TAB non-nomination
Hi All,
Several people have asked me to stand again for election to the TAB, so I thought I'd give a general explanation of why that isn't going to happen. For background: I was one of the people who lead the charge in getting the Linux Foundation predecessor OSDL to create the TAB to give developer input into what was then seen as a body trying to speak for the Linux Kernel. I was actually TAB chair for 8 years from 2006 to 2014. The job of the TAB, as I saw it, was to solve a lot of the political friction issues around the places where the Linux Kernel community interfaces with the Linux Foundation and Industry and in those 8 years I gained quite a lot of expertise in political relations trying to do that.
However, TAB member and TAB chair aren't roles people are born to fill, they're roles people have to grow into. What our community needs is more people willing to grow into these roles to ensure effective succession and if I stepped into one I'd be denying others that opportunity, which would be bad both for succession planning and the growth of the community in general.
I think the reason I'm getting these requests is angst over this CoC debate, so I'll go so far as to detail my political instincts over this below ... if you've no interest in politics (as most of you won't have) stop reading now. If you are interested, perhaps you should consider standing for the TAB yourself.
James
---
The biggest political mistake was actually doing anything with the Linux Kernel CoC at all. The object was to deflect a highly unfavourable article in the New Yorker. With hind sight, that could be achieved simply by Linus' personal apology, statement that he was stepping aside and going for assistance to understand others' emotions.
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today. Panic is very infectious so it can be extremely difficult in these circumstances to stand up and say "stop, we need more information" ... and if you think you'd be the one always to demand more information remember that there's a time a decision has to be made and it always passes before you can get complete information, so you'd basically be rendering the TAB indecisive and useless. Recognising when it's time to stop and ask for more data and when you have to make decisions with what you have is a key political skill.
The second mistake was picking the wrong CoC. I'm not talking about the wording, which has been discussed on this list, but the politics surrounding the choice: The original author of the current CoC was unsupportive to the point of attacking the kernel community in public. That drove a huge amount of me too attacks plus an equally large amount of anti-me too hysteria and lead to enormous external awareness and friction plus a not inconsiderable amount of unwelcome personal email to various people. This could largely have been avoided by either evolving our existing CoC through a community process or by picking a CoC whose original author would be willing to stand up and be supportive of our desire to change.
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input which has generated a lot of obvious anger within our community itself. All I'll say on this is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...
Quoted 21 lines · On Thu, 08 Nov 2018 16:04:02 -0800 James Bottomley…
Hi All,
Several people have asked me to stand again for election to the TAB, so I thought I'd give a general explanation of why that isn't going to happen. For background: I was one of the people who lead the charge in getting the Linux Foundation predecessor OSDL to create the TAB to give developer input into what was then seen as a body trying to speak for the Linux Kernel. I was actually TAB chair for 8 years from 2006 to 2014. The job of the TAB, as I saw it, was to solve a lot of the political friction issues around the places where the Linux Kernel community interfaces with the Linux Foundation and Industry and in those 8 years I gained quite a lot of expertise in political relations trying to do that.
However, TAB member and TAB chair aren't roles people are born to fill, they're roles people have to grow into. What our community needs is more people willing to grow into these roles to ensure effective succession and if I stepped into one I'd be denying others that opportunity, which would be bad both for succession planning and the growth of the community in general.
Quoted 5 lines
I think the reason I'm getting these requests is angst over this CoC debate, so I'll go so far as to detail my political instincts over this below ... if you've no interest in politics (as most of you won't have) stop reading now. If you are interested, perhaps you should consider standing for the TAB yourself.
If I understand this correctly, this is your way of telling those that asked you to run for TAB that you are not doing so.
Quoted 5 lines
James
---
Quoted 7 lines
The biggest political mistake was actually doing anything with the Linux Kernel CoC at all. The object was to deflect a highly unfavourable article in the New Yorker. With hind sight, that could be achieved simply by Linus' personal apology, statement that he was stepping aside and going for assistance to understand others' emotions.
Hind sight, though is always perfect. At the time, as a TAB member,
Hind sight is actually far from perfect, because we really don't know what would have happened if we did things differently.
-- Steve
Quoted 35 lines
all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today. Panic is very infectious so it can be extremely difficult in these circumstances to stand up and say "stop, we need more information" ... and if you think you'd be the one always to demand more information remember that there's a time a decision has to be made and it always passes before you can get complete information, so you'd basically be rendering the TAB indecisive and useless. Recognising when it's time to stop and ask for more data and when you have to make decisions with what you have is a key political skill.
