Fun thought experiment:
What if the patent office was only allowed to issue a certain number of patents per year, and was directed to pick the most worthy ones.
That hasn't been true for several years, and the PTO has been very tough on software patents since the Supreme Court decided Alice v. CLS. The PTO is currently rejecting applications that fail the "machine or transformation of matter" test, although sometimes they can be argued out of a rejection. The Alice decision is rather unclear, which annoys everybody.
The "patent troll" problem isn't that big. The EFF's data indicate that almost all the bulk infringement claims come from three lawyers in the Eastern District of Texas.
But there are strong forces, that are pulling the system into the wrong direction. Big corporations, patent lawyers and investors want the overstretched system, because it is good for their profit.
It is good for dinosaur corporations that want to defend their "claim" against smaller and more agile competitors.
Also the numbers of patents are presented every year to prove, how grand the nation is. Nobody wants to spoil that.
(BTW: That would be a real patent reform)
Not exactly, the USPTO is funded by the patents it examines and keeps active (maintenance fees). Issuance fees are actually small compared to the rest of the fees involved in examination. Additionally an issuance happens only once, but rejections are unbounded. If making money was an incentive, they'd do much better by repeatedly rejecting applications and forcing applicants to keep paying for continued examination rather than issuing them.
But I don't think, that repeated rejections would make the patent office happier. First of all, they would create much (political) pressure from the big corporations and second if they reject to much, the number of patent applications would decrease.
Third, they would cash in less maintenance fees.
Maintenance fees are quite a bit higher, but they are bounded (only 3 times over the life of a patent). Also, they are not guaranteed, because often a patentee finds out that the patent is not worth much due to the vagaries of the market and abandons it. However, it would be interesting to model this as a maximization problem given enough statistics about the probabilities of applying, continuing prosecution and abandonment...
In any case, this is all academic because the PTO gets to keep only flat amount of the money it brings in. Everything extra goes back to the Fed. Money is really not an incentive for the PTO.