Subject: Re: [Freesw]
From: "Frank BENNETT (フランク ベネット )" <>
Date: Sat, 12 May 2001 19:15:37 +0900

On Fri, May 11, 2001 at 11:42:05PM -0600, Richard Stallman wrote:
> There are major problems with  Actually, two
> problems--one tactical, and one strategic.
> In the US patent system, if the PTO looked at certain prior art and
> decided to issue the patent anyway, the court is supposed to presume
> the PTO was right to regard that prior art as insufficient.
> But if the PTO was unaware of the prior art, then the court can look
> at it with an unbiased eye.

I'm always ready to go back to school ...

With a little effort, I have been able to confirm that this is in fact correct
(citation is as important to the law as it is to ephemeral publications :-).
For anyone who is interested in seeing this rule in action, try looking at In
Re Portola Packaging, Inc., 110 F.3d 786; 1997 U.S. App. LEXIS 6443; 42
U.S.P.Q.2D (BNA) 1295.  On the Web, the text of the opinion can be found at:

The text includes quotations from 1980 deliberations in Congress and the
reference for the bill that subsequently introduced this rule.

> As a result, prior art is more effective against patents if the PTO
> does not know about it.  For potential patent victims to inform the
> PTO about prior art is a self-defeating project.

As Bernard Lang said, this is an important issue, so it is worth taking a
close look at the arguments.  This conclusion depends on two assumptions:
that, other things being equal, a court is more likely to deny a given
software patent than is the patent office; and that the potential victim has a
large litigation budget and can justify expending it in defending against a
given patent attack.

The first assumption is trickier than it might at first appear.  Any given
instance of patent litigation begins with a grant of a patent by the PTO. It
may then be subjected to reexamination proceedings in the PTO. If this is
followed by a judicial appeal, the court then has just two choices: overturn
the patent; or uphold it.  Once you reach the litigation node, you can say
that better knowledge of prior art in the PTO has hurt the case.  But this
does not tell us anything about the likely effect of better publicity at the
examination and reexamination nodes -- on whether we reach the litigation
node in the first place.  Legal rules can't give us the answer to this: we
need empirical evidence.

The second assumption will depend on the victim and the invention concerned.
The cost rises at each step in the process.  That is, persuading the examiner
through the record of prior art is cheapest, reexamination is more dear than
passive defense during examination, and litigation is more dear than
reexamination. It's not a pretty equation, but in most cases the rational
strategy (at the individual level) is to raise the bar in the patent process
for inventions that may be targeted.  We can try on three possible
assumptions.  If the PTO does not apply a systematically looser standard of
novelty and obviousness than the court would do, then filing with
always makes sense (if, that is, it's worth the 20 bucks).  If the PTO issues
patents without regard to the prior art in the examiner's file, then filing
with is a waste of time.  But if the PTO does take prior art into
account, but is less likely than the court to reject an application on prior
art grounds, then the incentive to file is inversely proportional to the value
of the invention.

I agree that eliminating software patents altogether is the best solution to
this.  But if they exist, an individual's tactical reaction to them will
depend on his or her relative chances at each of the three steps in the
review process.  Does anyone know of studies that have been done on this

> The effect of this is worse than you might think, because of the way
> the PTO uses prior art.  The question they are suppose is, "Is this
> idea unobvious given the known prior art?"  But their threshold of
> "unobvious" is so low, that in practice the tiniest difference from
> the known prior art is enough excuse for them to issue a patent.  The
> courts are much more likely to apply a sensible definition of
> "unobvious", if they are not blocked by a prior PTO decision about the
> same prior art.

My reading today has also taught me that "obviousness" and "novelty" are
different concepts, with different tests attached to them.  The two are
smudged together in the paragraph quoted above, so perhaps it is worth
outlining the difference for the sake of clarity. If an invention is identical
to a single piece of prior art, it fails the novelty test.  The slightest
difference between the invention and any individual piece of prior art is
sufficient for the invention to clear the novelty test.  Obviousness, as far
as it is related to prior art, has to do with combinations of preexisting

That said, it seems to me that the prospect that a combination of
preexisting software elements could be found to satisfy the "non-obviousness"
criteria just serves to strengthen the objection to software patents
generally.  The idea that a software patent should be allowed on the basis
that it combines two preexisting pieces of software in a way that is not
obvious is, er, not very obvious to me.

> Then there is the strategic problem.  I have seen publicity associated
> with this activity, and it serves as an excuse to whitewash the system
> of software patents.  The publicity suggests that we could live with
> software patents, if only we "work to make the system function" in
> this way.  It encourages people to think that the only problem in
> software patents is when non-novel ideas are patented, and that
> software patents on new ideas (some brilliant, most pedestrian) are
> ok.  And that will undermine the efforts now under way in Europe to
> prevent software patents there.

I understand the concern. But it seems to me that if publicity is misleading
about the public interest, the best response is to keep things clear and
respond on point.  I don't really see a strategic connection between the
_existence_ of and lobbying efforts in Europe.

Frank Bennett