silencing our more outspoken members; sometimes they are saying something we should hear, even if it makes us uncomfortable.

(Log in to post comments)

Case study: GCC

Posted Jun 8, 2006 6:05 UTC (Thu) by JoeBuck (subscriber, #2330) [Link]

The GCC project doesn't have any hard rules about politeness, it's just about social norms. People do occasionally get angry or rude, but you just don't see that much of it. When things get hot, people often contact each other off-list and smooth things out. Any development list has a norm; people read it for a while and they get a clue as to how people interact on that list, not just what's on topic, but the general tone.

In its entire history since the EGCS fork, only one person was ever expelled from the gcc mailing lists. The jerk in question flooded the list with his rants and made several threats of violence, both on-list and off-list, against the release manager at the time, and the steering committee discussed whether we'd have to get the police involved, but fortunately the guy just went away after he was banned.

With EGCS, we tried to set up a culture that was based on consensus. If you can work out a solution that everyone can live with, that beats a solution that wins by a 2/3 vote every time, because all of the serious objections have been answered. Of course, that isn't always possible; then if it's a technical issue, you want the best expert in that area that you have to decide, and if it's "political", you want to come as close to unanimous as you can.

Case study: Ubuntu Code of Conduct

Posted Jun 9, 2006 0:10 UTC (Fri) by nealmcb (subscriber, #20740) [Link]

Ubuntu is another good case study - there is a code of conduct, which people need to sign (PGP) before becoming a full member.

http://www.ubuntu.com/community/conduct

The outline is:

Be considerate
Be respectful
Be collaborative
When you disagree, consult others
When you are unsure, ask for help
Step down considerately

I've seen some good things said about it, and liked the idea myself.

It appears that Gentoo has one also, based on Ubuntu's

Behavioral standards in the free software community

Posted Jun 8, 2006 3:23 UTC (Thu) by dang (guest, #310) [Link]

Some other thoughts:

One huge key really involves the ability of people to work together and to *want* to help people. I can think of some prominent developers in important projects who gleefully rip people and code in pretty salty terms, but who clearly rip their goofs just as hard, clearly bend over backward to help people, and who do enough heavy lifing in general so that everyone knows that they can bark and sometimes even bite, but they are treasured anyway. Part of the read is "what are they trying to say when they flame?"

On the flip side, a steady stream of even small flames can just wear you down after a while. It is sad when prominent developers just drop off project because life is too short to slog through the crap.

Even for a project where contributers are generally polite and discuss the merits of code not coders, things really can heat up when people are talking past one another or when people are advocating mutually exclusive solutions. In both cases, it helps a lot if someone mediates or makes the call in a way that respects all contributors. Managing frustration levels wins as much as does basic decency.

Behavioral standards in the free software community

Posted Jun 8, 2006 5:11 UTC (Thu) by roelofs (guest, #2599) [Link]

One huge key really involves the ability of people to work together

I was just thinking that another prominent example (besides the GCC case) involved Linus + various folks (e.g., I believe one of the IDE subsystem developers two or three years ago). It's not necessarily a matter of conduct in the sense of rudeness (although that may well be/have been part of it), and neither is it necessarily a matter of getting booted off LKML. But it is about conduct in some broader sense, and insofar as it's direct access to Linus, it's the next best thing to LKML exile...

Greg

Behavioral standards in the free software community

Posted Jun 8, 2006 5:43 UTC (Thu) by k8to (subscriber, #15413) [Link]

This is kind of off-topic, my quite subjective view is that problem which
occurred with Andre Hedrick, who you refer to indirectly, is that he just
couldn't communicate very well with the other LKML developers. It wasn't
really a matter of hostility, just that it was very very difficult to
pass information in or out of Andre. My perception was that he wasn't
really in very strong command of certain issues that ended up being his
achilles heel, but I really couldn't say because every conversation I had
with him ended up with us both agreeing to give up trying to pass
information back and forth.

So there was _a_ problem there, but I don't think it was communication
style, not conduct.

Andre, if you read this and disagree, please do correct me. This is
_very_ subjective.

Behavioral standards in the free software community

Posted Jun 8, 2006 5:00 UTC (Thu) by cventers (subscriber, #31465) [Link]

Sometimes I wonder to what extent the Linux kernel model works only for the Linux kernel. I really assume it should be applicable in many other places (though I think it's clear you need a leader like Linus to make it happen).

