From my personal experience, NIH corresponds to the need to DIY to proper understand. I often feel like I can't decide between options unless I at least built a super-basic, minimal version of the functionality. The other aspect is this idea of mine that technology is ready for primetime as soon as it can be easily reproduced. People in our field have a sense of elegance - like every problem becomes trivial once you find the best abstraction. And there is this "do one thing an do it right" idea.

How many bugs do fix by removing complexity that is unnecessary for your special case?