Github is an amazing service to host and share any kind of code repository. I’m a big fan of github, ’cause I’m an avid user of git, and to be honest if you are not, just have a look how to get you started in 15 mins.
With the rising movement of openscience and reproducibility, the necessity for science to share their code and result is getting higher and higher.
For example, figshare is doing a great job to get the people the ability to share their own results quickly online, providing useful metrics and feedback to the uploader.
As far as we advanced in reproducibility and code sharing, in some disciplines we are still at cowboys stage, where the code is not shared and it’s
very difficult to reproduce a figure published on a paper, or at least get the code that comes with it.
One way to get over this could be to give to the scientists an easy way to play with their own code, on the safety of a private repository.
It would be great if the repo could be set up as open from day 1, but we know this is not the case for some projects. Therefore the ability to have a private repo will encourage scientists to set up their code under a VCS (git) and take confidence with the system. I know from experience and from friends that tried that there is not going back once you get the handle of it.
Now, let’s try to propose one way how github could be friendlier with academia. Right now github as an entry only for student, teacher or organized group, sitting at http://github.com/edu. This unfortunately does not cover at all the requirements of academic world,therefore I’ll take my chance and propose a very special role, which should cover most of the reqs for a single researcher.
The Researcher account should give the possibility to a scientist to get comfortable with the system.
I think the account should give the ability to create 5 time-based private repos, and the possibility to create one organization with at least 1 time-based private repo.
Let me try to explain the rational behind this numbers. First, the 5 repo are more than enough to get the people started. If they need more, they can always go to a paid plan, which it’s just fair. In second instance, the idea of having the ability to create on orgs with 1 private repo is a good idea because of the collaborations.
Scientist usually collaborate on big consurtium, and the collaboration is project focused instead of people focused, therefore the abitlity to have an
organization makes easier to share the control. Last but not least the url will be more community friendly, emphasizing the projects itself.
The time-based in front of the private, is to make clear that these repositories will automatically opensource in 5 years time. This is to encourage to opensource the repo when the paper gets written, and to help the sharing of the code. As soon on of the repo gets opensource, the researcher re-gain one repo on the total cout as private time-based one. Of course, if the research does not want to opensource the repo and wants to keep it private indefenetely, she can enroll in one of the paid github plan.
In conclusion to get more academic friendly:
- github should divide education and academic stream
- github should create a new type of account, the researcher
- the researcher should be entitled to new time-based repo
- these are going to be automatically opensourced after 5 years time
Github is, at the moment, the best website to share code and do collaboration. I hope they can take the lead and became the best way to get academics into
VCS and code sharing. They have just nominated a new educational liason, which maybe can help into bringing this issue up.
August 23, 2012 at 7:57 pm
Agreed, it would be nice to have some trial free private repos for academics. I just pay the $7 per month as it really is not that much money, but if I was a grad student I probably wouldn’t want to pay.
Having said that though, I love that GitHub is public by default. It facilitates code to be open, and we need that in science.
August 23, 2012 at 10:05 pm
that’s why I think the time-based repo makes sense.
It is to encourage people to opensource and share their code, howeve giving them time to publish it before.
August 23, 2012 at 8:16 pm
Why don’t you just use bitbucket? The entry account already comes with 5 private repositories.
August 23, 2012 at 8:51 pm
when I’ve started to use github bitbucket did not have any kind of support for git, but only mercurial.
When they actually started to support both, I think they did a very swifty move.
I thought about moving there, but a lot of my repos where already on github anyway and I manage to get one of this early thing for student with 5 repos, and I have discovered that they are pretty important and it makes sense to have them.
September 7, 2012 at 6:00 pm
Also BitBucket offers unlimited repositories, public or private, for people associated with an academic institution (i.e., with an .edu or .edu.* email address).
August 26, 2014 at 12:16 pm
I’m not that much of a internet reader to be honest but
your blogs really nice, keep it up! I’ll go ahead and bookmark your website to come back down the
road. Cheers