Linus Torvalds addressing maintainer fatigue

Linus emphasizes the challenge of finding and retaining maintainers due to the complexity of people's relationships.

[Aaron-slov]

It made me chuckle that when he said 'people are hard' that someone in the front row decides it's a perfect time to pull out their phone..

[SpencerHHO]

Despite being a passionate and fiery guy himself, Linus has a lot of charisma and social skills too. Open source is built on the work of thousands but I think people underestimate how lucky we are that one of the most prominent figures in opensource isn't a brilliant but anti social asshole. We need to protect Linus at all costs, he does a lot for keeping things on track.


EDIT:

As some have pointed out and I failed to properly acknowledge in my phrasing above Linus was and is still capable of being a very person. But as others have also pointed out, and I'd tend to agree, he has certainly made a concerted effort to work on people management/social skills for the betterment of the community.

I haven't met Linus but he reminds me of people I know who I'd describe as "caring/nice assholes" IE: abrasive and blunt but with good intentions and a willingness to learn and engage productively with other people even if this involves a degree of self criticism and reflection. I should have made it clear that I am comparing this with some other infamous/famous brilliant minds that seemed almost impossible to work with.

[DrakeDealer]

This man is such a treasure. The whole world is LUCKY to have him.

[andersjohnson]

Whether your project is open source or commercial, you really need to cherish your maintainers enthusiastically. I left Google shortly after finding out that there was a de facto "no promotion for maintenance" policy. (I was a tester, not a maintainer, BTW.)

[jonkoops]

The problem with getting new maintainers for the Linux kernel is barrier to entry. Having code reviews and patches go through mailing lists without any kind of CI is just an incredibly repetitive and frictionful experience. There are ways to improve this, and open source tooling around this exists, the project needs to make a move.

[TheUnkow]

He is spot on.

It's the people that don't read or follow guidelines or just have their own ideas that can downright steer the ship into oblivion in a matter of few small commits.

Having rules is very important.

Think of a great maintainer as a great forum moderator.

If they notice stuff off-topic they must either close or ban or move the discussion elsewhere.

Even posting similar bugs into a bug thread must instead go into a separate bug discussion, because in a bug thread you can only post info or discuss about the primary post of the reported bug.

Then you must ban those that make too many duplicates ... and so and so and so ...

[praecorloth]

Sounds like apprentice maintainer could be a thing.

[lashlarue]

I am high-key worried about what will happen to the Linux kernel if anything ever happens to Linus. I'm really worried that everything will all just get corrupted and atrophy.

[zxcvb_bvcxz]

Which conference did you rip this from without attribution?

[jorge.r.garciadealba]

It is fascinating to learn a little bit about how the kernel project is run in real life. There should be a book about this topic.

[Shifter-bphu]

Just hire full time developers as maintainers. Setup crowdfunding if the money isn't already there.

[varshneydevansh]

Maintaining the entropy is always difficult

[khuntasaurus]

most of the kernel code is very complex and undocumented. Coupled with the fact that changes are made through mailing lists, I don't think he will find anyone for the job any time soon until he gets with the times.

[fennyfinari]

So proud of you Torvalds, your communication to humans and computers alike have grown and soared, though developers still make up secret languages to code, I understand because of you alone, why certain steps may be necessary for adaptation into the coming of a new era. You as well as the GNU foundation inspire me to keep trying in all of life's endeavors. Thanks for being awesome!~

[Oodle-oxvf]

Yes, people are hard, and the Kernel must be protected.
An influx of enthusiasm and productivity may sound good, but, it would all crash without moderation, and revue from maintainers.
Even my best young inovators often write risky code. Only time and experience fixes this trait..

[IIIspirit]

Sounds you guys need a padavan system for maintainers somewhat to get practice?

[jbragg]

According to Wikipedia Linus gets paid to work on the kernel. I don't know the economic situation for most people working on the kernel, especially maintainers. I don't really know that ecosystem. But if you expect skilled people to work for free in this day and age, you live in a fairy-tale world. People need to eat, want to have income so they can live decently without struggling. Making this about ethics and morals of single developers and depoliticizing the issue is dumb.

[UniverseRobGames]

If you are the guy in the front row holding your camera up in the air completely oblivious to your surroundings; maybe skipping this lecture and watching it on YouTube would have been a better option.

[drac]

02:48 - It's incredibly frustrating to see someone blocking the view by raising their phone right in front of Linus's face during such an important discussion. We need to remember the basic etiquette of being in an audience, respecting both the speaker and those around us who are trying to engage.

[merthyr]

I don't think its a relationship issue, it's a technical one.

It's helped by the introduction of Rust into the kernel, but the fact a monolithic kernel like Linux is incredibly hard to maintain without serious manpower. It also makes managing the complex and busy git history of Linux very hard to do without relying on a handful of super-developers (who Linus claims you don't need to be to maintain the kernel, but with the thousands of commits, branches, and hardware required, you need to be) because any more than the few that can be easily managed within a mailing list quickly spirals out of control and into bikeshedding (or whatever corpo term for it).

The kernel was originally, and still is, designed around sending commits and patches to a single developer, and not the idea that multiple systems can be maintained by distinct teams. It's only barely working for Linux because it has so many stakeholders and contributors, but under most conditions this kind of developer system is unsustainable.