Here's an issue we don't talk about enough. Every year, peer-reviewed research journals publish hundreds of thousands of scientific papers. But every year, several hundred of those are retracted — essentially, unpublished. There's a number of reasons retraction happens. Sometimes, the researchers (or another group of scientists) will notice honest mistakes. Sometimes, other people will prove that the paper's results were totally wrong. And sometimes, scientists misbehave, plagiarizing their own work, plagiarizing others, or engaging in outright fraud. Officially, fraud only accounts for a small proportion of all retractions. But the number of annual retractions is growing, fast. And there's good reason to think that fraud plays a bigger role in science then we like to think. In fact, a study published a couple of weeks ago found that there was misconduct happening in 3/4ths of all retracted papers. Meanwhile, previous research has shown that, while only about .02% of all papers are retracted, 1-2% of scientists admit to having invented, fudged, or manipulated data at least once in their careers.
The trouble is that dealing with this isn't as simple as uncovering a shadowy conspiracy or two. That's not really the kind of misconduct we're talking about here.