I happen to like this observation in Documentation/ManagementStyle:

Most people are idiots, and being a manager means you'll have to deal with it, and perhaps more importantly, that _they_ have to deal with _you_.

It turns out that while it's easy to undo technical mistakes, it's not as easy to undo personality disorders. You just have to live with theirs - and yours.

However, in order to prepare yourself as a kernel manager, it's best to remember not to burn any bridges, bomb any innocent villagers, or alienate too many kernel developers. It turns out that alienating people is fairly easy, and un-alienating them is hard. Thus "alienating" immediately falls under the heading of "not reversible", and becomes a no-no according to Chapter 1.

There's just a few simple rules here:
(1) don't call people d*ckheads (at least not in public)
(2) learn how to apologize when you forgot rule (1)

The problem with #1 is that it's very easy to do, since you can say "you're a d*ckhead" in millions of different ways (*), sometimes without even realizing it, and almost always with a white-hot conviction that you are right.

And the more convinced you are that you are right (and let's face it, you can call just about _anybody_ a d*ckhead, and you often _will_ be right), the harder it ends up being to apologize afterwards.

To solve this problem, you really only have two options:
- get really good at apologies - spread the "love" out so evenly that nobody really ends up feeling like they get unfairly targeted. Make it inventive enough, and they might even be amused.

The option of being unfailingly polite really doesn't exist. Nobody will trust somebody who is so clearly hiding his true character.

In a community like LKML, it seems that people being mean are either tolerated because of their history and value, or they do a very effective job in isolating themselves and reducing their own value and trust from the community.

Hence I think behavioral codes aren't really necessary. I'm also not a woman, and can't speak for any, but I don't think the lack of a behavioral code has much to do with it.

I watch the mailing list of several projects, and it seems to me that while people like to be passionate from time to time, these issues seem to regulate themselves best when left nicely alone.

Behavioral standards in the free software community

Posted Jun 8, 2006 7:03 UTC (Thu) by lysse (guest, #3190) [Link]

The XFree86 project's decision to expel Keith Packard might not have been terribly good for their continued relevance, but it could be argued that it has turned out to be the right decision for all concerned - XFree86 have been allowed to continue up their own cul-de-sac unimpeded, and Keith's work has been allowed to make uninhibited progress. Moreover, without actually knowing anything about what happened, it seems to me that the view that the XFree86 team took of Keith's actions was not a surprising one; revolutions don't tend to start from a premise of civil obedience and submission to authority.

Even if the debate is framed in terms of personal conduct ("being a jerk"), there's the example of the notoriously abrasive Theo de Raadt to consider. NetBSD has continued rather well without him - but he has also turned his efforts to a project which has achieved considerable success and earned its place in the world in its own right. Being a jerk and having a valuable contribution to make are clearly not antithetical.

So perhaps the whole idea of a "code of conduct" shouldn't be phrased in absolutist terms ("behave like this or you're BAD!"), but in terms that reflect the reality of group work - which is that if one individual's views, orientation, or personal conduct threatens the survival of the group, the best solution for all concerned is for that group and that individual to part company, without issuing judgements on either side. Of course, personal opinions will abound, but they will, of necessity, be biased; it's perhaps unfair to give them any more status or weight than "we've reached a point where there's no way we can work together any more". Let disinterested observers make up their own minds as to who was in the right, and why.

Behavioral standards in the free software community

Posted Jun 8, 2006 8:12 UTC (Thu) by ekj (guest, #1524) [Link]

You have a choise.

Either you make your rules very general, very much subject to interpretation, very little concrete. ('Dont be a jerk') In which case everyone will agree, but on the other hand nobody agrees where exactly the border of jerkiness lies.

Or you make your rules concrete, precisely defined. In which case nobody agrees with the what exactly the rules should be.

I don't see a benefit in either. I also don't see any evidence that the way we do things currently needs changing. Essentially policing of a set of undefined, unwritten, unspecified, everchanging codes of conduct by the community itself.

It works like this: If you're a jerk, people don't like dealing with you. If people don't like dealing with you, they will avoid dealing with you as much as possible. Which again tends to work as a pretty efficient exclusion-mechanism.

