My general point of view is that FOSS software is a public good, and that we should aim to maximize the availability of that public good.<p>This point of view contrasts with other views, for example I don't see much value in ideological purity for its own sake.<p>For that reason, I think it's a bit foolish to dissuade people from hosting on GitHub as long as Microsoft is subsidizing bandwidth and hosting costs. You might as well stick MSFT with the bill.<p>But if you go that route, you do kinda have to keep your eyes open to the real possibility of an eventual rug pool. Or even just that MSFT will make incremental cost saving changes that render some uses of GitHub increasingly unusable over time.
The network effect really is important if you don't use email for patches and want to capture an audience greater than your typical zealot coder (of which I am one).<p>Github makes it easy for non-coders to submit documentation, test their changes, and genuinely be part of something that they are not an expert in. Sure there are disadvantages there too in terms of the quality of the submissions, but the benefits of a busy community outweigh the drawbacks of too many PRs in my opinion
Protesilaos is a legend and I agree with everything he's said on this post.<p>I myself put all my projects on GitHub because of visibility and the free CI to run tests on Windows/Mac/Linux which is a godsend... but I also normally keep an alternative upstream on all my projects with <a href="https://www.opencode.net" rel="nofollow">https://www.opencode.net</a> (it's as easy to do that as adding a new remote with git) so I feel like I have zero dependency on MSFT, all the services they provide to me for free are just "nice-to-have" things. But they are very nice indeed.<p>The only danger I see is that GitHub becomes so widespread that other tools start only integrating with it (instead of using pure "git" integration). People are saying NPM already does this (I don't know, I don't do NPM)? That's a real danger, but hopefully other projects will have the common sense to not fall into this trap. For now, at least, I feel like that's not been a big problem.
> I do not blame SourceHut, as they clearly state they are still in alpha<p>The decision to stick to the clunky email interface isn't due to stage name
I made the same excuses in the past, and I could not even keep making them to myself seriously after a couple years of watching Microsoft censor various repos.<p>At some point people put code on GitHub before it had "network effects".<p>If a package is useful enough, people created accounts on GitHub to get help or contribute.<p>Maintainers need to stop rationalizing away their -choice- to centralize code on GitHub.<p>We do not need governments to help here. The political votes that matter are every repo that someone chooses to not host on GitHub, or at a minimum mirror elsewhere. The more useful your software with fewer alternatives, the more "network effect" power you wield to get someone to sign up for something besides GitHub.<p>I started Stagex, which is as far as I know still the only oci-native, deterministic, multi-signed, and full-source bootstrapped Linux distribution that exists.<p>If you want easy deterministic containerized builds of software without trusting any single human in your compilation path, Stagex is the only option.<p>My shameless plugging aside, I chose to host this exclusively on Codeberg in spite of the major financial sponsors all being on GitHub.<p>Already we have 8 contributors and the ~200 most common packages needed for most python/tcl/perl/lua/c/c++/go/rust software with more language support in progress by volunteers.<p>We built Stagex to ensure a fully open and transparent supply chain for any software to be built with, and that starts with using a Git host that shares these values.<p>Codeberg also donated unlimited free CI/CD via woodpecker as is available to most projects that apply since their funding is exclusively for social good.<p>We will self host eventually when forgejo supports federation so our instance can still get contributions from Codeberg users.<p>Git was built to be decentralized and that is never going to happen under Microsoft.<p>I know it is harsh but if you have a popular project and exclusively host on GitHub, you are why we are in this monopolozed-network-effect situation, and you have the power to change it.<p><a href="https://codeberg.org/stagex/stagex" rel="nofollow">https://codeberg.org/stagex/stagex</a><p><a href="https://sfconservancy.org/GiveUpGitHub/" rel="nofollow">https://sfconservancy.org/GiveUpGitHub/</a>
I agree with many of OP's points, but I don't think OP is addressing the suggestion of using Codeberg, which I believe is a reasonable alternative to GitHub.<p>It provides a GitHub-like experience, so it solves the OP's problems with SourceHut.<p>Regarding discoverability, I doubt anyone finds projects through GitHub's search or discoverability features, but rather through a search engine like Google or DDG and online communities. So the project should be similarly discoverable if the source was in Codeberg.<p>The main friction point I can see is that many won't have a Codeberg user, so they'd need to sign up in order to collaborate, while most already have a GitHub account.
