Now here's a disturbing case: research sabotage. It involves a (former) postdoc at Michigan:
(Vipul) Bhrigu, over the course of several months at Michigan, had meticulously and systematically sabotaged the work of Heather Ames, a graduate student in his lab, by tampering with her experiments and poisoning her cell-culture media. Captured on hidden camera, Bhrigu confessed to university police in April and pleaded guilty to malicious destruction of personal property, a misdemeanour that apparently usually involves cars: in the spaces for make and model on the police report, the arresting officer wrote "lab research" and "cells". Bhrigu has said on multiple occasions that he was compelled by "internal pressure" and had hoped to slow down Ames's work.
The student's account of what happened (later in that linked article) is creepy and compelling. Things started going wrong with her experiments, one after the other. At first she couldn't figure out what was happening, then the suspected her own mistakes, but ultimately (like the man who furnished the title for this post), suspected sabotage.
What tipped her off, apparently, was the the same sorts of things went wrong over and over - and when one was fixed, something else would appear. Lanes looked switched on her Western blots, which turned out to be because the labels had been switched on her cell cultures. That happened a few times, then when she switched the labeling system to something that couldn't be messed with, contaminants started showing up in her media. Running experiments in another lab late at night showed that they were working the way they should - when something (someone) wasn't messing with them.
It would certainly take a while for sabotage to become a working hypothesis - after all, there are a lot of ways for things not to work. But this case seems to have been helped along by the crudity of the tampering. Even so, there was suspicion that the grad student herself was trying to blame someone else for her own failures. The university's public-safety officers put her through interrogations before they would go as far as installing cameras in her lab. But those cameras caught the post-doc messing around in the lab fridge in the hours before yet another experiment went awry, and he confessed when confronted.
How often does stuff like this go on? To be honest, I'm surprised that there isn't more of it in academic labs. The competition between individuals is much more fierce than it is in industry (where people tend to work much more in large teams), and frankly, there are more unstable personalities in academia than there are in industry as well. At the same time, this is a thoroughly nasty thing to do, striking right at the basic workings of any research lab. You have to be able to reproduce things, of course, and you have to trust that the reagents and equipment are going to allow you to do that.
Most of all, in science we have to take the word of others on trust pretty often. Experiments are always out there to be reproduced, but you really can't do that to everything, every time. It's just impossible. When someone says that they got a particular reaction to work, or protein to express, the default setting is to assume that yeah, they probably did. If you have to reproduce it and there's trouble, well, then you start checking things out step by step. But there's no way science could work if you automatically assumed that everything in the literature or in every presentation was probably a lie.
And there's no way it can work if someone's going to sabotage experiments, either. I have been around two or three situations in my career when there was a suspicion of this happening. For most of the cases I'm pretty sure that this wasn't the explanation, but in one other (which I was the most removed from) I still don't know. That one got me to thinking, though, about how terribly easy it would be to do such things. As I said, this case was pretty crude, but there are many, many more subtle ways of messing things up. Some of them would be quite hard to detect, but would definitely indicate foul play if they were, and some of them would still be obscure even if tracked down. Truly excellent sabotage, though, would require as much work as generating real results.
I'm not going into details - any scientist with sufficient imagination can think of such things (homo sum, humani nil a me alienum puto). But it is interesting, when you do that thought exercise, how strange it feels. You can see how it would be done, you can see what would motivate someone to do it, but it's something of a relief to find out how little thought you've given to the mechanics of it all.