I actually wanted to create something like this a few months ago, but the idea never got any momentum. Instead, I've been working on software that facilitates prior art search by semantically analyzing "literature," generally. I'm primarily focused on using genetic algorithms to construct a complex search query (compatible with LexisNexis/Westlaw) to find even more prior art (in the form of research articles and whatnot) -- not just patents. Luckily, every patent issued since 1976 already has a full-text version, and there are OCR'd versions of the patents from around 1910-1976 available on google. Unfortunately, academia is much larger than just the patent database, so it's important to broaden our searches further.

I was hoping to use this program to build support for, and ultimately raise enough money to request, ex parte reexaminations of issued patents.

Would any of you have interest if I kept pushing for that idea?

Good way to get more comprehensive results, but how would the wexis queries be funded?

Good question. Right now, Westlaw charges something like $100 per search in ALLCASES, which is their database of all federal and state court opinions ever written. Just to retrieve a single document by citation is something like $10.

The general trick is to try to craft a very broad 'initial' search that will encapsulate all of your results; you can create sub-searches within that set of retrieved documents (limited to at most around 3,000) at no additional cost.

Well, you're not talking ALLCASES, you're looking at academic commentary. Maybe better to pursue a solution that uses HeinOnline?

> What we really need is a crowd-sourcing platform for prior art research

Commercial example: http://www.articleonepartners.com/ "The World's Largest Patent Research Community"

Though apparently they've filed a patent application for crowdsourced vetting for compensation... http://arstechnica.com/tech-policy/news/2008/11/startup-crow...

I just checked PAIR and as of March 5, 2011 it's with an examiner. Most recently, on June 12, it looks like the USPTO wants them to restrict the claims in the patent to a single independent and distinct invention; i.e., it doesn't look like they'll get their way with having their patent broadly cover all forms of reward.

The problem with paying rewards for prior art is that it discourages community collaboration, finding prior art is a mix of inspiration and perspiration and by incentivizing everyone to work alone it actually substantially decreases the chance of building a strong prior art case.

I've been thinking on this area and my conclusion is to do the simplest thing possible: just publish blog posts, github dumps, etc of your prior art...tag it as #prior-art and let twitter and google help you find stuff.