The U.S. patent sys­tem is bro­ken. In the old days, the goal was to pro­tect the small-time inven­tor from get­ting ripped off by big com­pa­nies. That would seem­ingly encour­age inno­va­tion, which is regarded as a Good Thing. For decades how­ever, the sys­tem has shifted to favor the deep-pocketed com­pa­nies, who have dis­cov­ered that with enough legal wran­gling they can gen­er­ally get util­ity patents on any­thing they want. Such patents are assigned a the­o­ret­i­cal value and become bar­gain­ing chips. So that’s a bum­mer, but not the end of the world.

Lately, a num­ber of hos­tile players–patent trolls–have taken to the scene. These are indi­vid­u­als, com­pa­nies, or even shell com­pa­nies who are nei­ther inven­tors nor pro­duc­ers of goods. They sim­ply buy the rights to patents and sue the shit out of peo­ple who may or may not be in vio­la­tion. Most defen­dants set­tle because the cost of pur­su­ing a legal bat­tle over arcane and tech­ni­cal mat­ters is pro­hib­i­tive for all but the most well-funded com­pa­nies. (For more infor­ma­tion, see exhibit A and exhibit B.)

I pro­pose a sim­ple three-step fix to the system:

1. Elim­i­nate soft­ware patents entirely. This was a really bad idea. In soft­ware, last year’s novel inven­tion is this year’s stan­dard indus­try prac­tice. Does the patent you got for an Inter­net shop­ping cart in the ’90s enti­tle you to bil­lions of dol­lars of roy­al­ties today? No. And how can you really prove when some­one is in vio­la­tion? Spe­cific exe­cu­tions of soft­ware con­cepts are pro­tected under exist­ing copy­right law. What hap­pens under the hood is just dig­i­tal plumbing.
2. Pro­hibit “non-practicing enti­ties” from own­ing patents. All patents should be assigned to either the actual inven­tors or to a com­pany actively engaged in the busi­ness. Any other own­er­ship is not in the best inter­ests of society–remember that we want to pro­vide incen­tives for only the cre­ation and exe­cu­tion of ideas.
3. Elim­i­nate jury tri­als for patent cases. All patent lit­i­ga­tion should be resolved by a judge assisted by a panel of experts drawn from indus­try and acad­e­mia. I’ve served on a jury of my “peers,” and I can cer­tify that nobody on that panel would have been qual­i­fied to eval­u­ate a sub­tle ques­tion about pro­gram­ming, engi­neer­ing, or pretty much any tech­ni­cal sub­ject. It pains me to imag­ine what the jury had to say about the inner work­ings of Java APIs dur­ing May’s Google-Oracle trial.