>>13726745I've work in the sciences and have ~5 first-author publications and I'm on a dozen or so papers, and you hit the nail on the head pretty much.
I work in cell biology, and my phd lab went through this cycle a few times: when I want to replicate an experiment from a paper, it didn't work. Then I would email the PI/first author and ask for details. Most were nice and would email me their protocol they had in their lab notebook, and it would look absolutely nothing like their published methods, and they ALWAYS had this "one weird trick" (put it in the -80 after this step for 5 minutes; double the salt for this buffer; run it at this voltage for 10 minutes before the protocol voltage) that wasn't even in the protocol, just a trial-and-error they figured out and is written in their notebook. I understand it somewhat; you really care about getting the science itself out, and methods are the biggest PITA to write out, and its very easy to forget steps if you are writing about 6months-2 years of work.
I've personally seen pressure to publish lead to some sketchy stats and results by a friend of mine in another lab; all your experiments show negative results and you need the first author publication so you can graduate in the next six months; or the most common one I've advised against but they went ahead anyways with anyways: 5/6 experiments show your result is true, the last experiment disproves it/doesn;t line up. So sketchy stats come in to save the theory and everyone is happy (protip: you can't keep adding n until you get significance and then stop).
It's part why I left science, most papers are total horseshit. Maybe 1/1000 are worth reading.