The second mistake was picking the wrong CoC. I'm not talking about the wording, which has been discussed on this list, but the politics surrounding the choice: The original author of the current CoC was unsupportive to the point of attacking the kernel community in public. That drove a huge amount of me too attacks plus an equally large amount of anti-me too hysteria and lead to enormous external awareness and friction plus a not inconsiderable amount of unwelcome personal email to various people. This could largely have been avoided by either evolving our existing CoC through a community process or by picking a CoC whose original author would be willing to stand up and be supportive of our desire to change.
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input which has generated a lot of obvious anger within our community itself. All I'll say on this is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...
_______________________________________________
Tech-board-discuss mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/tech-board-discuss
Quoted 4 lines · James Bottomley
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today.
I think panic is the wrong word to attach to Linus' response, especially around the code of conduct.
Quoted 5 lines
The second mistake was picking the wrong CoC. [ ... ]
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
The update was entirely based on community input.
Quoted 2 lines
which has generated a
lot of obvious anger within our community itself.
It's absolutely true that some members of the community were upset.
We'll never know if there could have been a better time to make code of conduct changes. There are a wide range of deeply held beliefs in this area, and every choice would have eventually led to major disagreements.
But what we do know is that everyone sat down and did their best to
find compromise. That doesn't mean we found the right compromise for every developer, but I still really appreciate how much time and energy everyone spent explaining their point of view and looking for common ground.
Quoted 6 lines
All I'll say on this
is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...
It's really important the TAB is full of people that care about the kernel. Anger about the code of conduct isn't a great qualifier, but I'll happily encourage anyone who cares deeply about the kernel community, even if they disagree with my opinions about how to best support it.
-chris
Quoted 15 lines · Chris Mason
Quoted 4 lines · James Bottomley
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today.
I think panic is the wrong word to attach to Linus' response, especially around the code of conduct.
Quoted 5 lines
The second mistake was picking the wrong CoC. [ ... ]
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
It is unfortunate it had to start that way. I also understand at times it might be necessary to do so based on my experience with the Linux Kernel Community Enforcement Statement process. What should TAB do as a body if it needs to take action without an option to initiate an open discussion?
I am of the opinion that that was the situation the TAB was in a couple of months ago. Whether that the right choice or not, it is important to continue the conversation and come to an understanding that this is a problem that needs to ab addressed. I think we are on the path to doing that.
Quoted 7 lines
The update was entirely based on community input.
Quoted 2 lines
which has generated a
lot of obvious anger within our community itself.It's absolutely true that some members of the community were upset.
Quoted 9 lines
We'll never know if there could have been a better time to make code of conduct changes. There are a wide range of deeply held beliefs in this area, and every choice would have eventually led to major disagreements.
But what we do know is that everyone sat down and did their best to
find compromise. That doesn't mean we found the right compromise for every developer, but I still really appreciate how much time and energy everyone spent explaining their point of view and looking for common ground.
The positive take away for me in what transpired these past couple of months is just that. The community came together to discuss and express their point of view to move the nudge process forward and speak out to say "we didn't like the way it was handled".
I don't see this as an end and we have to continue to evolve CoC as a living document.
Quoted 7 lines
Quoted 6 lines
All I'll say on this
is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...
Quoted 6 lines
It's really important the TAB is full of people that care about the kernel. Anger about the code of conduct isn't a great qualifier, but I'll happily encourage anyone who cares deeply about the kernel community, even if they disagree with my opinions about how to best support it.
Right. This is the reason why I am standing for TAB and I want to be part of the solution and not part of the problem. I am looking for constructive ways to keep our community viable for years to come.
thanks,
-- Shuah
Theodore Y. Ts'o
maintainerQuoted 7 lines · Shuah Khan
Quoted 2 lines
Quoted 2 lines
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
It is unfortunate it had to start that way. I also understand at times it might be necessary to do so based on my experience with the Linux Kernel Community Enforcement Statement process. What should TAB do as a body if it needs to take action without an option to initiate an open discussion?
As Chris mentioned, there was a large amount of community input on the update. It just didn't happen in an open fashion. One of the challenges with e-mail discussion is that it can end up get dominated by a small number of people who send a large number of messages. In an in-person meeting, a good moderator can say, "Alexis, you've been talking a lot; perhaps we should hear from some other people who have been quiet. Drew, what do you think?" It's a lot harder to do this on a mailing list.
The second challenge is that we were getting trolled by people who were *not* members of the kernel development community. I was able to track down one such troll to their social media presence on gab.ai. (Yeah, that same lovely "free speech absolutists' site, for when Twitter considers you too toxic" which got deplatformed after the shooting at the Tree of Life Synagogue in Pittsburgh.)
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs. So there *was* community input, and that input did result in changes to the CoC update.
Could there be a better process? I think we're all open to input. If someone would like to suggest a better way to handle things, that would be great. I will disclose upfront, though, that I will have to politely disagree with the proposition that completely free and open discussion is always the magic bullet solution.
