Subject: Re: Examples needed against Soft Patents
From: Ben Tilly <btilly@gmail.com>
Date: Wed, 29 Dec 2004 15:04:24 -0800

On Wed, 29 Dec 2004 11:43:05 -0500 (EST), Russell McOrmond
<russell@flora.ca> wrote:
> (Wow -- these are getting longer each time.  As wonderful as these are I
> think we're going to have to slow down a bit.  I suspect we all have
> year-end paid work that needs to get done ;-).

I'll try to make this one shorter.  (I may fail.)

And next week I'll likely have to completely exit this discussion. :-(

>   While there are many arguments against software patents, the one that I
> bring to policy makers are not the ones we are using now.  I use two:
> 
>   a) Patent policy is *economic* public policy that exists to provide
> incentives for innovation.  Different subject matter exists in an entirely
> different economic context, and thus economic analysis and not legal
> analysis is needed to determine if a specific subject matter class should
> be patentable.

This argument is one that could convince me.

>   b) While good quality patents may provide incentives for innovation, bad
> quality patents act as a chill on innovation.  Unless patent quality above
> a certain level (I believe over 90% good quality is a good number) can be
> achieved in a specific subject matter, then patents should not be granted
> at all.  Patent quality in software within the USPTO has been found to be
> approximately 40% to 5% good quality patents.

Too dangerous.  You're accepting some patents as quality patents.

Don't concede anything.  Instead argue something like this:

  b) While it is doubtful that legally valid patents provide
  incentives for innovation, invalid patents certainly hinder it.
  Depending on who does the estimate, somewhere between
  60% and 95% of software patents in the USPTO would not
  stand up in court.  But they still are threats because nobody
  wants the expense and risk of going through litigation to
  prove that.

  Canada should not create the same problem here.

[...]
> On Tue, 28 Dec 2004, Ben Tilly wrote:
> 
> > It was not my goal to say that the reasoning above should be
> > convincing to you.  It was my goal to say that there are people who
> > find that reasoning convincing.  And they're not trivially wrong.
> 
>   It is my belief that once someone has an adequate understanding of what
> software is that they will then realize that the logic you expressed
> (on behalf of others you have heard it from ;-) is trivially wrong.

I beg to differ.

I'm a professional programmer.  I do not find the logic that I
expressed to be trivially wrong.  Nor would your arguments
convince me that software patents are bad.

> > The goal of the patent system is to pay people a temporary monopoly on
> > their solution in return for them revealing it.  If the consequence of
> > their revealing their solution is that someone else can readily
> > implement it, then the temporary monopoly is not very effective, is it?
> 
>   This is one of the goals of the patent system -- revealing a process
> that would otherwise be kept secret and then possibly "lost" to the
> world..
> 
> Lets look at this for software:
> 
>   a) Do software patents reveal the process to someone skilled in the art?
> How many of us who are software developers know what is claimed when we
> read a patent?  If we were writing proprietary software, could we read a
> patent, implement the process, and license the patent?  I would suggest NO
> for most software patents, and that is not just the poor quality (would
> fail useful/novel/unobvious tests) software patents.

The fact that software patents do not reveal the process to
someone skilled in the art does not prove that patents do not
inform those skilled in the art.  All that you need is to have
people who are able to understand the patent who can
communicate with someone skilled in the art.  That person
can then express the idea in a way that others skilled in the
art can understand.

Furthermore once a company has a patent, it has less
incentive to keep the technique secret.  As a result the
existence of the patent may facilitate communication even
if nobody learns from the patent itself.

In fact both of these happen in other fields.  For instance the
makers of generic drugs frequently study patent applications
to learn how to make generic drugs.  There is nothing to
suggest that they cannot happen in software, and there is
some evidence that they do.  (If not in the numbers that
would be required to justify software patents.)

>   b) Is it possible to *USE* software outside of a private environment
> without revealing the process?  Unlike manufacturing processes there isn't
> a separation between the process and the thing produced through the
> process, as the process is the product.  It it a machine?  No more than a
> thought is a body part.  Is it a chemical composition?  It isn't tangible
> at all, but it is a patent where it is the result and not the process that
> leads to the result that is being patented.

Non-sequitor.  There are many patented machines which, if
used publically, are easy to disassemble and reverse
engineer.  In fact patents are often sought explicitly to
prevent this, so that people will spend energy thinking of
useful innovations that they otherwise would not bother
with because the innovations are easily copied to the profit
of others.

>   c) Like other claimed purposes for patents we need to do analysis: is it
> better to provide an exclusive right or to allow things to possibly get
> lost.

Yes.  But given the stakes you'll find plenty of experts on
both sides of the issue, and it is hard for a layperson to
figure out which ones are more credible.

