Slashdot Mirror


Prior Art to Pinpoint vs. Amazon, from 1980's?

Gary Robinson writes "I'm in a fairly unique position with regard to the Pinpoint vs. Amazon case since I built a system in the mid-1980's which is commonly regarded as the first active service based on collaborative filtering. It was a voice-mail-based dating service called 212-ROMANCE. I still have the 8-inch CP/M source code disks as insurance against CF-related patent lawsuits. Today I've posted a discussion of the Pinpoint vs. Amazon case in the context of that prior art as well other prior art from the 1980's."

4 of 139 comments (clear)

  1. Re:One should have to prove "no prior art"! by ivan256 · · Score: 4, Interesting

    One should have to prove "no prior art"!

    The way it works is exactly the opposite of that. Prior art does not invalidate a patent, it just make the patent dependent on having the rights to the prior art as well.

    Unless your patent is exactly what the prior art is (hardly anything is exactly something else (!)) or some subset of it, finding prior art will not prevent the granting of a patent. Also, many patents that are flaged as outrageous by armchair lawyers on slashdot are not as broad of scope as they seem due to the prior art that is listed in the patent.

    The process may be broken, but not as badly as you'd think from all the postings you see around here.

  2. Don't discount such patents by mike_lynn · · Score: 2, Interesting

    If you look at the patents that reference patent 5,443,036, you'll see several people have already built upon the concept. What's sad isn't the concepts that people patent but the extents to which a corporation will go purely for extortion. Maybe they need a new department that will determine the generality of a patent and decide if it needs to be public domain/free. Yay more government. :p

  3. Re:One should have to prove "no prior art"! by alangmead · · Score: 3, Interesting

    Patent examimers are supposed to do a check for prior art. Unfortunately, their usual body of work to check from is other patents. From their point of view, useful items are patented, so examining other patent filings is the best place to find already existing inventions.

    The problem is, there patent office refused patents on software until forced to by the Supreme Court in the 1981 case Diamond v. Diehr The software industry developed without any consideration to patents, and now the patent office is missing decades of the basic building blocks needed for determing the novel from the mundane.

  4. Re:collaborative filtering and different domains by js7a · · Score: 2, Interesting
    Oh come on. That there is no guarantee that the same algorithms will work with different data won't mean squat to a jury if, in fact, the algorithms do work with the different data sets.

    Don't try to convince me that there is anything fundamental going on here beyond sparce matrix distance metric theory. Trying to recommend dates is essentially similar in Joe Random Juror's mind to trying to recommend music and books. Even if one algorithm is entirely unsupervised and the other depends on a huge preference-screening front end, the jury only sees the black boxes doing the same things.

    In this labor market, the defendant in infringement suits can get the more convincing experts on the stand, even if poorer than the plaintiff. Do you know how many linear algebra Ph.D.s are driving taxis right now?