An [announcement](https://bitbucket.org/blog/sunsetting-mercurial-support-in-bitbucket) went out today from BitBucket, one of the largest players in the version control platform market. They let the world know that due to declining popularity, they were going to stop allowing [Mercurial](https://www.mercurial-scm.org/) repositories to be hosted with them. But not only will they stop allowing new repositories and new commits, they will completely delete all existing repositories by the middle of next year.

Of course, a private company such as [Atlassian](https://www.atlassian.com/) is allowed to make their own decisions about their web presence and what they do with it, it is within their rights. But erasing from public history all of the work contained in those Mercurial repositories should be some cause for concern and discussion.

The web upon which Atlassian bases it's services was created by public funding. It was created by joint research projects and gifted as open standards and protocols for the betterment of all. Volunteer run committees and open source projects (upon which the entire foundation of BitBucket revolves) are the lifeblood of modern software and the internet. And yet Atlassian can't see a way to host even a read-only archive of these projects, tiny fragments of the whole that is open source software. Instead they will cast these projects and fragments of history into oblivion.

This isn't the first time we've seen this on the web. Many of us remember GeoCities either fondly or with warm nostalgic embarrassment. Whole generations of us had our first experiences building for the web on that platform. And yet it disappeared in a second, as Yahoo! saw no continuing value in it. But shareholder profits are not the only measure of value in this world. That shortsightedness forgets the value in having the history of the web and the preservation of our digital legacies.

Of course, Atlassian has provided us with a grace period, and a handy list of other free and paid Mercurial hosting services. But that does nothing for [link rot](https://en.wikipedia.org/wiki/Link_rot). Tim Berners-Lee puts it very well in his [1998 article on URIs](https://www.w3.org/Provider/Style/URI). Unless there is some threat of insolvency, what harm is there maintaining links to old, static content? By forcing users to find new hosting, not only are you foisting the work of curation upon them (without even providing an automatic redirect service), but you're also allowing many stale projects die. Stale projects can come about not just when a creator loses interest, but also when a creator considers it feature complete, when they lose access to their old accounts, or even when the creator passes on. There is now no way for these projects to be saved.

What dies on June 1st, 2020, is yet another digital cultural legacy. We'll see it when some build pipelines fail because a repository is no longer available. Others will discover tutorials they're following rely on code now lost to the ether. But most of all we'll not know what we've lost, and maybe that's even worse.