I'm standing up against verbal abuse on LKML.  I will happily stand alone, however you can also support this cause.  Please speak up, either here on Google+ by resharing this post, or commenting on this post with words of support.  If you dare, you can also reply to my lkml email.

http://marc.info/?l=linux-kernel&m=137390362508794&w=2

"Where do I put this fire? This bright red feeling? This Tiger Lily down my mouth? He wants to grow to 20 feet tall... I'm so tired of being shy; I'm not that girl any more. I'm not that straight-A anymore."

Paula Cole - "Tiger" - Live at Pershing Square

#linux #kernelhackers #kernel #abuse #resistance  

Update

Examples of verbally abusive behavior on the Linux kernel mailing list:

http://marc.info/?l=linux-kernel&m=135628421403144&w=2
http://marc.info/?l=linux-acpi&m=136157944603147&w=2

Further Update

I don't have time to moderate comments on this Google+ post, and I don't want to leave personal threats visible in this thread.  Please comment on my blog post instead: http://sarah.thesharps.us/2013/07/15/no-more-verbal-abuse/
[prev in list] [next in list] [prev in thread] [next in thread] List: linux-kernel Subject: Re: [ 00/19] 3.10.1-stable review From: Sarah Sharp Date: 2013-07-15 15:52:48 Message-ID: 20130715155202.GC29526 () xanatos [Download message RAW] On Fri, ...
390
140
Raúl “Nacho” Gómez's profile photoMatthias M.'s profile photoEdwin Knuth's profile photoRobbie Chen's profile photo
32 comments
 
I've heard the argument time and time again that you have to curse or scream at contributors.  I couldn't disagree more.  We get by fine in QEMU without it.

We police qemu-devel pretty strictly for that kind of thing.  The word "fuck" was used 4 times in the past 5 years on the list.  Profanity in general usually only comes in from cross-over discussion with the kernel community.  Personal attacks are unheard of and I wouldn't tolerate it.

We admittedly are only 1/10th of the kernel's size in terms of community but we are growing fast.  We're always looking for new contributors too.
 
Thanks for speaking up like this. You might also be helping out some of us who thought we were the only ones who dislike the hostile atmosphere in some open source projects.
 
I see lots of strong words in LKML. So much that in some other lists there's the say "hey, this isn't LKML and this behavior is not appreciated here".

However in this case I don't see it. Linus and Greg were in no way calling out for violence, but instead making joke about it.

What I am seeing more often than I'd like is people losing the ability to joke at ourselves. For me over reaction is as disturbing as the strong words.  Go to the link you pointed out here, read it again. Knowing the people who wrote them can't you imagine their troll face?
 
I think you're quite sane to call them out on that. It is a mailing list. Miscommunication and misunderstandings happen all the time in such inhuman low bandwidth electronic mediums as not everyone can or will ever read things with the same frame of mind and intent as the authors.

I really appreciate your, "Let's discuss this at Kernel Summit where we can at least yell at each other in person", comment. Face to face is the ideal way to resolve these things so that everyone understands (when that is even possible). Electronically this sort of thing has strong potential to devolve as misunderstandings explode and are jumped on by other people with an axe grinding agenda of their own.

Best of luck! You are quite sane. :)

I'm not responding on lkml as I haven't actively participated there in over 10 years so that would look a little weird... as if I were a random popup axe grinder.
 
I'd support a nicer tone on LKML, so hope you win through with this. Mostly because we see lots of people trying (and failing) to imitate Linus, but just ending up with lots of non-technical ad hominems that derails otherwise sensible discussions.

That said, I think you are in danger of over-stating the problem. In particular when you accuse Linus of "advocating physical intimidation and violence": the email you quoted was clearly a good-natured joke...

I'm also not really sold on the idea of any of this constitutes "verbal abuse", if you take away all the colorful adjectives the underlying messages are usually not problematic.

The tone tends to be a bit harsh though, and I think that's indeed something that could be improved without sacrificing clarity, so good luck! :)
 
Kudos Sarah. What you're doing is courageous and the Right Thing.

I haven't been part of the kernel community for a number of years now, else I'd respond on lkml too.
 