Behavioral standards in the free software community

Posted Jun 8, 2006 12:26 UTC (Thu) by liljencrantz (guest, #28458) [Link]

I strongly disagree. I think there are rules of conduct that are reasonably clear and concrete that would help a lot in these things.

The most important one would be this:

* Praise people in public, flame them privately.

If you subject someone to public flaming, then he will feel much more assaulted and will be much more likely to hold a grudge or even strike back in an irrational way. There are of course exceptions, mostly if repeated private mails do not yield any result.

Other rules that I think are reasonably clear and easy to follow include:

* Try not to treat people differently based on on name, gender or ethnicity.
* Don't repeat yourself. Specifically, don't spam a list with rehashes of your previous messages.

Behavioral standards in the free software community

Posted Jun 8, 2006 13:33 UTC (Thu) by csamuel (✭ supporter ✭, #2624) [Link]

I've always liked "Focus on the issue, not the person".

Chris

Behavioral standards in the free software community

Posted Jun 8, 2006 15:14 UTC (Thu) by wilck (subscriber, #29844) [Link]

Try not to treat people differently based on on name, gender or ethnicity.

Another thing that comes to mind is language. For non-native English speakers, it's more difficult to get heard (and easier to offend) than for "natives".

It's hard, but I wish we all would try not to discriminate by English language abilities.

Behavioral standards in the free software community

Posted Jun 8, 2006 16:17 UTC (Thu) by liljencrantz (guest, #28458) [Link]

Oh, right.

That's a pretty hard issue. I've seen a lot of witty, subtle and ironic responses to messages from people who are obviously struggling with their english. The result is generally that the true intent of the response is completetly misunderstood.

But on the other hand a code of conduct that says "don't try to be funny" won't really help in building a community.

A related issue is that a lot of the time, attacks are disguised as irony. On mailing lists you'll see all kinds of 'Ha ha, just seriously' statements.

Behavioral standards in the free software community

Posted Jun 8, 2006 12:40 UTC (Thu) by liljencrantz (guest, #28458) [Link]

Laws often state that everyone is equal before the eyes of the law; they rarely state that you have to treat everyone the same.

Laws often state that you aren't allowed to abuse another person; they rarely state that you have to be polite to people.

The law specifies a bare minimum of what is acceptable brehaviour from a person, which is as I think it should be. I would not like to live in a country where it was against the law to have an argument with someone, or that I had to treat me son the same way as I treat the man on the street.

Behavioral standards in the free software community vs law

Posted Jun 8, 2006 21:24 UTC (Thu) by giraffedata (subscriber, #1954) [Link]

I don't know what countries you're thinking of, but I know in the US, laws don't come anywhere close to the rules being talked about here.

It is not a crime to defame someone. Defaming someone can make one legally liable for the damage it causes, but even then it has to have a clear negative impact on the person's business or profession; the insults on OSS mailing lists probably wouldn't qualify.

It is not a crime to reject someone's code because of his race or gender. There are about half a dozen ways you cannot discriminate based on race (employment, housing, etc.), but merging code isn't one of them.

It is sometimes a crime to do indecent things such as use offensive words, but since it's always based on community standards, I don't think it would apply to OSS mailing lists, especially if said list does not have a code of conduct.

If there are other jurisdictions in the world where common decency is codified into police law, that might be a good start as opposed to generating a code of conduct for an OSS project from scratch; someone already suggested approximately the same thing by mentioning the ACM's code.

Behavioral standards and power

Posted Jun 8, 2006 15:41 UTC (Thu) by wilck (subscriber, #29844) [Link]

There is a relation between behavior and "social power" in the community. In all social systems, people at the bottom of the hierarchy are more likely to misbehave than people at the top. Reasons for this are (among others)

lack of education (in the community terms: lack of netiquette experience),
the feeling not to be heard unless you shout,
the fact that an action by a well-respected person is less likely to be regarded as misbehavior than the same action by a "nobody".

The free software community has a pretty strong social divide, although the criteria by which people are discriminated are different than in the normal society.

I think that the feeling of being oppressed by someone ranked higher is a frequent reason for misbehavior. From this viewpoint, the attempt to establish general Rules of Conduct can be interpreted as a potential repressive measure against those on the bottom side of the hierarchy. Care should be taken that this impression is avoided.

