Why bad scientific code beats code following “best practices” (2014) | Hacker News