The problem I've run into over and over with research code is fragility. We ran it on test A, but when we try test B nothing works and we have no idea why becau

The problem I've run into over and over with research code is fragility. We ran it on test A, but when we try test B nothing works and we have no idea why because god forbid there is any error handling, validation, or even just comprehensible function names.