Posted Jul 30, 2010 18:02 UTC (Fri) by daglwn
Parent article: Realtime Linux: academia v. reality
Comparing and contrasting research results is almost impossible.
This is absolutely true. Half of my Ph.D. thesis was about the great difficulty of doing this. The research practice in the computing fields is broken. The hard sciences require reproducible results. We don't even require all of the assumptions to be stated. In my experience, it is impossible to reproduce experimental results from any computing paper because the experimental setup is not described fully.
Free Software could be a big help. If researchers were required to submit their code along with the paper it would go a long way to allowing others to not only verify the results but build upon them. But there's a great cultural fear of getting "scooped" that needs to be addressed before this can happen. Academic research in computing is much more competitive than collaborative and that's really the fundamental problem.
to post comments)