I strongly disagree. For the types of code you mention we already have avenues for this -- academic papers and publications and libraries. Publishing code, say

I strongly disagree. For the types of code you mention we already have avenues for this -- academic papers and publications and libraries. Publishing code, say, in a GitHub profile, can serve other purposes though. Sometimes you're applying for a job and the employer just wants to assess that you can write syntactically valid, human-readable, not-too-terrible code and use source control. For this, I'm not too terribly concerned about whether this is the millionth time solving the problem, so much as I am about excluding people who aren't ready to be interviewed by a bona fide developer.