The alternative I favor is adding an element to infringement that requires some evidence of copying, reverse-engineering, or appropriation. That's an interesti

The alternative I favor is adding an element to infringement that requires some evidence of copying, reverse-engineering, or appropriation. That's an interesting suggestion, but what are the odds that such evidence will come to light when one of those things occurred? Is the discovery process reliable enough that we'd want to require patent holders to depend on it? After all, in such a case, the infringer knows they're infringing and can expect a lawsuit -- it doesn't seem like it would be that hard to destroy any evidence of reverse engineering in advance. Here's another alternative: require that the plaintiff in an infringement case show objective evidence that their invention was nonobvious. Basically, this means admitting that the PTO is only a first-level filter and removing the presumption that a granted patent is valid; the patent holder would have to defend their patent in order to litigate it. Along with that, we put some teeth into the nonobviousness requirement by requiring some evidence along the lines laid out in Graham v. Deere: commercial success, long-felt but unsolved needs, or failure of others. (In software, publication of a paper in a recognized journal or conference proceedings would probably be a common form of evidence.) Like your proposal, this would be a huge change that would make most extant software patents worthless. But it doesn't depend on discovery to determine malfeasance by the defendant. What do you think?