Questions About "Scaling The Patent System" Answered
Professor Mark A. Lemley was briefly mentioned in the following article by Forbes' Timothy B. Lee on patent infringements.
Sam Bell asks a couple of good questions about our new patent paper:
“(1) Does the patent office itself not have an index of some sort? If not, how do they check applications against existing patents? (Surely there are not more patent officers with the time to do laborious discovery than patent lawyers with the time to do laborious discovery).
There are two points to make here. First, there’s a significant difference between trying to find patents infringed by a particular product, and trying to find prior art for a particular patent. In the latter case, your starting point is already a textual document, and so you can use machine learning techniques to try to find other documents that appear to describe the same technology. I believe this is the approach that M-CAM takes. Finding patents related to a particular real-world product is more difficult because the product might have many patent-eligible features or concepts, and you need a human being to produce a comprehensive list of those features or concepts (which might be extremely long) before you bring search technologies to bear on it.
In any event, software firms really do ignore patents when designing new products. You should read Mark Lemley’s excellent paper on the subject if you don’t believe me.