The Problem with Putting All the World’s Code in GitHub (2015)
wired.com
wired.com
This is also the reason that GitHub is fundamentally different from source forge. If GitHub were to turn evil tomorrow, everyone could just upload their project to gitlab or bitbucket and carry on.
It's much easier for people to contribute to projects now, and a lot of projects might not have grown as quickly without it. The central directory also makes it a lot easier for other people to find stuff which probably also plays a role.
I doubt Github will go down the same path as Sourceforge.
You might not lose the latest code, but you'd lose the history and the commit comments. Those are significant.
With github you'd lose stuff like issues, but I find those not nearly as useful as the commit history.
As I understand it, the main benefits are still risks though (e.g. central directory). The code itself isn't really at risk, especially with signed commits.
Seriously, tho. The code history is the important part and DVCS provides plenty of redundant offline backups.
If people want issues saved offline, GitHub provides an API that makes it relatively easy to pull down the issues.
It wouldn't take a monumental effort to create an offline client.
Maybe if GitHub added some metadata to their models/api, it would be possible to diff the changes, cache data offline, and reduce requests.
Just don't version issues because recording all of the changes over time isn't worth saving.