Science typically builds on the body of available knowledge – the more knowledge available the faster science goes. It’s striking when you visit other labs in person; you see all of their unpublished work, and you know that most of their results and data won’t be available to the bulk of the scientific community until a year after each particular scientific project is finished. By the time papers are in print, it’s old news to the insiders. More striking is when you visit labs whose work you’ve thought about replicating and expanding on. It’s not too uncommon to find that only one person in the entire lab is able to get the technique to work, and even for him the technique only works on Wednesdays. This type of information would be useful to know before you embark on a useless three months trying to adapt their method. But scientific publications are covered in a thick coat of high-gloss finish, making these unacknowledged difficulties hard to detect.
It’s interesting observation that chimes with my own very limited experience. There is much larger number of things that can go wrong than can go right. All the experiments in our lab benefit from the experience of our senior scientific and RA staff. And then we almost always have several pilot phases. But even then, of course, the experiments don’t always work. What’s more the failures are often nearly as informative as the successes.
Perhaps what we need is a Journal of Articles in Support of the Null Hypothesis?