[dkirkman]

The problem is that in terms of getting jobs, promotions, and tenure, producing code has no value. Except for the people that actually write it, "programming" is generally considered to be a trivial task. So if you invest time writing data analysis code, the only way you can get a return on that investment is to keep it proprietary to your group and use it to get results before your competitors do.

We can't expect source code publication until we change the perception that it's trivial. If published code were as valuable as refereed papers for career advancement, the problem would be solved instantly. But if that change does not occur, requiring that analysis code be published will just stop the development of any new code (at least by career rational people) and probably hurt the field more than it helps.