> http://press-pubs.uchicago.edu/founders/documents/a1_8_8s12.html
> "Considering the exclusive right to invention as given not of natural
> right, but for the benefit of society, I know well the difficulty of
> drawing a line between the things which are worth to the public the
> embarrassment of an exclusive patent, and those which are not. "
>    Thomas Jefferson
> 
>   I believe if we as a society did this analysis, rather than leaving it
> to the legal profession to argue over the "true meaning of the word
> technical", then we would conclude that the embarrassment of an exclusive
> patent is far higher in information processes than any perceived benefit.

You believe this, and I believe this.  But you need to convince
people of this.

> > I agree that it is an independently implemented solution.  But it is
> > clearly a *derived* solution and I don't see the innovation in saying,
> > "We don't need to put these physical transistors here, we'll just write
> > the same logic in software."  Particularly not with decades of
> > engineering experience in moving implementations either way between
> > silicon and bytes.
> 
>   I believe this looks at the situation backwards.  I believe if you have
> hardware that could be so trivially duplicated in software logic on a
> generic computer then you would find that the original hardware was not
> really worthy of the embarrassment of an exclusive right.  You don't get a
> patent on all combinations of physical transistors, but a "new and useful
> art, process, machine, manufacture or composition of matter".

God forbid that we should patent any machine simpler than an
operating system!  Not.

Microprocessors are among the most complex machines in
existence and a tremendous amount of effort and skill goes into
their design.  They are also routinely emulated in pure software
during the development process.  Are you really saying that
microprocessors are not worthy of being patented?

Of course microprocessors are not alone in being emulated.
But the key difference between a microprocessor and, say, a
bridge is that the software emulation of a microprocessor is
useable as a microprocessor while the software emulation of a
bridge is not a useable bridge.

>   You may have an exclusive right in the form of an industrial design, but
> this type of exclusive right is closer to copyright than patents.  While
> you are still infringing this right if you simply change the layout on a
> circuit board, it would not infringe this right if you solved the same
> problem in software logic that was previously implemented in circuit
> design.

Under Canadian law, you're right.  Under US law, you're wrong.
Under the opinion that your listener has about what is fair, you
could wind up right or wrong.

>   It is only a derived solution if you believe that the solution is "the
> set of all possible solutions that bring us from A to B".  Claims of that
> type would generally be considered too broad and such a patent would be
> considered a poor quality patent.  Just because you reached B does not
> mean that you went there by the patented process.

This is clearly wrong.  If I take your solution and modify it in a
fairly straightforward way to get a new solution, most people
without an axe to grind will say that I've created a derived
solution.

Since this is possible when going from hardware to software,
it is possible to translate a hardware design into software and
produce a derived solution.

>   The engineering experience, and possibly the patented processes, would
> be in the manufacture of the silicon, and possibly the processes by which
> implementations are moved between silicon and bytes.  There are new
> learnings about the forces of nature that possibly exist in both cases.

Where did the phrase "new learnings about the forces of nature"
come from?  New learnings about the forces of nature applies to
*scientific discovery* (which isn't patentable).  There was no new
learning about the forces of nature in, say, the design of a zipper.
But the design of a zipper was patentable (and was patented).

>   The question we are asking here is if there should be exclusive rights
> on the processes embedded within the bytes themselves.

Yes.

> How about the reverse.  Mechanical calculators and later electronic ones
> were patentable.  Given they only carried out mathematical problems that
> had been carried out in the heads of humans for thousands of years
> (obvious prior art in the field of mathematics ;-), then how could these
> calculators have been patented?

Prior art existed for how to have humans do the work.  Prior art
did *not* exist for how to have machines do the work.  Automation
is a "useful improvement" and is therefore patentable.  It is no less
automation if we're talking a mechanical calculator than if we're
talking about Eli Whitney's cotton gin.

>   Maybe the patent is not "the path to go from A to B", but a new and
> useful way of doing so that is considered entirely unique from previous
> ways of doing this.
> 
>   I don't see how implementing "A to B" in hardware, software or wetware
> (in our brains) can be considered equivalent.  If they are between
> hardware and software, then how can any machine that carried out a task
> previously done by humans ever have been patented?

My understanding is that the tools that exist to design certain kinds
of hardware (including microprocessors and similar devices) can
be used to realize that design either as a specification for the
hardware or as a software program which emulates the final
hardware.  There is no difference in the work needed to design
either.  Both will accomplish the same job in the same "way".
How can you not see the equivalence between these two
implementations?

There is no analogous comparison that can be made between
how one builds a machine and trains a human, so that
equivalence does not exist.

