*BSD News Article 5801


Return to BSD News archive

Xref: sserve comp.org.eff.talk:9208 misc.int-property:551 comp.unix.bsd:5849
Path: sserve!manuel!munnari.oz.au!sgiblab!sdd.hp.com!caen!uwm.edu!linac!att!ucbvax!virtualnews.nyu.edu!brnstnd
From: brnstnd@nyu.edu (D. J. Bernstein)
Newsgroups: comp.org.eff.talk,misc.int-property,alt.suit.att-bsdi,comp.unix.bsd
Subject: Re: Patents:  What they are.  What they aren't.  Other factors.
Message-ID: <10880.Sep3008.43.0892@virtualnews.nyu.edu>
Date: 30 Sep 92 08:43:08 GMT
References: <1992Sep25.185314.8872@gvl.unisys.com> <1992Sep26.161204.24573@rwwa.COM> <BvBp1v.16J@lerami.lerctr.org>
Organization: IR
Lines: 57

If I show you two physical processes you can easily tell whether they're
the same. They achieve the same physical result---moving things around,
changing one chemical into another, whatever---in the same way. Patent
law is founded upon the principle that you can tell when two processes
are the same. The USPTO has to be able to tell in order to grant a
patent. The courts have to be able to tell in order to determine
infringement.

(Does ``same'' mean ``exactly the same''? No. It means, according to the
courts, the same up to ``colorable changes,'' or ``inessential
modifications.'' Yes, this is subjective. ``Same''---the legal word here
is ``equivalent''---is a matter of consensus among normal people. Almost
all legal terms are like this.)

If I show you two mental processes you won't have a clue whether they're
the same. If you're very lucky then the two processes will be expressed
in the same way. But what if one process is my Algorithm W and the other
is the subject of the Miller-Wegman patent? Yes, they happen to be
exactly the same process, but Algorithm W is a whole bunch of
mathematical gobbledygook while the Miller-Wegman patent talks about
storage devices and strings. There are no objects being moved around, no
chemicals being transformed into other chemicals, no creation, no
destruction, _nothing_ which you can use to test whether two mental
processes do the same thing.

The same remains true even if you attach ``insignificant post-solution
activity'' (that's another legal term, despite Scott's claims to the
contrary) to a mental process. If the physical applications of two
mental processes are essential (curing rubber, developing film) then you
can see whether they do the same thing. If the physical applications are
inessential (writing bits to disk, or tape, or memory, or your head, or
reading data from a similar place) then this test disappears.

This is the fundamental problem with mental process patents. I don't
mean to imply that the XOR cursor patent---which doesn't suffer this
particular problem---is fine; I find that patent shockingly far from any
sane standards of unobviousness and originality. But mental processes
are inherently unpatentable. You simply cannot tell when two of them are
the same.

In article <BvBp1v.16J@lerami.lerctr.org> merlin@lerami.lerctr.org (David Hayes) writes:
> Yet, unless *someone*
> publishes the technique, the PTO will never have any documented case of
> prior art on which to rule.

Unfortunately, even if the PTO had a perfect prior art database, the
fundamental problem of mental process patents would still exist. Witness
LZW versus MW. MW was prior art for LZW. Yet LZW was granted a patent.
Not only was MW published: it was sent to the USPTO! The USPTO had two
equivalent patents under its nose and granted _both_. How would it have
helped if the USPTO examiners knew every iota of prior art? They
wouldn't be able to tell which prior art applies to the patent at hand!
If you can't reliably detect the equivalence of mental processes, then
you certainly can't search the prior art for anything equivalent to the
process at hand.

---Dan