There is a difference between wholesale making things up and conveniently leaving out variables that alter analysis of your data. I do agree overall that this is not trivial and there are tremendous disincentives; that is evidentally not a good tactic to reproducing experiments in the general case.