A lot of this hit home to me as similar to my experience in bioinformatics - where all the formats are standardised and everything should be reproducible - but almost every project ends up completely different.

Software + Research = Chaos