Regards,
- Ted
Quoted 3 lines · Theodore Y. Ts'o
Quoted 2 lines · Shuah Khan
Quoted 2 lines
Quoted 2 lines
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
I agree with this statement.
Quoted 4 lines
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs.
I believe that did not happen as described as at least I was not asked for input/comment/sign-off and I am well within that top 200 or so list.
Quoted 6 lines
So there *was* community input, and that input did result in changes to the CoC update.
Could there be a better process? I think we're all open to input. If someone would like to suggest a better way to handle things, that would be great.
Open posting of suggested changes with a waiting period for comment of at least a week.
Theodore Y. Ts'o
maintainerQuoted 3 lines · Joe Perches
I believe that did not happen as described as at least I was not asked for input/comment/sign-off and I ahm well within that top 200 or so list.
I'm not sure precisely which metrics Greg used when he created his list, but using "git log --since="July 2017" v4.18" and piping it into the perl script I use to count Signed-off-by, Acked-by, and Reviewed-by, lines, you're about #240; so you're in the top 250, but not the top 200.
There are other metrics that could be used as well, such using the number of lines added/removed to create the ranked list. Different metrics will have differing amounts of bias for or against people who generate large number of "rename variables to random new types" or "whitespace cleanups" commits.
To be honest, that's one of the weaknesses of only using git statistics. They have the advantage that they are objective; however, especially if the exact metric which is used is revealed, such systems can also be easily gamed. (e.g., do you create a single commit that fixes all of the white space in one subsystem or use a separate commit for each file?)
And of course, using commit statistics may not accurately measure the value of say, a set of patches which add support for a new architecture or fixes a tricky locking problem, versus commits which only fix checkpatch complaints. So I will be the first to admit that using git statistics has its limitations.
Regards,
- Ted
Quoted 10 lines · Theodore Y. Ts'o
Quoted 3 lines · Joe Perches
I believe that did not happen as described as at least I was not asked for input/comment/sign-off and I ahm well within that top 200 or so list.
I'm not sure precisely which metrics Greg used when he created his list, but using "git log --since="July 2017" v4.18" and piping it into the perl script I use to count Signed-off-by, Acked-by, and Reviewed-by, lines, you're about #240; so you're in the top 250, but not the top 200.
Acks are nearly useless as a metric.
Non-primary Sign-offs are less so.
By author is a rather more useful one.
$ git shortlog -n -s --since=<foo>
Separating whitespace/checkpatch style patches from actual defect fixes and other enhancements is, of course, a more difficult automation problem.
Quoted 10 lines · Theodore Y. Ts'o
Quoted 3 lines · Joe Perches
I believe that did not happen as described as at least I was not asked for input/comment/sign-off and I ahm well within that top 200 or so list.
I'm not sure precisely which metrics Greg used when he created his list, but using "git log --since="July 2017" v4.18" and piping it into the perl script I use to count Signed-off-by, Acked-by, and Reviewed-by, lines, you're about #240; so you're in the top 250, but not the top 200.
Please also consider maintainers who are pulling in sub-trees without adding signed-off-by to each patch.
Signature · 2 lines
James Morris <[email protected]>
[Ksummit-discuss] better hot-topic discussion processes was: Re: TAB non-nomination
In reply to Theodore Y. Ts'o+Kostantin
Hi Ted,
Quoted 11 lines · Theodore Y. Ts'o
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs. So there *was* community input, and that input did result in changes to the CoC update.
Could there be a better process? I think we're all open to input. If someone would like to suggest a better way to handle things, that would be great. I will disclose upfront, though, that I will have to politely disagree with the proposition that completely free and open discussion is always the magic bullet solution.
Ok, I'll take a stab at that. :)
I'll make the assumption that there was nothing said in the "invited" discussion that the speaker would object to being a matter of public record.
So I see two possible avenues to improve the process. Keeping in mind that these are rare, and thus the process shouldn't be a burden.
The first is what I'll call a "delayed archive." Basically, you Cc the archive in along with the invitees to the discussion. The archive server then adds the emails to the public archive 90 days after the email was sent. This might also work well for the security ML.
The second is more of a "distributed moderation" model. A mailinglist would be added to the Cc where only members can post. Non-member posts are dropped. However, there are also "subscribers"; they can't post, but they receive posts to the ML.[1] There is no delay.
So, if I'm just a subscriber, but I know James, and I want to contribute to the discussion, I send my reply to him, and he can forward it if relevant.[2] James can filter any emails coming in against his address book so he's not bothered with people he doesn't know (Don't we all already do this? :-) ). The same goes for all the other members. The idea here is avoid the single-point-of-failure of having a single moderator. Optionally, the "members" could be automatically assigned that status based on the git history.
thx,
Jason.
[1] Another way to describe this is an ML for which posting is limited to subscribers that meet a given threshold within the git history.
