Too many lords, not enough stewards
Too many lords, not enough stewards
Posted Feb 5, 2018 10:35 UTC (Mon) by farnz (subscriber, #17727)In reply to: Too many lords, not enough stewards by neilbrown
Parent article: Too many lords, not enough stewards
The issue with Crocker's rule, taken to extremes, is that it implies that it's OK to be as deliberately offensive as you like, because the negative reactions to deliberate offensiveness as the responsibility of the person offended. For example, if I insist on calling someone a monkey, not a person, knowing that they're (a) Black American, and (b) have experienced such comparisons being used to justify treating them personally as less than human, Crocker's rule says that it's their fault they got upset.
It's a good guideline for an individual to start from - not least because, taken seriously, you're always able to explain what about a statement that has upset you has triggered the upset, and thus explain it so that people can avoid giving offence later, but it's not a great guide for communication in general.
(Log in to post comments)
Too many lords, not enough stewards
Posted Feb 5, 2018 11:06 UTC (Mon) by neilbrown (subscriber, #359) [Link]
I think that extreme is distorted beyond recognition.
To quote "Declaring yourself to be operating by "Crocker's Rules" means that other people are allowed to optimize their messages for information,.."
So the primary goal is maximizing information. If someone chooses to be deliberately offensive, I fail to see how that can be construed as "maximizing information".
Also, if you choose to use Crocker's rules, it doesn't imply it is OK for "you" to do anything. It is permission for other people to do certain things. If a person feels that the risks exceed the benefits, they would be well advised not declare Crocker's rules. As described in the link, it is strictly opt-in.
Of course, people who are determined to be offensive are unlikely be deterred by someone's decision not to opt-in. So I cannot see how the "deliberately offensive" hypothesis is at all relevant - it is neither encouraged nor prevented.
Too many lords, not enough stewards
Posted Feb 5, 2018 11:30 UTC (Mon) by farnz (subscriber, #17727) [Link]
The issue is that it fails in the face of bad actors; someone who genuinely believes that there's a skin tone that implies that you're not human any more is optimizing to maximise information transfer. It's just that some of the information they're transferring (that they don't believe that you're a person because of your skin tone) isn't directly relevant to the task at hand.
Now, it may also be the case that some of the information they're transferring is offensive - but they've optimized for information transfer above all else, and you, by opting into Crocker's rules, have implied that it's fine for them to assert that you're not a real person, but a sub-human slave, because by doing so, they're maximizing information.
Too many lords, not enough stewards
Posted Feb 5, 2018 17:06 UTC (Mon) by dgm (subscriber, #49227) [Link]
Just remember that opting in to Crocker's Rules doesn't imply that I cannot discard information that I deem unrelated to the task at hand. It's not that different from the Robustness Principle "Be conservative in what you send, be liberal in what you accept".
Too many lords, not enough stewards
Posted Feb 5, 2018 17:55 UTC (Mon) by farnz (subscriber, #17727) [Link]
And it has the same failure cases as the Robustness Principle; where all actors involved are making a good faith effort to make things better, both Crocker's Rules and the Robustness Principle result in miscommunication being quickly resolved with the right outcome. It falls over when there are bad actors involved who don't actually want to make things better, but to make their ideas dominate come what may; if no-one is calling out bad behaviour, then bad actors have no reason to change their actions.
In turn, this leads to the toxic community issue, given enough time - when you see that a community is apparently accepting of racism/sexism/whateverism (because no-one calls it out under Crocker's Rules), bad actors who approve of racism/sexism/whateverism are willing to join, but past victims of racism/sexism/whateverism are put off because they fear a fresh bad experience. If left unchecked, that leads to an environment where only people tolerant of racism/sexism/whateverism take part, because people who just want to discuss technical details are put off by the racism/sexism/whateverism that pops up all the time.
Too many lords, not enough stewards
Posted Feb 5, 2018 22:25 UTC (Mon) by neilbrown (subscriber, #359) [Link]
Mapping them onto the protocol space, I interpret them as saying that it doesn't matter what message you send me, I promise that it won't trigger a buffer overflow or a NULL pointer dereference, or if it does, I won't hold you accountable. I certainly won't launch a DoS attack against you.
