Thanks for making my argument [LWN.net]

Thanks for making my argument

Posted Mar 6, 2013 21:25 UTC (Wed) by man_ls (subscriber, #15091)
In reply to: Thanks for making my argument by airlied
Parent article: Canonical reveals plans to launch Mir display server (The H)

Let me wear my sophist hat for a minute. Where to start.

If you cannot make technical criticism, why do any criticism at all? After all you started with your criticism before even knowing there was a public repo, making a blunder in the process ("I remember when open source project announcements used to come with source").

Once you found the missing repo, why not take your time and make criticism of the few thousand lines of code? If Wayland is also a few thousand lines of code, is it enough to define it or not? Apparently a few thousand lines of code is not enough for you to understand, since you cannot make technical criticism; but it should be enough for Canonical engineers to not only make technical points (which after all can be superficial), but to completely understand it. And apparently not taking the time to understand another project in depth before criticizing it is equivalent to not being competent. Or perhaps a few thousand lines of code are not sufficient to evaluate and understand a complex project.

As I've pointed out on my blog they are hiding something or they are technically incompetent.

Maybe they are hiding something; after all they work for a company as you do and it has its agenda, just as yours. But that is a false dichotomy; maybe they just need something that moves faster and believe they can pull it. Or maybe Shuttleworth as a kid promised himself he would one day build a replacement for X; use your imagination instead of accusing others of incompetence. It will be more interesting.

The other thing is Red Hat (as a corporation) doesn't even see this stuff, they do pay me to see this stuff, but they don't direct how I respond and I don't speak on their behalf. My job is to do what I think is best for Linux graphics in the long term, like seriously that is actually what they pay me for, and if I think pointing out flaws in Canonical's arguments for NIHing things is best for Linux then I'll go ahead and do it!

Read what you wrote carefully. First you say that you don't speak on Red Hat's behalf, but then you go on to say that they pay you to speak your mind if you think something is wrong. Amazing.

Red Hat as a corporation sees this stuff, since you are part of that corporation and are paid to see it. As others have said, you would not be a good employee if you had not interiorized some business principles, such as avoiding projects not under the direct control of your corporation. Sometimes the goals of that corporation will align with the interests of Linux, sometimes it will not (like not publishing a git repo for the kernel). You think Mir is a not good idea; fine. You might also say "Let us wait and see what they come out with; there may be good ideas that we can adapt, and it is Free software after all" or "They have come with good stuff in the past that has served Red Hat well, like Upstart", but hey.

However, you are not only pointing out flaws in Canonical's arguments; you are actively despising them in public ("They should call the next Ubuntu Jumping Sharks"). It points to some kind of internal hostility towards Canonical, which somehow cannot surprise anyone at this point.

(Log in to post comments)

Thanks for making my argument

Posted Mar 7, 2013 0:20 UTC (Thu) by man_ls (subscriber, #15091) [Link]

Great, so you dislike Canonical and badmouth them in public. That is exactly what I was trying to establish (and something that other commenters obtusely refuse to see). It is your choice, but expect no sympathy for your cause from those grateful for Ubuntu.

Do you apply the same criteria always? Because I am eagerly awaiting your criticisms of Google Android, Tizen and the GNU project, just to name three prominent examples that are run more or less behind closed doors. (GNU is more open but still requires contributors agreement; the others are worse because AFAIK they don't even accept external contributions.)

Me, I am grateful for what I get, and contribute back what little I can.

Thanks for making my argument

Posted Mar 7, 2013 0:34 UTC (Thu) by airlied (subscriber, #9104) [Link]

Android never posted a criticism of a project I cared about, they did their own thing, I don't like how they do it, but I feel no need to comment on something that exists in its own universe. If they attempted to bring that universe into collision with mine, then yes I would be vocal about it.

If you believe signing the GNU contributors agreement equates to the Canonical contributor agreement, I have no further time for you, because frankly I expect to discuss things with people intelligent enough to interpret the agreements they sign and who they assign copyrights to.

Thanks for making my argument

Posted Mar 7, 2013 0:47 UTC (Thu) by man_ls (subscriber, #15091) [Link]

I find strange that doings things behind closed doors would be more palatable (or at least less deserving of your ire) than posting public documents. But everyone is entitled to their own contradictions, aren't we?

As to the GNU project, I think they do a great and necessary job, but their contributor agreement (reasonable as it is) and their development policies (cathedralicious according to some) have been criticized before. They are certainly less agile than they might be, and probably unlike every other project where you have worked before. Hence the comparison.

Thanks for making my argument

Posted Mar 7, 2013 1:35 UTC (Thu) by airlied (subscriber, #9104) [Link]

Android development model gets lots of criticism from lots of people, in the embedded space and everywhere else. Look at the kernel merge fun ongoing for years.

I have given out about Android and their encouragement of closed source graphics stacks before, and I do quite often at conferences and anywhere else. I've bitched about HW designs like the raspberry pi and I believe the FSF encouragement of such designs is totally anti their real mission.

