Subject: mandatory patent auctions
From: <stephen@xemacs.org>
Date: Thu, 5 Oct 2006 22:24:36 +0900

Seth Gordon writes:

 > I have seen a proposal floated around that the government should reform
 > the patent system as follows:
 > 
 > (1) When a patent is issued, the owner must put it up for auction.  (You
 > can bid on your own patent in the auction.)
 >
 > (1.5) [my friendly amendment:] When putting the patent up for auction,
 > the owner may specify nondiscriminatory terms under which the
 > auction-winner and all successive owners must license the patent.
 > 
 > (2) After the auction is over, the government flips a coin, and if it
 > comes up heads, the government buys the patent for 150% of the winning
 > bid, and releases the patent into the public domain.

It seems to me that this invites predation from deep-pocketed
companies that do no research.  They can afford to pay up to the full
monopoly operating profit for it; researchers can pay only that minus
the development costs.

There's also the consistency problem of "who pays whom here?"  If the
winner pays the developer, you have "Marti's Money Machine".  If the
winner pays the government, who pays the developer?

 > A company that gets a patent will have to weigh the benefit of
 > cashing in to the highest bidder against the risk that the highest
 > bidder will be a direct competitor;

"Cashing in" is already possible, and quite common at least in
anecdotes.  Eg, in the research-only biotech sector and also the
famous "write great software, get bought by Microsoft" strategy for
garage programmers.  That latter gives me pause.

 > it can hedge against that risk by, for example, specifying that all
 > open-source software that practices the patent may do so without
 > paying a licensing fee.

I don't really see risk; your own bid will take account of this.  If
your own bid is not highest, on average it's just free money from a
competitor, who is subject to the winner's curse.[1]  If so, as Tom
points out, what's left is you lose out in case (2), and this is a
definite loss (no "risk", just lose), and further leveraged by the 50%
(or whatever).


Footnotes: 
[1]  It's very likely that the average estimate of the market demand
is a better guess than the maximum estimate---but the maximum estimate
will win the auction.  In theory, you can compute how big this effect
is, and shade your own bid appropriately.  In practice, nobody is very
good at this---not college students in experiments and (according to
Paul Milgrom, a leading expert, holding forth with a strong wine buzz
at a conference :-) not big oil companies in real auctions of drilling
rights.