[2] programmatically, we could say that James' address is in the To: and all others in the Cc:. This would allow for easier spotting of "I'm being asked to moderate this."
Re: [Ksummit-discuss] better hot-topic discussion processes was: Re: TAB non-nomination
Quoted 17 lines · Jason Cooper
+Kostantin
Hi Ted,
Quoted 11 lines · Theodore Y. Ts'o
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs. So there *was* community input, and that input did result in changes to the CoC update.
Could there be a better process? I think we're all open to input. If someone would like to suggest a better way to handle things, that would be great. I will disclose upfront, though, that I will have to politely disagree with the proposition that completely free and open discussion is always the magic bullet solution.
Quoted 5 lines
Ok, I'll take a stab at that. :)
I'll make the assumption that there was nothing said in the "invited" discussion that the speaker would object to being a matter of public record.
I'm snipping the rest because you lost me right here ;)
-chris
Re: [Ksummit-discuss] better hot-topic discussion processes was: Re: TAB non-nomination
Hi Chris,
Quoted 14 lines · Chris Mason
Quoted 13 lines · Jason Cooper
Quoted 11 lines · Theodore Y. Ts'o
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs. So there *was* community input, and that input did result in changes to the CoC update.
Could there be a better process? I think we're all open to input. If someone would like to suggest a better way to handle things, that would be great. I will disclose upfront, though, that I will have to politely disagree with the proposition that completely free and open discussion is always the magic bullet solution.
Quoted 7 lines
Quoted 5 lines
Ok, I'll take a stab at that. :)
I'll make the assumption that there was nothing said in the "invited" discussion that the speaker would object to being a matter of public record.
I'm snipping the rest because you lost me right here ;)
In light of your other, slightly more verbose email (venue for objectors to speak freely), I agree. I was not aware that was a primary reason for the "community, but non-public" discussion.
That is a tremendously valuable aspect of the decision, that I'd not seen conveyed clearly until your email. Thanks for sharing it.
thx,
Jason.
Quoted 25 lines · Theodore Y. Ts'o
Quoted 7 lines · Shuah Khan
Quoted 2 lines
Quoted 2 lines
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
It is unfortunate it had to start that way. I also understand at times it might be necessary to do so based on my experience with the Linux Kernel Community Enforcement Statement process. What should TAB do as a body if it needs to take action without an option to initiate an open discussion?
As Chris mentioned, there was a large amount of community input on the update. It just didn't happen in an open fashion. One of the challenges with e-mail discussion is that it can end up get dominated by a small number of people who send a large number of messages. In an in-person meeting, a good moderator can say, "Alexis, you've been talking a lot; perhaps we should hear from some other people who have been quiet. Drew, what do you think?" It's a lot harder to do this on a mailing list.
The second challenge is that we were getting trolled by people who were *not* members of the kernel development community. I was able to track down one such troll to their social media presence on gab.ai. (Yeah, that same lovely "free speech absolutists' site, for when Twitter considers you too toxic" which got deplatformed after the shooting at the Tree of Life Synagogue in Pittsburgh.)
Quoted 5 lines
So what was done with the update to the CoC was that a proposed set of changes was sent out to the top 200 or so contributors to the kernel, by git statistics over the past year, asking for their comments and their sign-offs. So there *was* community input, and that input did result in changes to the CoC update.
Good to know this did result in changes. But probably only for the first version?
I hadn't checked before, but the actual patches that went in are exactly the same as the patches I acked.
Nevertheless, if one is asked through a private channel if one wants to ack some patches that have already 10+ acks, there are basically 3 options (excl. forcing a public disclosure):
1. Ack the patch, if you (mostly) agree with it,
2. Nak the patch, which probably won't have any effect, due to the private channel,
3. Ask for changes to be made, which may be complicated, as the version you received already has several acks that may become invalid, and it's a private channel. Hence this may also don't have any effect.
So even if one considers option 3, depending on the amount and severity of the changes one has in mind, option 1 might be the better option. There are still opportunities for improvement later, quoting the second patch "[...] can be seen as a bug, though; such bugs will be fixed more quickly if any interested parties submit patches to that effect."...
However, the latest message seems to be "don't change the CoC anymore", and no more patches have been (re)sent...
(Yes, I'm "guilty" here too, perhaps everyone is waiting for someone else to (re)submit first?)
Gr{oetje,eeting}s,
Geert
Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected] In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds
Quoted 10 lines · Chris Mason
Quoted 4 lines · James Bottomley
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today.
I think panic is the wrong word to attach to Linus' response, especially around the code of conduct.
Quoted 1 line
Quoted 7 lines
Quoted 4 lines
The second mistake was picking the wrong CoC. [ ... ]
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input
The update was entirely based on community input.
