Why I started using Github

So, Google are closing down Google Reader, their web-based RSS feed aggregator, in July. This announcement is somewhat timely for me. I don't use Google Reader, and never have, but a little while ago I mentioned that I might write about my recent decision to start using Github. This thing with Google Reader actually ties in rather well with some of what I wanted to say about that, surprisingly.

I should start by saying that this is an explanation of why I started using Github as opposed to setting up my own repositories and web interface on my own servers. It's not about why I started using Github as opposed to any of the other websites offering similar services, which is a different and far less interesting matter, at least for me right now.

Modern web culture is for people to have a large number of accounts (I imagine a double digit number is not uncommon) at all sorts of "cloud"/SaaS providers. Their mail and perhaps their calendar is at Google (previously their feeds were!), their blog is at Wordpress, their photos are on Flickr, their bookmarks are on Delicious, and other parts of their life are all over the place: wish lists, to do lists, lists of stuff they've read, lists of stuff they've watched, records of dieting and/or exercise progress, and a whole host of other things. I'm no exception in having a lot of these accounts, but in recent years I've been trying to use them less and less, and I've slowed down the rate at which I acquire new accounts drastically. I think long and hard before signing up anywhere these days.

I could (and probably will) write a proper article one day on the various thoughts that have lead me to embrace a philosophy of minimising my use of third party web services, because there are a lot of them. But the Google Reader situation illustrates one of them nicely, and that's the inherent lack of certainty or control with regards to the future of these services. Free web services can be taken down at any time, they can be sold to other companies at any time and fundamental changes to what they do and how they work can be made at any time. As a user are absolutely powerless in the face of this. Another high profile example, perhaps even worse than the Google Reader case, would be the aforementioned bookmarking site Delicious. This was quite a popular and well-respected website a few years ago, but then Yahoo sold it to AVOS. AVOS made wide-sweeping, drastic changes to the whole thing, which were widely considered a complete disaster. A large proportion of the users hated the changes, many of them lost data they'd carefully entered over years, and a whole lot of people left, leaving Delicious a mere shadow of its former self (I see apparently it's relaunched itself yet again just this January - not sure how that's working out). This situation is very different to either running your own web-based software if you have the skills and a server, or using locally run desktop applications if you don't (I'm terrified that in ten year's time I'm going to have to explain that concept to young people who've never known it - natives of the cloud?), where nobody can forcibly uninstall programs you are using, and where you can continue to run an old version of a program if you don't like the latest changes.

Before signing up to anything new on the web, it's rational to ask (among many other things): will this site allow me to easily export all the information I put into it, at any time? Are there other websites that offer similar functionality, and can I easily import my data into them if this site disappears or changes for the worse? If there aren't, can I be certain that I won't end up using this site in such a way that I would be seriously inconvenienced or upset if it suddenly disappeared or stopped offering some of the features it offers now?

In the case of Github, answering these questions was actually pretty easy. Because git is a decentralised revision control system, they're not hosting anything that I don't already have copies of in my working directories (except issue reports, but those aren't super precious), so even if all of their hard drives failed tomorrow I wouldn't lose anything. Because git is a widely used system, with open source tools, there are a large number of Github-like sites (such as Bitbucket and Gitorious) which I could easily move my data to with very minimal fuss if anything ever happened to Github. So my risk of data loss or lock in are both basically zero. Something else I always consider, but won't talk about at length here for the sake of brevity, is privacy. This is usually a pretty big deal for me, but in the case of Github it's, refreshingly, a non-issue - I'm only using it to store free software I write, which is a case where I actually explicitly want a total lack of privacy, because that's the whole point! I also really like that Github appears to let you genuinely and irreversibly delete your account (and this option is easy to find, rather than being buried deep in menu hierarchies or fine print). Of course, there's no way you can verify they actually delete it instead of just permanently making it inaccessible from the web and keeping it to themselves (though, again, this is free software so I don't care), but it's a rare thing these days to find somewhere that will even offer the pretense of genuine account deletability. Normally the best you can do is "disable" your account temporarily. This can be reversed at any time, which means your data is basically kept forever. Very often there's not even a way to do that, and all you can do is abandon the account. I really don't like this, for several reasons I'll write about another time.