My working days are spent dealing with the intricacies, finer points and frustrations of the academic peer review process - but it seems that some science journals take a rather less scrupulous approach, to the extent of not even bothering with it despite claiming to do so.
When it is implemented rigorously and reviewers truly engage with papers, offering measured assessments and constructive critique rather than brief approval or ego-driven demolitions, the process is of enormous value in sorting out the wheat from the chaff and making already sound pieces even better.
Sadly, the fact that the problem of fraudulent journals appears to be growing is probably an inevitable consequence of the ever increasing pressure to publish that quality is disregarded in favour of sheer quantity - something that suits many academics as much as it suits the journals. There's no point in crowing about the open and free dissemination of knowledge, though, if that "knowledge" is not worth knowing.
(Thanks to Shaun for the link.)