I am going to try to parse that sentence very carefully and narrowly.
If you are saying that the update (that is, code-of-conduct-interpretation.rst) then I would agree that the document appears to have been created based on community input. But that is merely a conjecture on my part since the document was created in a small closed group.
If you are saying that the creation of code-of-conduct-interpretation.rst was done in a process that was open and visible to the community, then I would disagree. I don't know if this is what you meant to convey, but it is very easy to interpret the sentence in this way.
-Frank
Quoted 33 lines
Quoted 2 lines
which has generated a
lot of obvious anger within our community itself.It's absolutely true that some members of the community were upset.
We'll never know if there could have been a better time to make code of conduct changes. There are a wide range of deeply held beliefs in this area, and every choice would have eventually led to major disagreements.
But what we do know is that everyone sat down and did their best to
find compromise. That doesn't mean we found the right compromise for every developer, but I still really appreciate how much time and energy everyone spent explaining their point of view and looking for common ground.Quoted 6 lines
All I'll say on this
is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...It's really important the TAB is full of people that care about the kernel. Anger about the code of conduct isn't a great qualifier, but I'll happily encourage anyone who cares deeply about the kernel community, even if they disagree with my opinions about how to best support it.
-chris
_______________________________________________
Tech-board-discuss mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/tech-board-discuss
Quoted 30 lines · Frank Rowand
Quoted 12 lines · Chris Mason
Quoted 5 lines · James Bottomley
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation
that we needed an updated Kernel CoC ASAP, like today.I think panic is the wrong word to attach to Linus' response, especially
around the code of conduct.Quoted 1 line
Quoted 8 lines
Quoted 5 lines
The second mistake was picking the wrong CoC. [ ... ]
The third mistake was dumping the fully formed CoC and a later update
into the tree with little to no community inputThe update was entirely based on community input.
I am going to try to parse that sentence very carefully and narrowly.
If you are saying that the update (that is,
code-of-conduct-interpretation.rst)
then I would agree that the document appears to have been created based on community input. But that is merely a conjecture on my part since the document was created in a small closed group.
Quoted 6 lines
If you are saying that the creation of
code-of-conduct-interpretation.rst
was done in a process that was open and visible to the community, then I would disagree. I don't know if this is what you meant to convey, but it is very easy to interpret the sentence in this way.
Ted's earlier reply has a good summary, but the part I want to underline is that we sought out people who strongly disagreed with us, and we did our best to understand their concerns. It was important to me that we give people a private channel to express themselves, especially considering that the topic at hand was behavior on public lists.
It was a tradeoff, but I was really happy with the number of people who participated who might otherwise have stayed out of the discussion completely.
-chris
Hi Chris,
On Sat, Nov 10, 2018 at 07:15:16PM +0000, Chris Mason wrote:
...
Quoted 9 lines
Ted's earlier reply has a good summary, but the part I want to underline is that we sought out people who strongly disagreed with us, and we did our best to understand their concerns. It was important to me that we give people a private channel to express themselves, especially considering that the topic at hand was behavior on public lists.
It was a tradeoff, but I was really happy with the number of people who participated who might otherwise have stayed out of the discussion completely.
This is the most concise and clear explanation for the path taken. Thanks for sharing it with everyone.
thx,
Jason.
Quoted 41 lines · Chris Mason
Quoted 30 lines · Frank Rowand
Quoted 12 lines · Chris Mason
Quoted 5 lines · James Bottomley
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation
that we needed an updated Kernel CoC ASAP, like today.I think panic is the wrong word to attach to Linus' response, especially
around the code of conduct.Quoted 1 line
Quoted 8 lines
Quoted 5 lines
The second mistake was picking the wrong CoC. [ ... ]
The third mistake was dumping the fully formed CoC and a later update
into the tree with little to no community inputThe update was entirely based on community input.
I am going to try to parse that sentence very carefully and narrowly.
If you are saying that the update (that is,
code-of-conduct-interpretation.rst)
then I would agree that the document appears to have been created based on community input. But that is merely a conjecture on my part since the document was created in a small closed group.Quoted 6 lines
If you are saying that the creation of
code-of-conduct-interpretation.rst
was done in a process that was open and visible to the community, then I would disagree. I don't know if this is what you meant to convey, but it is very easy to interpret the sentence in this way.
Quoted 5 lines
Ted's earlier reply has a good summary, but the part I want to underline is that we sought out people who strongly disagreed with us, and we did our best to understand their concerns. It was important to me that we give people a private channel to express themselves, especially considering that the topic at hand was behavior on public lists.
OK. So the update was done in an opaque closed fashion, which involved soliciting input from some unknown fraction of the community. Do I understand that correctly?
