I've written before that I believe academics should share their code (<a href="http://news.ycombinator.com/item?id=3845070" rel="nofollow">http://news.ycombinator.com/item?id=3845070</a>).<p>This greatly improves repeatability of results, and saves other researchers a huge amount of time.<p>My idea---which unfortunately I came up with towards the end of my academic career---was to create a public github from day 1. Because it made me more conscious that other people would be able to see my code, this is a vast improvement in usability over deciding only halfway through that the code would be public.<p>Unfortunately, many academics are scared of this approach. Sharing their earliest thoughts and meanderings makes them feel exposed, and strips away the veneer of unassailability that only showing finished work gives. And I will admit that I deleted repos that of work that was failure to find.<p>One potential solution is conferences that give priority to researchers who share their code. (I've seen this occasionally, but can't remember where right now.)
This mostly has to do with Life sciences. Most other fields have embraced open publishing and data collaboration, with remarkable results. Life scientists stubbornly insist on defending their tiny castles and not changing the rules with which funding is given.