I might shut down the communication channel immediately.
I might enable fail2ban and respond to repeated unhelpful messages by ignoring all messages from you for a period of time.
If the information content of your messages makes the parsing overhead worth while, then I'll probably keep listening (maybe after a break while I re-architect my parser).
If, however, the information content of your messages is (by my assessment) harmful to the community - if you refuse to use secure encryption or masquerade as a host that isn't you, or consistently deliver unsolicited bulk email or engage in DDoS attacks - then I don't see that Crocker's rules prevent me from highlighting this fact in the community and suggesting that they are not helpful behaviours.
> if no-one is calling out bad behaviour, then bad actors have no reason to change their actions.
I completely agree with this. Crocker's rules are only half a solution. They allow us to communicate openly without fear of a misunderstanding escalating. They don't tell us what we need to communicate.
If you receive a message which you think it unhelpful or harmful, it is perfectly within Crocker's rules to say so - to call out the bad behaviour. Don't say it in a way to attack anyone - say it in a way to inform people. Maybe your correspondent will learn something and change their ways. Maybe they will persist and you will determine that they are an unreliable source of information, and act accordingly.
I'm imagining a new tag in the MAINTAINERS file which declares:
1/ I accept Crocker's rules on kernel related email lists
2/ I undertake to be polite and respectful in kernel related email. Please tell me if you notice me behaving otherwise.
Purely opt-in of course.
If you are having a problem getting a patch accepted, try sending it to someone who has declared that tag.
Too many lords, not enough stewards
Posted Feb 6, 2018 0:36 UTC (Tue) by excors (subscriber, #95769) [Link]
I think a problem with that approach is that when someone sends a message to you on an email list, you are not the whole audience for that message. Possibly hundreds of other people will read it too, and they haven't all signed up to Crocker's Rules.
E.g. Alice says something to Bob, Eve is on the same list and reads that message and silently agrees with Alice's arguments, but Bob disagrees and calls Alice a moron. Alice doesn't care and shrugs it off, but Eve feels like the insult applies to her too (as she shares Alice's position) and is offended, and will be reluctant to join the discussion in support of Alice, and might become nervous of talking to Bob even in unrelated discussions, and everyone will lose out on Eve's valuable technical input.
Then there's the positive feedback cycle that farnz suggested, where newcomers who are crocks like Alice and Bob will happily join in all the discussions, while non-crocks stay quiet in any discussions that involve at least one crock, and it will only reach an equilibrium once all the non-crocks have been driven out of the community.
Too many lords, not enough stewards
Posted Feb 6, 2018 13:29 UTC (Tue) by farnz (subscriber, #17727) [Link]
Worse; if Bob is a bad actor (rather than a good actor with poor communication skills), Bob and their ilk can ensure that the Crocker's adherents don't get to the point of calling them out, but that Eve and their ilk are made aware that joining in here will result in the same bad outcomes that they've already experienced elsewhere. Things like "dogwhistles" are useful here - chances of you even recognizing them if you're in neither the abusive nor the targeted group are low, but they set expectations.
A better alternative I've seen described as "assume good intent" is to start from the assumption that Bob doesn't mean the bad thing he just said, and to respond by first rephrasing what Bob said into a good description of what you understood him to mean in Crocker's terms, and then to respond to the rephrasing you did.
Something like Bob says "WTF, this code is awful - time of the month?!?", and you respond with "Bob, that's not a good way to express things. I'm assuming you meant 'how did you test this code - I can't see how it could work?'" and continue from there.
If Bob is a good actor, then you've not harmed anyone - you've given Bob a chance to understand that they didn't express themselves well, you've given Alice and Eve a sign that bad behaviour is not considered acceptable here, and you've moved on quickly from the bad phrasing to the technical detail. If Bob is a bad actor, you've either made them realise that bad behaviour is not OK here, or you've pushing them towards being explicitly offensive.
Too many lords, not enough stewards
Posted Feb 6, 2018 21:57 UTC (Tue) by paulj (subscriber, #341) [Link]
Great comment, thanks.