I want to comment on comparing things that seem similar, but aren't.<p>In particular, OP wants readers to believe that taking drugs developed by "big pharma" is the same, or is morally equivalent to hosting one's project on GitHub.<p>The problem with this is that the relationship between the drug manufacture and the drug user isn't the same as with the repository service and the user, and in a very important way.<p>* By using the drug the user doesn't contribute anything back to the manufacturer beside the monetary compensation. The equivalent of repository service would've been a drug user who agrees to run experiment for the company developing the drug in exchange for the drug. Such things do happen, especially in desperate cases... but, certainly aren't the norm of pharma company vs drug user relationship.<p>* Pharma companies are heavily regulated, with lots and lots of drug users protections in place. If a pharma company is found to have walked back on the safety or other promise about a particular drug -- there's a good chance they'll be taken to court, and, hopefully will have to pay reparations. GitHub, on the other hand, makes no promises to the non-paying customers, and, like a famous jedi, will force choke you when they alter the deal they made with you, while telling you to prey they don't alter the deal any further.
A lot of you are not taking about the biggest reason to avoid GitHub. Having the rug pulled from you.<p>Terraform? Remember that? VMWare? Redis? What happens when a company stops providing a good deal? With GitHub you have lock in to the way they do things. So many of you complained about CentOS and then go lock yourselves in to far worse when you have CentOS Stream which was mostly good enough.<p>You have a perfectly good option: GitLab. Yes, parts of it are closed source, but most of you will be perfectly fine with the open source stuff. You can move your repos over to a self hosted option for your independence with ease. Most of you do not care about that because you want to seem like you have a lot of activity on your GitHub.<p>I do not blame you. I literally had an employer reach out to interview me directly because of my GitHub activity. His job offer was absolutely terrible but it was nice for them to do that this time.<p>But most of you also want to entertain the idea of moving up in the world from a simple coder. A tech lead, a tech business owner, vc startup, run a nonprofit, etc. Then you care about the rug being pulled from you, either because it means less money for what you want to do or it is a massive hassle to move, and this is an excellent way to prevent that.
I don't think the author is painting a correct picture here.<p>Yes, SourceHut has a different modus operandi compared to GitHub, GitLab, Codeberg et. al when it comes to merging patches. However, this doesn't invalidate Codeberg/GitLab as a viable primary repo and patch collection path.<p>If the author is so adamant on network effect advantages, they can host a mirror on GitHub with an appropriate README.md pointing folks to right direction.<p>If the author needs to write a long post to validate themselves about using GitHub, there's no need. At the end of the day, what you say to others doesn't matter much. The biggest wisdom is to not lie to yourself about why you're doing something.
I think the root thing I disagree with in this essay is whether (GNU definition) freedom has anything to do with successful, thriving <i>economic markets</i>. No; they're completely orthogonal. The logic of this article might suggest Shenzhen is the freest place on earth—a fast-moving, healthy marketplace with a vast diversity of economic actors to transact with with low friction. Yes, but also no.<p>- <i>"Freedom and diversity go hand-in-hand. We empower people to express their individuality. The cumulative effect is a richer corpus of shared resources, from which we can all draw from to elevate our experience."</i><p>This is like a definition of wealth, not of freedom.
Wonderful article. Protesilaos does great work, and doesn't indulge in popularity contests, or shy away from uncomfortable realities as he sees them (which apparently is difficult for some HN readers). Keep it up, Prot.
There are many software collectives offering Gitea or similar for free or almost free, so there has to be a good reason for giving away one's code to MICROS~1.
this is the reasons why I kinda doesnt like it when an opensource package hosted on gitlab I mean idk if gitlab still foss or opensource company its just have same vibe like redis terraform or any other VC backed open source that try to search for the money