is certainly an incredibly expensive problem, even in single cases;
http://www.space-travel.com/reports/Boeing_Patent_Shuts_Down...
And I am pretty sure, even ignoring broadness as a factor and just sticking to patents granted despite prior art, that if there were only a few dozen patents causing trouble then the companies like Article One, who is getting patents revoked in the hundreds due to discovery of prior art, would not be able to exist while paying out millions in discovery fees to their researchers.
[edit] - a look at the problem of workload by a USPTO Patent Examiner;
I'm not counting any undesirable patents that are causing problems because patent terms are too long, or those that were granted because patent law/precedent/policy didn't give the examiner any cause to reject it, etc.
My point is only to refute the charge that examiners 'sloppiness' is common and/or a notable problem.