Patents are premised on a fundamental bargain: in exchange for exclusive rights, inventors must disclose how to make and use their inventions. But the patent bargain is eroding. This Article shows that patent examination and litigation overwhelmingly concentrate on whether an invention is new, while largely overlooking whether the patent’s teachings are practically usable or scientifically credible. When disclosure issues arise, they typically focus on what we call “scope-evidence mismatch,” which occurs when an applicant fails to provide any support for part of the claimed invention. In contrast, scientific peer review—the other significant context in which technical disclosures are scrutinized—routinely surfaces disclosure problems with the support itself, like flawed methodology or data analysis. It would be surprising if these defects in scientific manuscripts were absent from patent applications.

This Article surfaces the disclosure issues that peer review focuses on and that patent law neglects. Drawing from doctrine and scientific practice, we develop a taxonomy of disclosure breakdowns beyond the “scope-evidence mismatch” frame—aspirational disclosures, defective teaching, and unreliable support—and show how each maps onto existing but underused doctrinal hooks. We then present new mixed-methods evidence from interviews, public records, and examination and litigation decisions to explore why the patent system under-polices these problems. Examiners face severe time pressure, limited training, and weak incentives; litigation rarely corrects these errors because technical disputes are hard to resolve pretrial and difficult to integrate into a coherent trial narrative.

The stakes of these disclosure failures are rising. Modern innovation depends on data- and biology-driven techniques that require complex and validation-sensitive disclosures. And generative artificial intelligence tools are making it feasible to flood the system with plausible-sounding but erroneous patents. The lack of a robust disclosure screen means not only that examiners mistakenly grant patents that fail to deliver the promised teaching, but also that these poorly supported patents can improperly block later inventors who do meaningful inventive work. The result is a patent literature that often fails to educate patent readers or ensure that patent rights are limited to what the inventors actually contributed, increasing costs for follow-on researchers and the public. Drawing from our diagnoses, the Article concludes with concrete reforms to strengthen patent disclosure and restore the patent bargain.