Comment by yiyus

> Code used to produce published results absolutely should be accessible.

Although I agree with this sentiment, I think that if this was a requirement we would not get much more code but much less papers.