The big problem I've seen is being asked to do prototypes that absolutely will not be used in production, and will require rewriting half of it, rather than spe

The big problem I've seen is being asked to do prototypes that absolutely will not be used in production, and will require rewriting half of it, rather than spending an extra hour or less to do something resembling the production version. So I spend two hours, because half that time was prototyping something that's already very well known to be A Thing people can do, with only one obvious best practice everyone does, not really much that needs to be prototyped or explored. I'm talking about things like not bothering to get a JSON parser working if it's nontrivial in a certain language, but using string split nonsense and some simpler file, because managers like to see regular progress without large gaps. I've also done this kind of thing to myself plenty of times when in a hurry.