And I think it would be fair to say that the people who created the update were probably aware of the comments of a much larger group of people who had participated in the threads on various email lists, and also I suspect the comments threads on the related lwn articles. So likely also based on input from a (probably) larger fraction of the community who had been willing to publicly comment.
So based on community input, but the document was not reviewed by the broader community, or accepted by the broader community.
Note that my opinion is that code-of-conduct-interpretation.rst is a step in a positive direction. I'm just disappointed that it was not submitted through the normal process of review (and no, sending a patch to the mail list on Saturday and merging the patch two days later on Monday is not normal review, especially when some people were traveling to OSS Europe, ELC Europe, and the Maintainers Summit that weekend).
-Frank
Quoted 6 lines
It was a tradeoff, but I was really happy with the number of people who participated who might otherwise have stayed out of the discussion completely.
-chris
Theodore Y. Ts'o
maintainerQuoted 11 lines · Frank Rowand
OK. So the update was done in an opaque closed fashion, which involved soliciting input from some unknown fraction of the community. Do I understand that correctly?
And I think it would be fair to say that the people who created the update were probably aware of the comments of a much larger group of people who had participated in the threads on various email lists, and also I suspect the comments threads on the related lwn articles. So likely also based on input from a (probably) larger fraction of the community who had been willing to publicly comment.
Quoted 2 lines
So based on community input, but the document was not reviewed by the broader community, or accepted by the broader community.
"Community" is a very slippery term. I will note that there were *many* people who were participating on the threads, sometimes in very non-constructive or in a downright toxic fashion, who had zero commits in recent years. In some cases, it was zero commits, *ever*. I recall doing the research on one prolific author and found that while he did contribute the kernel, it was 3 or 4 commits... ~5 years ago... to a driver.
And then there was one person who admitted that while he was just a user, he insisted he had a right to weigh in the issue. They certainly have the right to have that belief, of course. Whether or not maintainers are obliged to cater to people with those beliefs is a very different question, however.
There seems to be an assumption that a open, public discussion will always give you the best review. I don't think that's necessarily true. It can often give you a very biased sample from the poeple who are most stridently on one side of the debate or the other, as well as being biased towards those who believe in the "last post wins" style of debate, since they end up speaking most loudly and posting most frequently and most aggressively.
I found it very interesting that by explicitly asking the top ranked developers by git statistics for comments and for their sign-off on the various update patches, we got a much broader read on what people thought, and received some very thoughtful comments --- from people who had *not* engaged on the public threads.
At this point, Linus has indicated that he would prefer that we not try to tweak the CoC any further, and let's see how it works out in practice. If new developers continue to report that they feel more welcome, and we get more news reports like this:
https://www.zdnet.com/article/a-kinder-gentler-linus-torvalds-and-linux-4-20/
... and no one is getting kicked out of Linux development for being politically incorrect, and the quality and quantity of kernel code continues to increase, it'll all be good and we can spend most of our time worrying about technical rather than political issues.
- Ted
Quoted 23 lines · Theodore Y. Ts'o
Quoted 11 lines · Frank Rowand
OK. So the update was done in an opaque closed fashion, which involved soliciting input from some unknown fraction of the community. Do I understand that correctly?
And I think it would be fair to say that the people who created the update were probably aware of the comments of a much larger group of people who had participated in the threads on various email lists, and also I suspect the comments threads on the related lwn articles. So likely also based on input from a (probably) larger fraction of the community who had been willing to publicly comment.
Quoted 2 lines
So based on community input, but the document was not reviewed by the broader community, or accepted by the broader community.
"Community" is a very slippery term. I will note that there were *many* people who were participating on the threads, sometimes in very non-constructive or in a downright toxic fashion, who had zero commits in recent years. In some cases, it was zero commits, *ever*. I recall doing the research on one prolific author and found that while he did contribute the kernel, it was 3 or 4 commits... ~5 years ago... to a driver.
Quoted 8 lines
And then there was one person who admitted that while he was just a user, he insisted he had a right to weigh in the issue. They certainly have the right to have that belief, of course. Whether or not maintainers are obliged to cater to people with those beliefs is a very different question, however.
There seems to be an assumption that a open, public discussion will always give you the best review.
Maybe not, but it does help create a sense of community. It encourages people to feel valued and included.
The new CoC suggests that our standards include
* Focusing on what is best for the community
Is having a perfect CoC best for the community? Or is having an open process best, even when it produces a suboptimal result?
NeilBrown
Quoted 4 lines · Theodore Y. Ts'o
At this point, Linus has indicated that he would prefer that we not try to tweak the CoC any further, and let's see how it works out in practice.
I do love this. Linus, who is apparently the cause of the problem, is seen to be wise enough to be able to decide when we should start the conversation, and when we should stop it.
To me, this just seems like Linus throwing his weight around in an area where he has already demonstrated his incompetence.