I agree with +Lucas De Marchi here. I really wish you had picked a better example, because I honestly believe that you are in principle correct, but making this instance representative of the problem was, in my opinion, a mistake. Unfortunately, this means that the entire problem will likely be cast as just someone being too PC. I would reccomend you add other, more representative instances to this post before that happens. Abuse is not okay, but this is a bit of a stretch.
 
+Matt Walker and +Lucas De Marchi: Examples added from the thread.  This is not about the "meta" thread, this is about the everyday abuse that is directed from Linus to his top maintainers when they make "stupid mistakes".
Dave Airlie
+
29
30
29
 
Do we have any scientific studies into abuse making the kernel community less or more productive vs other similarly sized community development projects? Changing Linus management methods from ones he considers as working for him might require a bit more evidence that things would improve the quality of the kernel or will allow the kernel development to proceed at the same pace.

e.g. Linus merges 20-30 pulls during the merge window every single day, he has probably 10-15 mins to spend on assessing and merging each pull, now he finds one he doesn't like, he gets angry, now he
a) takes 5-10 mins to calms down, writes a response, makes it polite and professional, taking 15-20 minutes, the maintainer is professionally unhappy, nobody else reads the response.
b) fires off a flame, in under 2 mins, moves on to next pull request knowing that problem likely won't occur again, as all maintainers will read the flame.

The thing is without some evidence that Linus can do (a) and still scale, I'm not sure he'll change his methods.
 
Woooot! Go you +Sarah Sharp You are always awesome, but especially here. You are absolutely right. In fact we were discussing the abusive hostile nature of LKML over dinner the other evening. If I may, it's not one person at fault, it's a general thing. My argument over dinner was that many are put off by it, and in some parts of the world the culture leads to a more severe impact.
 
I re-shared this as well, but over in U-Boot land, we also manage to get by discussing things in a civil tone.
 
+Dave Airlie: that sounds a bit like a false choice to me. I'm pretty sure there could be a way to get both a scalable model and education such that others don't duplicate the mistakes.

I tend to always do a quick search for emails by Linus to see what's going on -- in particular during the merge window since if he seems to be in a bad mood today I might as well wait with pull requests until tomorrow. I'm pretty sure others keep an eye on email in similar fashion.

With a milder tone there's also the increased likelihood that +Jonathan Corbet covers it in LWN, for example.
 
+Olof Johansson Agreed, though I really do like his idea of getting some studies to show Linus. From the response I read, Linus isn't actually being a total asshole about this situation, so I think introducing that sort of evidence might start the wheels grinding -- the first step in changing someone's mind. 
 
+Olof Johansson oh that was just an example of two possibilities, I've no idea how Linus scales, but if being polite was to provide a net negative on the kernel development process throughput then its factor that needs to be considered.

I do think gathering evidence will be tricky, there is no other Linux project or Linus Torvalds, does any other engineering project come close to the scale? So the fact that Linux exists at all is always going to be used as a counterpoint to most of these arguments anyways.
 
Unfortunately, my medication has worn off for the day, so I can't do the research atm. Although I am quite good at physics, I lack social science expertise as well. If someone hasn't beaten me to it, I'll try to take a look for some tomorrow. I want to help, but I'm not exactly LKML material; is there anyone more familiar with the research in this area though?
 
I agree with +Sarah Sharp.  The boorish and abusive language that is accepted as "normal" on LKML reflects badly on all of us who work on Linux.  It does not serve to improve the quality of the kernel in any substantial way.
 
I'm surprised how many people felt the need to contribute to that LKML thread. If this is about the "everyday abuse that is directed from Linus to his top maintainers" as +Sarah Sharp puts it - shouldn't this kind of thing get sorted out between, well - Linus and the "top maintainers"? Or maybe there are way more "top maintainers" than I thought.
 
+Frerich Raabe LKML is the communication method between Linus and his top maintainers. There is no private communication lists and nor should there ever be. That raises the issue then that anyone can get involved in any discussion, and that is also a good thing, so really anyone can contribute is a good thing.
 
I'm not impressed by LKML, I certainly won't be doing any kernel development until a code of conduct is installed into Linus. I believe the corruption comes from the top, he's not a teenager despite acting like one and needs to develop some other parts of that brain of his.
 
