How can we fix the PTO, allowing examiners to effectively distinguish between patentable and unpatentable inventions, without slowing the process to a crawl or wasting a bunch of money? This essay reviews the recent literature and considers a number of proposals and their problems. It concludes that we can make the PTO better, but are unlikely to solve the problem of bad patents altogether. Rather than focusing on adding money, the focus must increasingly turn to understanding and changing applicant and examiner incentives.