4 ms·

As I was reading through the section on the difficulty of catching cheaters, I got to thinking whether the process of submitting a paper might be better done th

by jmknoll 10y ago

As I was reading through the section on the difficulty of catching cheaters, I got to thinking whether the process of submitting a paper might be better done through something like GitHub, where you then have a record of the person's commits, whether that be new sources found, time spent researching, a section written/edited. It then allows the professor to grade the process, not just the result. I mean, when I'm hiring software developers, and they send me a link to a site that they made, i find that basically meaningless, because it doesn't tell me how they work, when/where they decided to incorporate external resources, and how they collaborate. It also doesn't guarantee that they actually made the site. A link to a github repo doesn't automatically solve all of those problems, but it solves a lot of them, and it would be almost as much work to falsify well as it would be to just do it correctly.

ben_jones 10y ago
I don't think it would be difficult to falsify. How thoroughly do you go through the commits of candidates? My guess is it's only a cursory glance which would be pretty easy to fool. It makes me wonder though, could a pull request contain changes to previous commits to a project and thus "steal" credit from other authors?
gengkev 10y ago
For papers, Google Docs basically does this — except that the revision history isn't as accessible as it is with Git. I think this would be a good idea to implement, but it still doesn't solve the main problem that this article presents, which is contacting out the entire research/writing process itself.
divbit 10y ago
If you don't mind people seeing your occasional stupid ideas that don't pan out, then yeah, it works fine: https://github.com/divbit/phd-thesis https://github.com/divbit/phd-thesis edit: as an addendum to this, overleaf has built-in git integration, which I found to be useful

by jmknoll 10y ago

ben_jones 10y ago

I don't think it would be difficult to falsify. How thoroughly do you go through the commits of candidates? My guess is it's only a cursory glance which would be pretty easy to fool. It makes me wonder though, could a pull request contain changes to previous commits to a project and thus "steal" credit from other authors?

gengkev 10y ago

For papers, Google Docs basically does this — except that the revision history isn't as accessible as it is with Git. I think this would be a good idea to implement, but it still doesn't solve the main problem that this article presents, which is contacting out the entire research/writing process itself.

divbit 10y ago

If you don't mind people seeing your occasional stupid ideas that don't pan out, then yeah, it works fine: https://github.com/divbit/phd-thesis https://github.com/divbit/phd-thesis edit: as an addendum to this, overleaf has built-in git integration, which I found to be useful