Of course we cannot really see how the CoC "works out in practice" because you changed two variables at once. Linus decided to behave differently and a CoC was "adopted". We will never know which change had which effect.
NeilBrown
Quoted 7 lines · On Mon, 12 Nov 2018 15:54:56 +1100 NeilBrown <[email protected]> wrote:
Quoted 4 lines · Theodore Y. Ts'o
At this point, Linus has indicated that he would prefer that we not try to tweak the CoC any further, and let's see how it works out in practice.
Quoted 6 lines
I do love this. Linus, who is apparently the cause of the problem, is seen to be wise enough to be able to decide when we should start the conversation, and when we should stop it.
To me, this just seems like Linus throwing his weight around in an area where he has already demonstrated his incompetence.
No, think if it this way. Code got into the kernel. Perhaps how it got in wasn't the best way, but regardless it's in the kernel. Now we are not going to touch it until someone shows that the code is actually broken, with real examples. In other words, it's about not trying to fix things for theoretical situations or hand waving. When it proves that there is an actual bug, we'll fix it then.
I personally do not agree with this method, but I'm willing to try it out. Linus has made that same argument for real code in the kernel before.
-- Steve
Quoted 14 lines · Theodore Y. Ts'o
"Community" is a very slippery term. I will note that there were *many* people who were participating on the threads, sometimes in very non-constructive or in a downright toxic fashion, who had zero commits in recent years. In some cases, it was zero commits, *ever*. I recall doing the research on one prolific author and found that while he did contribute the kernel, it was 3 or 4 commits... ~5 years ago... to a driver.
And then there was one person who admitted that while he was just a user, he insisted he had a right to weigh in the issue. They certainly have the right to have that belief, of course. Whether or not maintainers are obliged to cater to people with those beliefs is a very different question, however.
Quoted 7 lines
There seems to be an assumption that a open, public discussion will always give you the best review. I don't think that's necessarily true. It can often give you a very biased sample from the poeple who are most stridently on one side of the debate or the other, as well as being biased towards those who believe in the "last post wins" style of debate, since they end up speaking most loudly and posting most frequently and most aggressively.
I think it's interesting to note that ksummit-discuss seemed to be a much more fruitful list for discussion than LKML in this matter. Would we benefit from a non-technical mailing list to accompany LKML? Something less "seasonal" than ksummit-discuss.
Could also restrict posting to subscribers only, with public archives, and limit subscribers to, say, people in MAINTAINERS and/or some (fairly low) number of required commits in the past years. I presume ksummit-discuss had fairly good signal-to-noise because it's not well known or publicized.
BR,
Jani.
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
Hi Jani,
Quoted 23 lines · On Tuesday, 13 November 2018 18:49:58 EET Jani Nikula wrote:
Quoted 14 lines · Theodore Y. Ts'o
"Community" is a very slippery term. I will note that there were *many* people who were participating on the threads, sometimes in very non-constructive or in a downright toxic fashion, who had zero commits in recent years. In some cases, it was zero commits, *ever*. I recall doing the research on one prolific author and found that while he did contribute the kernel, it was 3 or 4 commits... ~5 years ago... to a driver.
And then there was one person who admitted that while he was just a user, he insisted he had a right to weigh in the issue. They certainly have the right to have that belief, of course. Whether or not maintainers are obliged to cater to people with those beliefs is a very different question, however.
Quoted 7 lines
There seems to be an assumption that a open, public discussion will always give you the best review. I don't think that's necessarily true. It can often give you a very biased sample from the poeple who are most stridently on one side of the debate or the other, as well as being biased towards those who believe in the "last post wins" style of debate, since they end up speaking most loudly and posting most frequently and most aggressively.
Quoted 4 lines
I think it's interesting to note that ksummit-discuss seemed to be a much more fruitful list for discussion than LKML in this matter. Would we benefit from a non-technical mailing list to accompany LKML? Something less "seasonal" than ksummit-discuss.
Or could it be because the list is less known by the general public, and thus not as targeted by trolls ? Maybe I'm being overly negative there, a new mailing list might help. We would need to make sure to get people to subscribe (yet another mailing list...), but on the other hand the much lower traffic compared to LKML might help regain people who are not subscribed to or just don't read LKML.
Quoted 3 lines
Could also restrict posting to subscribers only, with public archives, and limit subscribers to, say, people in
MAINTAINERSand/or some (fairly low) number of required commits in the past years.
Limiting subscription to MAINTAINERS would in my opinion keep out too many people with a real interest and useful feedback to provide. Limiting it to contributors would be better, even if possibly too limiting. We could experiment with that though and see how it works out.
In any case, if we want to restrict list subscription (or at least post rights) to kernel community members, we will need to answer the elephant in the room question: who is the kernel community ?
