Sunday, November 18, 2012

SCU Software Patent Conference Recap

Friday was Santa Clara's Solutions to the Software Patent Problem conference, which I've previously blogged about, and I was able to follow some of the panels from New Haven via the live stream. Here's a recap, based on the parts I watched online and the detailed #HTLI tweets—thanks to all the tweeters who made this possible!

Keynote #1: What is the Problem?

Richard Stallman said software needs protection from software patents (or "computational idea" patents), and he proposed a safe harbor for software on generally-used computer hardware. The only good computational patent is a dead computational patent, he said. Kent Walker said there is a patent quality problem, a litigation problem (with internet patents litigated nine times more), and a troll problem (90% of cases settle, but when they do go to court, trolls lose 76% of the time, and they cost $29-80B). Patent litigation is a tax on innovation. Walker said we need to stop bad patents from issuing, weed out existing bad patents, and have clearer rules for damages and awarding costs. An attorney from Intellectual Ventures asked, if software patents are such a problem, why doesn't Europe dominate software since they don't patent software? Walker said US dominance came in the 90s before meaningful software patents.

Panel #1: Legal Reform, Part 1

James Bessen said we've had restrictions on patenting abstract ideas since 1972, but money trumps precedent, and lawyers have played word games to get software patents approved. In 2011, the straight legal cost of defending troll suits was $7B, and the deadweight loss from suits was $70B. He proposed adjusting maintenance fees to deter NPEs (i.e., "make pollutors pay" through a Pigovian tax). Colleen Chien explained why TRIPS Article 27 ("patents shall be available ... in all fields of technology") does not prevent industry-specific software patent reform: TRIPS allows exceptions (like our surgical methods and tax strategy exceptions) if they are narrow, well-defined, and don't prejudice legitimate interests. Brian Love proposed a patent term reduction: because most NPE suits are filed at the end of the patent term, even a 3 year reduction would impact 60% of patent troll claims. And while getting a term reduction is not politically feasible, we could effectively do it with higher maintenance fees. Christal Sheppard said Congress can affect patents by simply passing a "Sense of Congress" resolution. The audience poll for "whose solution do you like best?" was close, with Brian Love at 37% and Colleen Chien at 32%.

Panel #2: Agency Reform

Peter Menell argued that the problem is unclear claims and suggested requiring patent applicants to fill out standardized templates to improve patent clarity (and this solution won the audience poll). John Allison said we shouldn't single out software patents and suggested coming up with techniques to make it more costly to game the regulations. Arti Rai said software should learn from bioinformatics and focus on getting rid of patents that don't satisfy written description or definiteness; she thinks the enablement standard is too fuzzy to be useful to examiners. We don't need new law; we just need to enforce 112. Michael Risch said poor examination led to bad patents, but the problem will subside; we just need to tweak the current rules. He noted that coming up with the right obviousness standard is difficult because for software, the hard part is identifying the problem to be solved. He also said that enablement is not a solution: you can't simultaneously claim that everything is obvious and that everything is not enabled. Christina Mulligan noted that even if all the patents are great (satisfying obviousness, 112, etc.), there is still a problem because no one can find the patents they might be infringing. She argued that we should reduce the costs of accidental infringement by (1) having an independent invention defense; (2) making it more difficult for NPEs to recover; and (3) having lower reasonable royalties.

Keynote #2: Views from the Trenches

Caroline Dennison (USPTO) announced that Michelle Lee, formerly of Google, will be the director of the new PTO Silicon Valley office. She also said the PTO will be soliciting comments on improving claim clarity. Hon. Edith Ramirez said the FTC is concerned about use of the ITC as a way of getting around eBay, and that they are urging the ITC to use a public interest test to determine when injunctions are appropriate. They are holding a workshop with the DOJ on December 10 to discuss patent assertion entities (PAEs), which will be available as a webcast. Hon. Paul Grewal emphasized that from a trial judge's perspective, the when of all these questions is as important as the what; he asked folks to consider how different reforms fit into a trial schedule.

Panel #3: Legal Reform, Part 2

Mark Lemley discussed his solution to software patent overclaiming, which I've blogged about previously; he also won the audience poll for this panel. John Duffy (who came in second in the poll) said most of these solutions are software-specific kludges (a characterization Stallman objected to during the Q&A), and that a more elegant solution is to use obviousness to restrict patents based on Graham's inducement standard (only allowing patents on "those inventions which would not be disclosed or devised but for the inducement of a patent"), an idea he and Michael Abramowicz described in their excellent Yale Law Journal article. Ted Sichelman suggested purging patent law of "private law" remedies, arguing that "make-whole" remedies are often unnecessary to incentivize innovation. Samson Vermont said patent law is out of touch with its purpose, which is to optimize the enjoyment of operable inventions, not to promote invention for its own sake. He argued that infringement should be excused when the patentee does not practice the invention, when the infringer is an independent inventor, and when there is a notice failure in the relevant field. Commenter Suzanne Michel of Google said she likes Lemley's idea and that it was also independently invented by someone at Google—another example for his Myth of the Sole Inventor? She also loves Duffy's inducement standard for obviousness. Heidi Keefe argued that clients need to be willing to stick their necks out for legal changes, even if it means waiting for the Supreme Court to reverse the Federal Circuit. She also argued that it is important to explain the ecosystem to jurors, including that high damage awards will come back to them as higher prices.

Panel #4: Self Help

I didn't watch this panel and the tweets were more confusing than for other panels, but it appears that Keith Bergelt discussed defensive publication of prior art, Dan Ravicher discussed public interest lawsuits, Wendy Seltzer discussed royalty-free standards, and Jennifer Urban (who won the audience poll) discussed defensive patenting. Twitter's Benjamin Lee said they've fought every lawsuit over the past six years and haven't paid a cent to an NPE. Brad Burnham of Union Square Ventures said 50% of the companies in his portfolio have been threatened by patents. He said the independent invention defense would have covered every suit they are facing, and Mark Lemley's functional claiming proposal would also help.

Keynote #3

Julie Samuels (who won the poll for favorite keynote and also deserves credit for dubbing the event the #patentprom) emphasized the importance of community building around these proposals, especially post SOPA/PIPA when policymakers are listening to the tech community. Although Congress is nervous about TRIPS, they shouldn't be. Pam Samuelson said there is no silver bullet because there are so many aspects to the patent problem.


  1. I still cannot understand the popularity of EFF propositions while they have been described as a major turn around, implying acceptance of software patents. Comments on their proposition are very telling. There is now a momentum in US to get rid of the threat of software patents altogether, partial solutions are just a waste of time and resources.

  2. Anything someone can implement in hardware, I can implement in software, and vice versa. Although "software patents" are the favorite hobby horse of a group of folks, software patents are not meaningfully unique from other patents. IBM was getting software patents in the 1970s, when software was not considered patentable, by writing the claims in terms of memories and processors. Unless you want to eliminate all patents (as Stallman certainly does) advocating against software patents as a whole is not feasible. All you would be doing is raising form over function.