[...]
> > The Canadian courts decided that the above is how they should interpret
> > Canadian law.  To patent owners whose patents are now easily
> > circumvented, this decision is unfortunate.
> 
>   Public policy like patent policy is not simply about what a narrow
> special interest such as patent agents and their clients want, but what is
> on balance of all the interests will benefit society.

The public policy question is whether it is in the interests of
society to reward inventors with effective patents.  You're
advocating making patents in certain fields of endeavour be
relatively ineffective.

> > And certainly it is within the rights of Canada's legislature to pass
> > laws which would make those patent owners far happier.
> 
>   Without public consultation?  Without economic analysis?  Without the
> simplest of cost/benefits analysis?

Does the Constitution require these things of Canada's
legislature?  Not that I'm aware.  Implicitly it is to be hoped that
politicians would not capriciously pass bad legislation, and it is
further hoped that somewhere between the House of
Commons, the Senate, and the Queen (or her appointed
representative in her normal state of absence) that someone
would take the public interest into account.

But it is not required.

>   While it is possible to pass legislation without any consultation or
> analysis, I'm not so convinced it is within the rights of a democratically
> elected legislature to abandon their representative roll by doing so.

Au contraire.  It is within the rights of a democratically elected
legislature to represent their constituents in any way that they
see fit.  If the constituents are unhappy with the result, it is the
right of the constituents to choose better representatives.  See
"elections".

Furthermore it is impossible for the legislature to examine
every issue before them in depth.  This is why the government
has cabinet ministers, and the opposition shadow ministers,
each with their own staff to provide analysis.  From time to
time the analysis is incomplete - perhaps they only heard
one side of the story.  This is part of life.  But the system is
intended to survive mistakes like this.  Government does not
need to be perfect, merely good enough.  And if it is not
good enough, well such is life.

[...]
>   It isn't about only applying to hardware, but the need to check for each
> subject matter class whether the first principles that justified patent
> policy still hold.
> 
>   a) will it provide incentives for innovation, or stifle innovation.
>   b) is the embarrassment of a patent greater or lesser than the potential
> loss of an "invention" due to secrecy or other lack of disclosure.

I don't see b being relevant.

> ...etc...
> 
>   Lawyers just look at the word "process" in the definition of an
> "invention" and say that anything that can be considered a process should
> be patentable.  This means to them that software (which is obviously a
> type of process) should be patentable, and if ever asked they would
> probably believe that all processes should be patentable including
> parliamentary procedures, Robert Rules, acts of parliament, etc, etc....

Some lawyers clearly do believe that, see "business process patents"
under which something like Robert Rules might indeed become
patentable. :-(

Your job is to prove them wrong, not absurd.  (It is possible to be
absurd and right...)

[...]
> > Likewise in a computer with a piece of software, there has been
> > an internal physical rearrangement.
> 
>   Just as new thoughts in the brain can be claimed to have a different
> physical arrangement.  This is totally unconvincing to me, and to be
> honest I have never been able to respond to this claim as it seems
> to totally ludicrous to me.

When I compare software to a chair, the comparison seems
ludicrous.  When I compare software to physical machinery
of similar size to the representation of that software, the
comparison becomes far less ludicrous.  In fact the division
between what we consider a "physical thing" and what we
don't comes to seem to be arbitrary.

Of course the question of what is or is not part of a "chair"
also comes to seem arbitrary when I try to examine it on that
scale, but I'm not in the habit of looking at chairs that
closely.

>   The idea that a computer with different software in it is a "new
> machine" is something I can argue against as well as I can argue
> against someone who truly believes the world is flat.
>
>   Can anyone else come up with a response to this?

"You might as well argue that a piece of paper that I've
written on is now a new piece of paper."

But the analogy breaks down.  By writing on a piece of
writing paper one cannot make a piece of paper that
functions well as toilet paper.  But by writing to the
memory of one computer you can make it act like
another.

[...]
> > I do not believe that I started or ended with the conclusion that
> > patent lawyers wanted a new type of monopoly.
> 
>   You are relaying all the reasons that we have all been exposed to to
> justify information process patents.  I don't know that these are reasons
> that you independently came up with, and thus your motives for exploration
> of these various excuses doesn't suggest why such time was spent to come
> up with these excuses.

I did not originate them, and my motivation in exploring them
was to understand how we came to have software patents.  As
for your statement about patent lawyers, I have a different
theory.  My theory says that a lawyer was trying to represent
his client's case in the best possible light and got creative.
His creative argument stuck.

It wouldn't be the first time.

