Bad apples in science
In science you occasionally get bad apples, i.e. scientists who fabricate results. Fortunately, science is a self-correcting activity in which results have to be reproducible to be accepted, and theories have to be falsifiable in order to count as science. If nobody else can reproduce a result that is obtained by only one scientist then it is erroneous, no matter how talented that scientist is. A theory that makes no predictions about things that we can measure is not science, no matter how elegant or philosophically deep that theory is.
There are some famous recent examples of bad apples in science, such as J Hendrik Schön and Luk Van Parijs, both of whom lost their jobs and no doubt their careers as a result of their reckless dishonesty. The self-correcting nature of science worked for us by exposing these rogues, but the fall-out is that people trust scientists a bit less than they did beforehand. I wonder how many more as yet undiscovered cases of fraud there are out there.
What I don't yet understand is why would any scientist think that they could get away with fabricating results? They must know that other scientists will be attempting to reproduce their results. They must know that they won't receive any long-lasting recognition until other scientists actually do reproduce their results. No Nobel prizes have ever been rashly awarded because someone suddenly produced a surprisingly impressive result.
Is it possible that some scientists can so completely delude themselves that they really believe their faked results? I know self-delusion can easily happen, but usually a good night's sleep and a pause for reflection cures the problem. Maybe some people can't reset themselves in this way.
Is it possible that the demand for a continuous stream of publications forces some scientists into grey areas where they publish what they think their results would be if only they had the time to do the work? I am very familiar with this pressure to publish according to a schedule that is determined by "bean counters", and it isn't nice.