>most of the time, when you are doing an experiment you never done, you just fuck it up.

"Experiments are supposed to be replicable. The authors should have done it themselves before publication"

That seems to imply the experiments aren't even replicated by the people who are supposed to know how to get it right.