The justification for patents is a bargain: a temporary monopoly in exchange for disclosure, which is supposed to buy more invention than the monopoly costs in restricted use. Whether the bargain is a good one is an empirical question, and the empirical literature is considerably less favourable than the institution's security suggests. Industries differ enormously — pharmaceuticals, where a molecule is cheap to copy and expensive to test, look nothing like software, where products combine thousands of components each potentially encumbered — and a single regime applied across all of them will be badly calibrated for most. Historical episodes in which patent protection was absent or unenforced show invention continuing at rates that are hard to reconcile with the strong version of the case. The system is defended today largely by those it enriches, which is not an argument against it and is worth noticing.