Scientists often get things wrong, at least if they're doing their jobs. That's because they're investigating things they don't understand, and in so doing they have to try out different answers. Many answers may be partially right, or perhaps a given matter may have more than one completely right answer. But there are at least an equal number of answers that are flat-out wrong.
This is OK, because each wrong answer is a possibility crossed off the list and one step closer to the right answer. The wrong answer may rule out a whole area of possible answers, too, taking the search several steps ahead.
It's also possible that the experiment which found the answer wasn't properly conducted. Human beings are the ones who do experiments, even with mechanical and computerized help, so their fallibility comes into play.
The only problem, the article notes, is when scientists act just like everyone else and won't admit they're wrong or that someone made a mistake somewhere. The dimensions of a person's head tell us nothing about that person's intelligence. The Earth and the other planets of the solar system orbit the sun, not the other way round. Light doesn't propagate via luminiferous aether. And yet, all of these ideas have been at one point or another in history been taken as accurate descriptions of things in the world around us by sober and sane people who relied on the best experimental knowledge they had.
If an experiment suggests a conclusion, but repeated experiments don't back it up, then something's obviously wrong. Yet some folks, even behind the rational armor of the white lab coat, won't admit that, the article says, and choose to select the result they prefer. That way, we have seen by experimentation aplenty throughout human history, lies madness.