[iamnothere]

Once again, the stability of a BDFL is vindicated.

Democracy works ok in extremely complex systems like giant distros—Debian’s governance is slow but functional—but it seems to need guardrails around the process. (From what I can tell, an excruciating process is the best guardrail. Glacially slow decision-making means that mobs burn themselves out before they get to voting.)

[dokyun]

I wouldn't call Drew DeVault stable, he crashed out after 4chan simultaneously exposed his creation of the RMS report alongside his huge loli porn fixation.

[iamnothere]

He may not be, but the product has been. Same goes with Theo and OpenBSD.

Hyper-productive people are often volatile. Putting up with it is a necessary tradeoff, as the systems they produce often fill a niche that nobody else will serve adequately.

[dokyun]

Guys like Theo have a baseline to uphold and I think decades of dealing with the public on an open project with specific goals that need to be met would turn anyone sour. People respect him for his actual work though, like they used to with Torvalds before he got bought out by IBM.

Drew I think is just kind of a dick who has a history of putting his personal crusades before delivering a product people want. I remember a few years back sourcehut being reported as unstable and he went and lashed out at people who were complaining to him about it instead of fixing the issues. And he banned crypto projects cause he doesn't like them, at least Codeberg held it to a vote.

Patrick Volkerding is a pretty stable guy, I think he's the model more BDFLs should strive to be.

[iamnothere]

Patrick is an unusual exception; I’d actually put people like Drew, Theo, and pre-corporate Linus in the midpoint. The opposite extreme of Patrick would be unfortunate cases like Reiser and Ian Murdock.

I think Drew suffers a little from lack of focus. I greatly appreciate his work on Harelang, sway, and aerc for instance, but these are quite far from his Sourcehut work. (I can see how aerc and Sourcehut dovetail with the plaintext email focus, at least.)

[tkel]

BDFL is worse for all the reasons you mention.

Stability? A BDFL can make decisions at a whim that radically destabilize and change the project. Lording over and causing immense harm to the community without any sort of accountability is a recipe for disaster.

Good for everyone? A BDFL can unilaterally make decisions that prioritize their own desires over everyone else. It becomes about their personality rather than the project.

Functional? A BDFL is less functional, seeing how there is no requirement that the BDFL do anything that the community of people actually wants.

If you're saying you're opposed to democracy because you don't like the decision that codeberg reached democratically, the BDFL can do the exact same thing. Except the process is far worse, because there is literally no input from anyone else on making the decision.

[iamnothere]

You can say all these things, and yet my experience with BDFLs in reality has been the opposite. More often, they make good decisions that continue their vision for the project, rather than abruptly redirecting the project. This is because productive people are often relatively consistent in their vision and direction.

Small community “democracy” OTOH often devolves into ochlocracy, or reactive mob rule. Small communities are easily swayed, don’t often have a durable vision, and can easily be infiltrated by people who don’t have the best interest of the project (or the users) at heart.

The only thing that can offset these effects is a heavy process, like Debian has, which slows down the mob and forces everyone to consider all sides of the issue. This can actually work! And for some projects it’s a better fit than a BDFL.

[tkel]

"Democracy works OK when I agree with the result"

[andai]

I think there is a slight difference between open source and government there. With open source you can just leave and pick another one. Or start your own.

Well I suppose that's also technically true of governments...