Part of the problem (as I understand it) is that when patent examiners reject a patent application, they are required to give a reason from an approved list of rejection reasons. The "this already exists" reason requires the examiner to cite some sort of approved publication which describes the invention. Unfortunately, the list of approved publications is essentially "existing patents and nothing else." The patent office recently (less than a year ago) started a wiki to allow experts in certain fields to document the state of the art in that field.

Another part of the problem is that patent examiners are required to examine X number of patents per unit time. Unfortunately, when a patent applicant resubmits a rejected application, the patent examiner does not get credit for examining the patent a second time. So an applicant can troll the patent examiner by repeatedly modifying the application. The examiner becomes so overwhelmed with resubmitted applications that he or she has no choice but to approve the application in order to meet quota.

The result of these rules is that a patent examiner effectively can't reject a patent application unless someone has already tried to patent the exact same thing. Or, "blame the game, not the player."