Anything available to the public can be used in a prior art rejection, not just patents. (I'm a former patent examiner and I frequently used "non-patent literature" as it's called.)
I wasn't around long enough for many to get to the point where they could be granted. I think I only granted around 3 or so. And those ones were unlikely to cause anyone trouble.
I've heard a second-hand story of one examiner who granted a patent that was later strongly criticized. They specifically requested more time from their supervisor to work on it, but their supervisor denied the request. Since they didn't have any valid reason to reject the application, they allowed it. The examiner suspected that it wasn't valid, but they really didn't have a choice as they have to meet a quota and "docket management" requirements. So it's not a matter of an examiner using their own discretion. The examiner knew it was likely a problem but they couldn't find any evidence which was legally enforceable.
Why do patent examiners sometimes grant invalid patents? For the same reason software developers sometimes release buggy software: Management doesn't give them enough time or resources to do a quality job.
