Academia should adopt code reviews. They are totally frictionless today and are the best learning tool I know of for a developer.
6 plus ones
Shared publicly•View activity
View 7 previous comments
- Mostly other students, but I would have been super happy to get the occasional code review from my advisor. Code is not that hard, and small enough reviews are super quick to do.Jun 18, 2012
- Academics should not even think about code quality :) Nobody cares about their code anyway. Is code review nice? Yes. And I use them all the time. Is it worth in a deadline-based science field? No. Do reviewers care about their code? No. The code review for academics is called paper. :)Jun 18, 2012
- Producing good code is motivated by the fact that good code gets used, and therefore, generates citations. The incentive is already there.
In a deadline-based science field, code can make or break a project. Random bug in your evaluation framework produces erroneous results? Oops. Paper redaction.
In a deadline-based science field, working with messy code slows you down in the long run.
A paper's results are often not reproducible unless the author also publishes their code, dataset, features, etc..
You can even do self-reviews. Just looking at the diff that you're about to commit can help you detect bugs that you are about to introduce. And you can run scripts to confirm that your code conforms to a particular style guide.Jun 18, 2012
- About academics not being ready for code review... what would it take to get them ready for code review? It's easy, it's useful... what more are they waiting for? (Edit.. I'm legitimately curious about what the barriers are.)Jun 18, 2012
- It has been a long time since I was in grad school, but it was common practice to pass around papers to other students/professors for review. So, it was low-tech compared to how we do code reviews today, but the essential elements were the same.Jun 18, 2012
- Jun 18, 2012
Add a comment...