Subject: Re: [openip] Re: "rights" and "freedoms"
From: "Stephen J. Turnbull" <>
Date: Fri, 22 Oct 1999 14:57:39 +0900 (JST)

>>>>> "kms" == Karsten M Self <> writes:

    kms> The problem with the patent system as it exists is that there
    kms> are few if any incentives to issue good patents -- patent
    kms> holders don't seek them, the USPTO's certification doesn't
    kms> select for them, and infringers often find it's easier to pay
    kms> or quit than to fight.  There is little (insufficient?
    kms> none?) risk for bad patents.  Economists have a similar
    kms> example in the phrase "bad money drives out good" --
    kms> imperfect analogy, but it might stir something in Turnbull's
    kms> head.

Again, I think that making the approval process itself adversarial
would help most.  I believe that there is a conservative bias in the
law; if the patent has been approved, the defendent in an infringement
action bears the burden of the proof (once the infringement of a claim
has been demonstrated, anyway).  OTOH, during the approval process,
the burden would shift to the applicant, in the presence of claims of
prior art.

I'm not sure what Ross Williams's "Patentbuster's Inc" is, but I
gather that is similar although perhaps more ex post the patent grant.

    kms> The one reform I think has a proverbial snowball's chance at
    kms> making it would be to add risk to the system.

I'm not sure what you're getting at; the USPTO would have to refund
application fees in the event of a busted patent?  Unless you also add 
bait for patent-busters, the rate of patent-bustings will remain at
X%, and the USPTO just raises the fee by (X%/(100 - X%)), and balances 
its budget.

    kms> I'm also trying to form an idea around the concept of a title
    kms> insurance company, hampered greatly by the fact that I don't
    kms> know exactly what one is, though the general sense I've got
    kms> is that they're an agency which will research a title
    kms> (usually land) and find out if anyone has a lein on it, with
    kms> insurance to compensate you in the event that there is one
    kms> that didn't turn up in their search.

Yeah, this is one of the "business opportunities" I've mentioned in
the abstract.

There's a big problem with the insurance aspect, though, and that is
that the patent insurance company has no idea how to assess the
probability that the invention you want to use or patent isn't already
in the pipe, and you do.  You may very well be aware that your
classmate is working on the same kind of project at a different
company, but you don't know who's ahead.  The insurance company lacks
that kind of information.

So this requires more transparency in the application process, but
there will be strong opposition to that, especially from applicants
whose applications might be denied, for whom the "option" on a trade
secret is very valuable.

Another problem is the degree to which a patent applicant would be
insured against "prior art."  The applicant is probably more expert,
and thus there is adverse selection (riskier candidates apply for more
insurance), just as in the "buddy works for a rival" case.

Just off the top of my head.  Doesn't mean "it won't work, period"
(the insurance premium can be adjusted) but it won't work as well as
we'd hope (a few candidates who have no adverse private knowledge will
be denied coverage unnecessarily).

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."