Codeberg went dark on Saturday. No warning. No graceful degradation. Just a blank page staring back at thousands of developers who trusted it with their code, their projects, their communities. The mastodon status account posted a terse update — the kind that says everything and nothing at once. By the time you read this, it might be back up. Or maybe not. That's the point.
We've built the modern world on platforms that can vanish in an instant. GitHub goes down, and half the internet stops. GitLab has an incident, and CI pipelines everywhere break. Now Codeberg, the darling of the open source purists, the ethical alternative, the non-profit haven for those who fled Microsoft's empire — it's down too. And we're all supposed to shrug and say, "That's the cloud for you."
But it's not the cloud. It's a house of cards.
The Myth of Ethical Infrastructure
Codeberg positioned itself as the good guy. No venture capital. No ads. No data mining. Just a community-run forge for projects that believe in software freedom. It was the place you went when you wanted to escape the corporatization of open source. And it worked — for a while. Projects that couldn't stomach Microsoft's cozy relationship with Big Tech migrated there. They brought their issues, their pull requests, their entire development workflows.
And now those workflows are stalled. Code reviews that were about to merge — frozen. Bug reports that needed triage — inaccessible. Releases that were scheduled for Sunday — delayed. All because one server, or maybe a whole rack, decided to take a nap.
This isn't a hit on the volunteers behind Codeberg. They're doing heroic work with shoestring budgets. But that's exactly the problem. We've outsourced the backbone of software development to entities that can't guarantee uptime, can't afford redundant infrastructure, and can't be held accountable when things break. Open source is supposed to be about resilience. Instead, we've built a system that's fragile as glass.
The Fragility of Centralization
I remember when "the cloud" was sold as this magical thing that would never fail. Redundant servers. Automatic failover. Data replicated across three continents. That was the promise. The reality is that every platform, no matter how well-funded, has a single point of failure. Sometimes it's a human — someone fat-fingering a config change. Sometimes it's a cable — a backhoe in the wrong place. Sometimes it's just bad luck.
Codeberg's outage is a reminder that the open source movement has become dangerously centralized. We laugh at the idea of CVS or SVN now, but at least when your local repo was on your own hard drive, you could still work when the network was down. Now we've traded that for convenience. And convenience is a cruel master.
Every time a forge goes down, entire communities grind to a halt. That's not progress. That's dependency.
"We've outsourced the backbone of software development to entities that can't guarantee uptime."
Think about what happens when a project's infrastructure is unavailable for a day. Or a week. Or forever. The code doesn't disappear — you have your clones, your backups (you do have backups, right?). But the coordination does. The context does. The issue threads where decisions were made, the wiki pages that explain why that critical function exists, the CI/CD configuration that automates releases — all of it becomes inaccessible. And if the outage is permanent? That's not just an inconvenience. That's a cultural loss.
The Real Cost of "Free"
People love free stuff. Free hosting, free CI minutes, free package registries. And the platforms love giving it away — up to a point. GitHub is free because it's a funnel into Microsoft's ecosystem. GitLab is free because they hope you'll eventually pay for the enterprise version. Codeberg is free because it's run by volunteers who believe in the cause. But free doesn't mean zero cost. It means the cost is hidden — in your time, your trust, your vulnerability.
When you host your project on someone else's infrastructure, you are at their mercy. You can fork the software, sure. You can self-host if you have the skills and the servers. But most people don't. Most people just want to write code and ship it. So they accept the risk, because the alternative is too much work.
And that's how we end up here. Another outage. Another round of frantic posts on Mastodon and Hacker News. Another reminder that the emperor has no clothes.
What Should We Do?
I'm not going to tell you to abandon Codeberg. That's not the point. The point is that we need to think differently about infrastructure. We need to treat code forges like we treat power grids — critical utilities that cannot fail. That means funding them properly. That means demanding transparency about uptime and incident response. That means, when necessary, building decentralized alternatives that don't have a single point of failure.
The irony is that the tools exist. Git is inherently distributed. Every clone is a full backup. But we've layered so much centralized crap on top of it — issue trackers, CI/CD, package registries — that we've lost the forest for the trees. We need to figure out how to make those services work in a peer-to-peer way, or at least with federation, so that when one node goes down, the network doesn't collapse.
Until then, every time you push to a remote repository, you're making a bet. A bet that the remote will be there when you need it. A bet that someone else is paying attention to the server logs. A bet that you're not one outage away from a complete halt.
Codeberg will probably come back. The maintainers are good people. But the question isn't whether this particular outage will be resolved. The question is how many more outages it will take before we start taking resilience seriously.
Open source is about freedom. But freedom without responsibility is just chaos. And right now, the chaos is winning.
