One thing I've learned in 25 years of software maintenance is that every time you commit crap that appears to work without being sure why it works, that's going to come back to waste a lot of your time later.