> > It is true that the line of reasoning was intended to result in a
> > specific conclusion (that if you accept patents, there is no obvious
> > reason to say that you shouldn't have software patents), but the
> > intention has no bearing on whether the reasoning is valid.
> 
> BTW:  There is another argument we have all heard embedded in there, which
> is that any opposition to information process patents is really opposition
> to all patents.

Let me formalize that argument slightly.  Questioning the
utility of software patents involves questioning the utility of
patents in general.

>   I'm an example of someone who believes that proper economic analysis
> done on each subject matter area would reveal entirely different results.
> And I do believe that there are subject matter areas where patents do
> promote innovation, and where th embarrassment of a 20 year monopoly is
> lesser than the loss of something learned through an invention.

I believe that there are people in areas that you think can
benefit from patents who think that their areas are better
off without patents.  I further believe that it is unlikely that
we'll have solid research at any time soon that all
reasonable participants can agree is a proper economic
analysis that settles this question.  I believe that because I
think that people have a hard time being convinced of
anything particularly inconvenient to them, but still to the
layperson this leaves the problem of, "smart people who
legitimately disagree - which to believe?"

[...]
> > I just want to be sure that I understand what you're saying.  You
> > mean that patent lawyers went to court and did their job - they
> > presented their clients' cases as convincingly as they could.  In
> > fact they presented the cases so convincingly that the judges
> > were convinced!
> >
> > I'm not entirely sure what is supposed to be wrong with that.
> 
>   I'm not saying that the lawyers on behalf of the patent filers didn't do
> their job.  I am suggesting that the other half of the case, the people on
> the side of "the people" such as government lawyers, government
> economists, public policy analysts, etc, etc -- that these groups who are
> supposed to protect our collective interests did not do their job.

That wasn't the implication that I got from what you originally said.

My rewrite is how I expect your opponents to mentally rewrite
what you said.  Perhaps they'll say it privately, perhaps they'll
say it to you, but it is a response to expect.

[...]
>   The underlying question is a bit different, which is that the economic
> question about whether patents do what they are supposed to do must be
> done on a per subject-matter category basis.
> 
>   It is not a simple question, it is a critical question, and all that
> people like me are asking for is that the government do their job and seek
> to answer this question before granting new types of patents.

You're asking the job to do a lot of work that it isn't clear how to
do.  The argument that you'll get back is that this is an obvious
ploy to stall software patents forever, no matter how
overwhelming the need you'll always need more research.
Plus existing patent holders in other fields are likely to see this
as very threatening to their interests since it throws their
patents (which they consider very valuable) into question.

I'm not saying you're wrong, just that there will be a natural
response.

> > Obviously a physical court house can't do that.  But it is possible
> > to literally turn software into hardware - take software and build a
> > custom piece of hardware that does exactly what that software
> > would do, in the exact same way, but will do nothing else.
> 
>   I simply don't understand this logic: If you can go from A to B using
> software, and can go from A to B using hardware, then the hardware and
> software are equivalent.

Yes, obviously you don't understand this logic.  It is not that the
hardware and software achieve the same effect.  It is that they
are essentially identical in what they do and *how* they do it.

In fact you literally can take a single description and, through
automated tools, wind up either with software or equivalent
hardware.  (This is useful because it runs faster as hardware
but is easier to debug as software.)  How can you not see that
hardware and software as completely equivalent?  In effect
hardware has just become another "architecture" that you can
"compile" to.

>   Patents were never intended to be an exclusive right on "all possible
> methods to go from A to B", they are a disclosure of something learned
> (traditionally about the forces of nature, but that is being eroded) as
> part of the inventive process.

When have patents ever been about something learned
about the forces of nature?

[...]
> > If the dividing line was truly as obvious as you're making it out to be
> > in talking about court houses and acts of parliament, then shouldn't
> > they have found it easy to clear up any attempt at confusion?
> 
>   Was there anyone in that courtroom as motivated as practitioners in the
> art to having good public policy in this area?  I think the answer to that
> was no.

I'd expect both sides to have consulted with practioners in the
art.  Furthermore not all practioners in the art agree with you.
There are programmers who believe in software patents, in
fact some have even filed for patents!

>   There were lawyers representing patent agents who obviously have a
> strong motivation to increasing patentability.
> 
>   There were lawyers representing patent offices, which also have a
> motivation to increase patentability.
> 
>   I didn't see evidence of other interventions.

Are you accusing the lawyers representing the patent offices of
deliberately throwing the case because winning it is not in
their best interests?  That's a pretty serious accusation...

>   The conclusion reached is not surprising given the circumstances of the
> case.  The problem I have is that the circumstances were very wrong for
> trying to set good public policy, and there was nobody in that courtroom
> representing the public interest.

True, courts are not good places to set public policy.  That's why
we have legislatures...

Cheers,
Ben