Quoted 2 lines
I presume ksummit-discuss had fairly good signal-to-noise because it's not well known or publicized.
I should have read your whole e-mail before writing the reply :-)
Signature · 3 lines
Regards, Laurent Pinchart
Quoted 8 lines · Theodore Y. Ts'o
There seems to be an assumption that a open, public discussion will always give you the best review. I don't think that's necessarily true. It can often give you a very biased sample from the poeple who are most stridently on one side of the debate or the other, as well as being biased towards those who believe in the "last post wins" style of debate, since they end up speaking most loudly and posting most frequently and most aggressively.
Quoted 5 lines
I found it very interesting that by explicitly asking the top ranked developers by git statistics for comments and for their sign-off on the various update patches, we got a much broader read on what people thought, and received some very thoughtful comments --- from people who had *not* engaged on the public threads.
One thing that other organisations that do this sort of consultation process (which is very good and useful) do which didn't seem to happen here is that they announce that this is going on. This is useful because it helps people know what's going on and can help make the public discussion a bit more useful as a result. If, as is common, there's also some way for people to push in comments those can turn up useful feedback that people you wouldn't have thought to ask aren't comfortable providing in public though you'll also tend to get a lot of noise in there as well if it's a contentious area so that's hard work.
Quoted 21 lines · On Thu, 08 Nov 2018 16:04:02 -0800 James Bottomley…
Hi All,
Several people have asked me to stand again for election to the TAB, so I thought I'd give a general explanation of why that isn't going to happen. For background: I was one of the people who lead the charge in getting the Linux Foundation predecessor OSDL to create the TAB to give developer input into what was then seen as a body trying to speak for the Linux Kernel. I was actually TAB chair for 8 years from 2006 to 2014. The job of the TAB, as I saw it, was to solve a lot of the political friction issues around the places where the Linux Kernel community interfaces with the Linux Foundation and Industry and in those 8 years I gained quite a lot of expertise in political relations trying to do that.
However, TAB member and TAB chair aren't roles people are born to fill, they're roles people have to grow into. What our community needs is more people willing to grow into these roles to ensure effective succession and if I stepped into one I'd be denying others that opportunity, which would be bad both for succession planning and the growth of the community in general.
Quoted 7 lines
I think the reason I'm getting these requests is angst over this CoC debate, so I'll go so far as to detail my political instincts over this below ... if you've no interest in politics (as most of you won't have) stop reading now. If you are interested, perhaps you should consider standing for the TAB yourself.
James
Thank you for your service.
Quoted 6 lines
The biggest political mistake was actually doing anything with the Linux Kernel CoC at all. The object was to deflect a highly unfavourable article in the New Yorker. With hind sight, that could be achieved simply by Linus' personal apology, statement that he was stepping aside and going for assistance to understand others' emotions.
Quoted 11 lines
Hind sight, though is always perfect. At the time, as a TAB member, all you saw was a panic driven by both Linus and the Linux Foundation that we needed an updated Kernel CoC ASAP, like today. Panic is very infectious so it can be extremely difficult in these circumstances to stand up and say "stop, we need more information" ... and if you think you'd be the one always to demand more information remember that there's a time a decision has to be made and it always passes before you can get complete information, so you'd basically be rendering the TAB indecisive and useless. Recognising when it's time to stop and ask for more data and when you have to make decisions with what you have is a key political skill.
Based on what it looked like from outside the process, this analysis matches what I observed.
Quoted 11 lines
The second mistake was picking the wrong CoC. I'm not talking about the wording, which has been discussed on this list, but the politics surrounding the choice: The original author of the current CoC was unsupportive to the point of attacking the kernel community in public. That drove a huge amount of me too attacks plus an equally large amount of anti-me too hysteria and lead to enormous external awareness and friction plus a not inconsiderable amount of unwelcome personal email to various people. This could largely have been avoided by either evolving our existing CoC through a community process or by picking a CoC whose original author would be willing to stand up and be supportive of our desire to change.
There is precedent for taking something which did not match the political alignment of the community (GPL license) and modifying the interpretation to meet the consensus of the a subset of the community. I don't think even today that the Linux kernel developers beliefs around GPL match the purist camp.
Quoted 7 lines
The third mistake was dumping the fully formed CoC and a later update into the tree with little to no community input which has generated a lot of obvious anger within our community itself. All I'll say on this is that revisiting the CoC is going to cause another huge cascade of externally driven attacks which I think we'd all rather avoid, so if you're still ticked, then perhaps you should channel that anger and stand for the TAB ...
This is where I disagree. Without doing something it would have caused even more attacks. This kind of messy surgery was necessary to fix the practice and perception of the kernel developers.
James, I am sorry to see you throw your hands up and give up. It is has always been valuable to have your considered insight even if at times we all disagree.