Hacker News Reader

Comment by microcolonel | Hacker News Reader

The problem here is that the "Collaborating" sector is the wrong goal for the Linux kernel. Linus Torvalds is uniquely qualified to be concerned about matters concerning the kernel; if his interlocutors "win" too much on their side of the "win/win", every user loses.

The public embarrassment of lazy and dishonest actors, especially those who are paid to collaborate, is an effective tool for maintaining order and standards on behalf of the quality of the product, and provided nobody gets hurt permanently at the end of the day, it is easily justified.

View on HN

> if his interlocutors "win" too much on their side of the "win/win", every user loses.

You seem to be trying to find a way to turn this back into "win/lose".

Try another interpretation.

For instance, a patch might call attention to a problem, but it takes the wrong approach to solve it. Could we find a better solution to the problem?

Nobody is suggesting giving bad code or bad ideas a pass. Are you under the impression that it's impossible to reject a patch or give technical feedback without being brutal or toxic?

A large chunk of frequent contributors are paid by companies to heave features/drivers at the kernel with no regard to doing it the right way. To them it's about landing code ASAP.

Linus is already working in the win/win quadrant by default by merging these things all in the name of making the kernel better.

Occasionally a frequent contributor who has gained trust will start to merge trash in the name of meeting deadlines. This is usually when linus loses his shit at the people responsible.

This has nothing to do with a silly quadrant of being competitive/neutral/anything towards the collaboration itself. Linus is still looking for a win/win, but he's requiring people to fix their shit.

That quadrant is irrelevant about Linus's behavior because he's always collaborating. What you need is a quadrant describing how feedback is delivered to subordinates.

I think generally you're right, but the comment you're replying to has something to it as well. If two parties are not "equals" in the arena they are in, things get Complicated. I'm not sure true collaboration is possible in situations where parties are on such uneven footing (and I mean this in the abstract sense as well: I don't think the model you've presented is complete as is does not give you a productive way to structure this kind of relationship).

In the situation where Linus is far more qualified, and obviously so, and someone is trying to get him to "approve" a patch, there's no way it's ever really going to be a truly collaborative dynamic.

This doesn't mean Linus has to be a dick about it, but it's definitely not collaboration between equals.

If you read what Linus said, nowhere did he conflate technical merit with social merit. Your comment very much does this. He specifically said there haven't been pain points around technical issues, but around process issues, which is what he's personally trying to fix.

Open source is about the community, and I think Linus figured this out later than most. Sure, you can dump some code out there and maybe even get users, but you won't have built a community that improves software until you've shown people can file bugs reports and patches or otherwise be a part of the community without being personally insulted. This is a pretty low bar tbh.

    The public embarrassment of lazy and dishonest actors,
    especially those who are paid to collaborate, is an 
    effective tool for maintaining order and standards on 
    behalf of the quality of the product, and provided nobody 
    gets hurt permanently at the end of the day, it is easily 
    justified.

Can you explain why you feel rants and insults are necessary to maintain high technical standards?

Literally nobody is asking Linus to relax his standards for what makes it into the kernel.

I suppose the theory is that fear of embarrassment makes people put in extra effort before submitting anything. This would reduce the time spent politely rejecting submissions.

Intuitively, in my opinion this sounds plausible.

I think fear also hinders innovation as it hinders thought and disruption at times. I am sure, to a certain degree, it's good to be fearful and understand the stakes but I doubt this has to be something maintained through verbal attacks, public humiliation and other forms of _abuse_.

I like Linus Torvalds. It's hard not to have a soft spot for him when his counterparts always were Bill Gates and Steve Jobs (for the longest time). But for this, there is no necessity of asshole-behavior.

Is that the "At least he's our asshole". That kinda makes sense to me.

Intuitively? have you ever contributed code to a FOSS project?

My experience is that I try to write the best code I can to be accepted, because I care about quality.

The embarassment (my feels) is gonna be because my code wasn't up to snuff.

I choose not to contribute to projects in which the feedback might be based on feels (vitriol) rather than reals (the code quality)

> Can you explain why you feel rants and insults are necessary to maintain high technical standards?

It is possible that they are not truly "necessary", but they are one often-effective way to discourage disruptive/unproductive behaviour. Linus ultimately should not be asked to review changesets he will be far from willing to merge; and for him there are only so many hours in the day. It is, as a practical matter, necessary to communicate that there is a cost to wasting his time with work that somebody should have known was substandard, and it is important that that cost is borne primarily by the originator of that substandard work (i.e. it needs to suck more for the committer to waste Linus' time than it sucks for Linus to spend time responding).

It is a high-risk, unappealing technique, but we should be careful not to impose alternatives which fail to meet the same criteria at least to some extent.

    there are only so many hours [for Linus] in the day. It 
    is, as a practical matter, necessary to communicate that 
    there is a cost to wasting his time with work that 
    somebody should have known was substandard

This is a really a restatement of the old (yet still relevant) "Linus doesn't scale" issue, right?

It's true: Linus' time is a valuable resource and shouldn't be wasted. This is of course true of any project maintainer, or any person at all really -- it's just that the stakes are a higher in the case of the Linux kernel because of the scope and impact of the project.

    it is important that that cost is borne primarily by the 
    originator of that substandard work (i.e. it needs to suck 
    more for the committer to waste Linus' time than it sucks 
    for Linus to spend time responding)

This is a nearly universal problem.

Imagine suggesting "hurl insults at the perpetrators until they learn" as a solution in any other field. Ridiculous!

The answer of course is to proactively reduce the number of time-wasting things that are raised to Linus' attention, not to reactively be meaner when those sorts of things do happen.