A bit out of context, I'd like to add that killfiles and filters already offer effective measures against cases of strong misconduct. The question must be asked why these aren't considered good enough any more.

Behavioral standards and power

Posted Jun 8, 2006 18:09 UTC (Thu) by ajcpi (guest, #1122) [Link]

Could it be that people who "misbehave" are less likely to acheive social power

Behavioral standards and power

Posted Jun 8, 2006 18:56 UTC (Thu) by wilck (subscriber, #29844) [Link]

Certainly. This is a self-enforcing process.

OTOH, frequent violation of explicit or implicit rules of conduct is not prohibitive for getting "powerful".

The community is no different from real life in both respects.

Behavioral standards and power

Posted Jun 9, 2006 4:01 UTC (Fri) by dang (guest, #310) [Link]

There is a flip side here, too, I think. You see a lot of flamage coming from people on top when you get the following mix:

- the person on top is a heavy lifter and plays a gatekeeper role
- the flamee is a newbie who hasn't done her or his homework
- about basic design or style
- about how work flows

You get frustration on both sides here. The gatekeeper wants to help but is frustrated because the submitter didn't do homework which makes it difficult to help and also eats up precious time ( because skilled people willing to do grunt work are in hot demand in open source projects ), and the submitter doesn't know initially what the just got themselves in to.

Typically everything shakes out quickly and you see a good working relationship, but it can be rocky at first. Which ain't none of it about a code of conduct, really, but about setting new contributors expectations well before they submit...and maybe about overloaded gatekeepers.

Behavioral standards in the free software community

Posted Jun 9, 2006 5:16 UTC (Fri) by paulmfoster (subscriber, #17313) [Link]

While I run several lists and am a heavy proponent of civilized behavior, I caution against giving in to "political correctness". We've created a world full of people whose job is to be aggrieved over extremely slight offenses, or none at all. The danger of codes of conduct like this is that they may give in too far to the easily aggrieved. It's impolite to make sexist jokes on a list populated by women, but the most they should be owed if someone told a sexist joke is an apology. No more. Continued, constant crude behavior might eventually rate an expulsion.

Unfortunately, hackers skew a bit in the direction of the juvenile and dorky. Goes with the territory. I have some of these guys on my lists. Every once in a while I have to snap at them to get their attention. They get back in line. I rarely find it necessary to expel one, and only do so when it's clear to one and all that they guy just can't live comfortably with other people.

Look at it this way: If we all adopted what might be called "reasonable" codes of conduct, RMS would have been kicked out of this movement long ago.

Behavioral standards in the free software community

Posted Jun 16, 2006 22:44 UTC (Fri) by kfogel (guest, #20531) [Link]

It's interesting that the community feels the need to spell out the
code of conduct (or at least some in the community do). My feeling
has always been that, since anyone who's kicked out can take a
copy of the code with them, all that's really happening is that one
group of people is saying they will no longer associate with another
group (perhaps the latter group numbers only 1, but the theory
is, er, fully general). In that circumstance, the question of what fairness
requires becomes more complex. The kicked-out person has not
been deprived of anything except the company of people who, by
definition, don't want his company. Of course, in reality he's lost
something very tangible -- the ability to influence the other developers
directly -- but he lost that because they decided they didn't want his
influence.

Which makes the Keith Packard / XFree86 example interesting: what
happened there was exactly what should have happened, really.
Packard proposed working in a way the others didn't like. They
said no (i.e., kicked him out), he went on to revitalize X.org and now
things there run the way he originally proposed. It's as true that he
kicked out the rest of XFree86 as vice versa, although of course no
one ever puts it that way :-).

So I guess I have to admit I don't see the need for a code of conduct
when sharing a replicable resource. If people aren't behaving well,
talk to them. If they don't change, no code of conduct is going to
make a difference, you just have to decide whether to remain associated
with them, or they with you.

(Of course, a code given as a guideline can be very useful -- "Avoid
personal insults and you'll get a lot more done around here" is great.
But when someone starts making personal insults, the reason to
kick them out then is not that they violated the code that they had
been previously notified of, it's that they're being insulting!)