Again you say the GNU project, but really its not one big monolith, you seem to be unable to distinguish individual pieces moving in their own directions, GNU projects are not beholden to some GNU development overlord.

some of them are more agile than others, the GNU project is an umbrella framework. Contributors to GNU projects again come from many companies and all believe that GNU is a good steward of the copyrights. (even if they don't use it enough to beat GPL violators with).

The thing is I don't believe Canonical is a good steward for my copyrights or anyone elses, and I think contributing code as an equal to their projects is very difficult. I've gone from 0 to being an integral part of 3 major projects *before* I joined Red Hat, and have never felt my company affiliation mattered in any of them, and was never asked to assign copyrights in any of them. Why would I want to give Canonical rights to take my code proprietary when they don't give me the same right? If you want the right to take code proprietary license it under MIT, and give the same right to other contributors (ala the Mesa 3D stack and X.org), if you don't want it use a GPL variant like the kernel. I also disliked Qt for doing the same for years, and MySQL also ran like that. It discourages individuals and other companies from making any decent contributions to your codebase, so in general you lose a major benefit of being open-source in the first place.

Its generally okay to give copyrights to a foundation that is setup correctly, but to individual companies, my personal believe is it doesn't end well.

Thanks for making my argument

Posted Mar 8, 2013 14:29 UTC (Fri) by nix (subscriber, #2304) [Link]

Sheesh. The GNU project was cathedralicious *sixteen years ago*. It's not any more.

Not a cathedral, not a bazaar

Posted Mar 8, 2013 15:05 UTC (Fri) by man_ls (subscriber, #15091) [Link]

You are right (and I should know since I have been a voluntary Savannah admin for a few years). When handling code it is still not as agile as other environments, but it is certainly not centralized. The decision process however remains largely in the hands of a couple of people, and there have been some complaints about it recently.

Let us see a practical example: inclusion in the project.The GNU project says about it:

Based on the evaluators' report, Richard Stallman (the Chief GNUisance) makes the final decision on whether to accept the contribution. [...] Thus, becoming a GNU maintainer is a somewhat formal process, since affiliating with the GNU project as a maintainer means you must agree to work (within the confines of the maintenance) with the GNU project's mission for software freedom.

Contrast with inclusion in other software collections:

Debian: have a Debian Developer sponsor your package.
CPAN: register, wait two weeks and upload your package.
The PyPI: register and upload.
NPM: npm publish.

The GNU project may not be a cathedral of software, but it is not exactly a bazaar just yet... and probably it never will given its goals.

Thanks for making my argument

Posted Mar 7, 2013 4:05 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]

"That is exactly what I was trying to establish (and something that other commenters obtusely refuse to see)."

Actually, no. What I refused to accept is that they are criticizing a project because of company affiliation. You are entirely wrong about.

Thanks for making my argument

Posted Mar 7, 2013 11:54 UTC (Thu) by paulj (subscriber, #341) [Link]

No one is accusing you of being a mouth-piece.

However, there is clear evidence that even well-intentioned, professional people tend to be affected by their affiliations. They tend to have biases toward arguing or acting in favour of the interests of their affiliations. This can happen subconsciously, even against the logical desires of the person. It's well documented in other fields, medical research particularly, and there are various psychological experiments you can do to demonstrate what is thought to be the underlying behaviour involved.

No matter how logical and dispassionate we try to be, our wetware just appears naturally to have a predilection for biasing us towards groups we are members of. There is no good reason to think software people would be immune to this.

Rather than deny this general fact, the best thing to do is to acknowledge it, so we then can find ways to minimise it or work-around it.

Thanks for making my argument

Posted Mar 7, 2013 12:23 UTC (Thu) by paulj (subscriber, #341) [Link]

Oh, and these effects occur even with full information. In the real-world, an obvious/frequent source of affiliation bias, is of having more complete or better information about your affiliations than of others.

Or as other commentators put it, communication issues. :)

Thanks for making my argument

Posted Mar 7, 2013 12:30 UTC (Thu) by HelloWorld (guest, #56129) [Link]

It's rather hard to be informed about what Canonical is doing when they do it behind closed doors. And that's actually what all this is about: the fact that Canonical utterly misses the point in Open Source Software: collaboration instead of competition.

Sources of bias

Posted Mar 7, 2013 12:41 UTC (Thu) by man_ls (subscriber, #15091) [Link]

In-group bias is another well-researched source of bias, that probably can be traced back all the way to primate groups. This is important not because it is more primitive, on the contrary: it is much more subconscious and entrenched than other, more cultural (you could say "human") biases. We want our group / tribe / company to thrive not because we will be better off, but because we are the best.

Sources of bias

Posted Mar 7, 2013 12:49 UTC (Thu) by paulj (subscriber, #341) [Link]

Yeah, I must have put it clumsily. My point was that lack of information is one obvious source of bias. But even when we have full information, we *still* have a definite tendency to be biased toward group members - even if subconsciously, as per your link.