Typically you'd do this in Linux or any software project by:

(a) making sure that the other kernel maintainers are trusted and skilled (b) making sure that everybody has a solid understanding of what is and isn't a good contribution (c) discussing contributions ahead of time so that there are fewer surprises at review/merge time (d) solve the "Linus doesn't scale" issue by delegating more of his responsibility elsewhere, cloning him, etc.

I should note that I have no idea about how the kernel maintainers function on a daily basis. Maybe they're already doing a really great job of those things, and it's only 0.001% of merge requests that result in spectacular Linus rants and therefore show up on HN or whatever. Since it seems like kernel development is humming along pretty well for the mooooooost part I'd put my money on that, but who knows.

> Linus Torvalds is uniquely qualified

Gotta stop you right there. Hero worship should not be considered a good thing in our industry.

The kernel subsystem maintainers, who actually review the code prior to getting it merged, are the ones with the real power and the necessary knowledge, and I'd argue they are more qualified to oversee the project's future than Torvalds.

What makes Torvalds uniquely qualified is he is:

1) Technically capable of maintaining the kernel

2) Known to be mainly motivated by technical concerns

If you role the dice with another governance structure, you might get one that is motivated by "collaboration" and then the integrity of the kernel might get compromised by corporate interests. We have no idea how the subsystem maintainers would deal with the politics of a situation without Linus calling shots. We know how Torvalds reacts; rudely :P.

The choice isn't ever going to be Torvalds vs. Sunshine & Roses, the choice is Torvalds vs. Unknown. Because of that, he is practically (but not technically) unique.

I am in complete agreement that Linus is one of the best we have, and I personally think it would be a huge loss if he were to pull back from the project, and not only for technical reasons. Without sounding like I’m engaging in hero worship, I think him being around is usually positive. However, I’m not convinced—and it would seem he isn’t convinced either—that treating other humans as if they’re inhuman is a fundamental part of what makes him him.

I don’t know how we ended up at a point where we believe the only way to tell someone “No” or “You need to do better.” is through incredibly aggressive interactions. I suspect it’s just another hold over from a different time period, and we just never asked “Why?”

It sounds like he believes that treating other humans with respect won’t shatter him and I tend to agree with this assumption. His fundamental personality won’t go away and his skills certainly won’t suddenly disappear.

I’m interested to hear more regarding your comments about corporate interests. While I strongly and completely agree that we must resist letting their interests work their way into being driving forces, I’m not clear why some believe his most egregious behaviors are needed to keep these separations of intersts in place.

Yeah, just like regular employees are better qualified to run a company than the CEO.

No one is more qualified than Linus to run this project and he's done a damn good job so far. Has he made mistakes? Sure. He's human.

I think that when a company reaches a certain size and stability level, as long as the CEO is competent and understands how to balance external and internal issues, the particular CEO doesn't matter that much. The real power over product direction comes from the PMs, VPs, etc. That's my argument -- Linus isn't uniquely qualified to run the Linux kernel project, since he's barely even involved in the direction of the project.

> the particular CEO doesn't matter that much

Business history & Jack Welch disagree with you. Individual project success is incredibly nuanced and individuals - even in 10k+ orgs - can have massive impacts. Trying to pretend that outliers don't massively influence company success is naive, and MBA programs study these individuals because they are the exception and not the rule.

There's a reason we have Madison Square Garden, Rockefeller Center and Carnegie Mellon. It's not hero worship to acknowledge the contributions of individuals - recognizing a great general doesn't diminish the accomplishments of his army.

Consider how few "genius CEOs'" reputations have survived their attempts to take over failing companies and turn them around.

What do "genius CEOs" have to do with anything? Most successful CEOs were only responsible for a single business. Multiple successes don't change what Gates, Ellison, Knight et el. accomplished because they only did it once. Plenty of instances of people that failed plenty before eventually succeeded elsewhere (Ray Kroc).

Well, if CEOs didn't have that much bearing on the success of businesses, or don't have much bearing on the success of businesses once they reach a certain size, then seeing a wildly successful CEO completely flop at the next company would make sense.

> Yeah, just like regular employees are better qualified to run a company than the CEO.

I think you could pretty easily adapt Tolstoy's argument to business and say that the primary role of a CEO is to ratify all the on-the-spot decisions his subordinates made as events got away from them.

Who will maintain the maintainers, though?

I think that's what he's good at. They know they'll get called out if they try any BS so they mostly don't. And sometimes Linus overdoes it, but Linux keeps getting better.

Anyone who can do better can make their own fork. So far, everyone still follows Linus.

> The public embarrassment of lazy and dishonest actors, especially those who are paid to collaborate, is an effective tool for maintaining order and standards on behalf of the quality of the product, and provided nobody gets hurt permanently at the end of the day, it is easily justified.

It may be effective, but aren't there other equally effective alternatives? How are hard situations handled by maintainers of other operating systems/kernels like FreeBSD? Do they also need public embarrassment?

This has absolutely nothing to do with Linus winning or losing. Only with how he goes about winning. If he wins an argument in ways that alienate contributors, then he’s ultimately losing. There’s no reason he can’t be assertive without being toxic. I for one give him more credit than that. Calling people names is he laziest and least productive way to argue your point.

You're an idiot if you think insulting someone is effective shaming or the only way to do so. As the CoC points out, you can reprimand a community member in any number of ways and keep track of it.

I think this can be further generalized, that there is a tier of talent that we must excuse from having to filter themselves and that this may serve humanity even if it rubs some the wrong way.

Really? If they're so "lazy and dishonest," maybe they just shouldn't be involved in the project, instead of being allowed but also publicly berated.

This mindset is a bit too fascist for the modern workplace. I still see this attitude a lot, so I think it's good to recognize and call it out for what it is.

Please let's not bring the f-word into this. That's a move to an inner circle of hell.

Reply on news.ycombinator.com