+Kyle Ladd
I don't hate Linus. I just think he's a dick to his top maintainers. As I mentioned, the standard you walk past is the standard you accept. I can't accept unprofessional, verbally abusive language on the mailing lists.

But you know, if you have a lead on where I can find special brownies in Edinburgh, let me know...
 
+Dave Airlie It's not that hard to collect the data.  Here's a spread sheet for 1H13 comparing Linux to QEMU.  The top kernel contributors are extremely productive.  Almost mind blowingly so.

Of course, when you look at the first 20 or so names, there isn't an immediate correlation between "yells a lot at people" and productivity.

http://www.codemonkey.ws/files/qemu-linux-1H13.ods
 
+Sarah Sharp I dont know know, I am somewhat conflicted on this one: "Let's discuss this at Kernel Summit where we can at least yell at each other in person. Yeah, just try yelling at me about this. I'll roar right back, louder ..." is not exactly deescalating -- also not cool. I never had the pleasure to discuss things in person with +Linus Torvalds, but I can very well imagine that while being stubborn and direct, he is not the one going for a shouting match (that is: moving from verbal to physical).

I also never met +Greg Kroah-Hartman in person, but from what I gather reading elsewhere, despite being a "giant" -- he is so in a kind Sulley of Monster, Inc. way -- And Linus and Ingo are teasing him for it as a wellknown friend (and they are likely very self aware that there is at least partly some envy as motivation too). As such, this seems not the best starting point to me -- unless GregKH is offended by this.

A better starting point would be when some innocent newcomer ends up as roadkill by a Linus rant. Im not a regular reader of LKML, but all the rants I saw also had wellfunded arguments.

Does it have to be that way? Surely not. Would I prefer my communication channels to stay clean of such strong language? Sure.

But in a perverted way, such rants by now provide an additional convenient function: A rant easily get widely published and prevents such some bikeshed discussions to be done again and again. Of course, this can be achieved in different ways -- with fewer collateral damage to the newcomers in the project. But change is really expensive. So if communication culture on LKML shall change, there is a lot of work to be invested there. Linus does not seem to be interested in that(*) -- so it needs to be done by others on that list. And no, Im not volunteering -- I have a small OSS project of about the same size as the kernel to hatch for right now.

(*) And btw, so it seems to be up to now for a wide majority of other contributors. Otherwise Linus would not be the canonical maintainer of Linux anymore, despite a wide set of branches/forks being available.
 
+Bjoern Michaelsen You obviously missed the last two links in my post. This isn't about the thread I replied to. This is about the larger issue that people are afraid to stand up to Linus about. He's the kernel release manager and forks of Linux have not survived (save for Android). However, that does not mean people like working with Linus.
 
I think that both Greg and Linus are behaving as antisocial creatures and the Linux Foundation is not doing anything to fix it.

Have a look at the talk of Jim Zemlin, http://www.linux.com/news/featured-blogs/185-jennifer-cloer/715856-jim-zemlin-at-tedx-what-weve-learned-from-linus-torvalds who explicitly says that their attitide for "Don't be nice" is something that works in the kernel development.

What Jim Zemlin fails to understand is that the Linux kernel is an extremely important piece of software, so anyone who is dissaffected will sadly go away and many more will come to replace them. Other free software projects are not so fortunate, so if those core developers are being dicks, then their project dies.

You should take the issue with Jim Zemlin, who is their boss. Jim is letting them act like that and he confessed to it at a recent talk. And the Linux Foundation is following the negative tradition of those old companies with a few star employees, that they also would put up with bad behaviour.
 
+Simos Xenitellis You watched that talk and don't understand Jim has no control over Linus, I think he explicitly says that. Linus is not a regular employee like you might think, you seem to be stuck in the negative tradition of thinking boss means must control everyone.
 
Good for you for speaking out. You are right, and very brave. I hope that your points at least give some who are reading them pause to consider their audience and intent before posting. 
 
i am understand with your stand Sarah, but for sure Linus is difficult to change his behavior (though not impossible) since he even showed his middle finger for the nvidia issue. 
 
+John Hass, I work as an embedded software engineer at a Fortune 500 company. We might curse, but we don't curse at people. 
 
This doesn't compile Sarah #.

I prefer the way to do things of Linus. Just look at the pyramids, the lash do miracles.