A different patent covenant...

[Brian Behlendorf]

...one that still accomplishes a patent holder's goal of realizing
economic return from a large number of potential users, but which places
the burden of worrying about the flaws of the patent system onto those who
want to play the patent game:

"I hereby grant an unrestricted license for implementation, use, and
distribution (etc.) of any software that implements my patents, for no
fee, to any entity who agrees to license all of their own patents under
this same license to any entity.  Alternate commercial licensing options
are available upon request."

This would preserve a revenue stream for the patent holder from other
large patent holders who have not made the same kind of grant.  Small
companies with few patents will be incented to make such a covenant; large
players with existing large revenue streams from patent licensing will
probably hold off on doing so.

Yes, "all their own patents" is very broad.  More "viral" than the GPL.
This is because patents themselves are a very broad enforcement mechanism
- I can be subject to a patent I've never even seen.

Companies who register patents solely for defensive purposes are still
protected if they make such a covenant, because the other party will
nearly by definition not agree to such a covenant.

I daresay IC would make just as much money from this approach as from
their proposed covenant.  I know it would be more acceptable to many of
the open source projects out there than a "non-commercial use" kind of
license.

  Brian

[Rob Cameron-2]

On September 28, 2006 01:59 pm, Brian Behlendorf wrote:

>
> ...one that still accomplishes a patent holder's goal of realizing
> economic return from a large number of potential users, but which places
> the burden of worrying about the flaws of the patent system onto those who
> want to play the patent game:
>
> "I hereby grant an unrestricted license for implementation, use, and
> distribution (etc.) of any software that implements my patents, for no
> fee, to any entity who agrees to license all of their own patents under
> this same license to any entity.  Alternate commercial licensing options
> are available upon request."
>
> This would preserve a revenue stream for the patent holder from other
> large patent holders who have not made the same kind of grant.  Small
> companies with few patents will be incented to make such a covenant; large
> players with existing large revenue streams from patent licensing will
> probably hold off on doing so.
>

This is a very interesting idea.   It has a nice kind of ironical
twist to it.   It reflects more of an anti-patent stance than an open
source stance, though.    

I still think there is merit in trying to create an equation between
FOSS and systems that promote the "progress of science and
useful arts."    It may help the cause of patent reform to get the
constitutional intent onside.   If IBM, Sun, and/or other major players
buy in, there might be a real chance at a legislated equation
one day (I dream).

> Yes, "all their own patents" is very broad.  More "viral" than the GPL.
> This is because patents themselves are a very broad enforcement mechanism
> - I can be subject to a patent I've never even seen.
>
> Companies who register patents solely for defensive purposes are still
> protected if they make such a covenant, because the other party will
> nearly by definition not agree to such a covenant.
>
> I daresay IC would make just as much money from this approach as from
> their proposed covenant.  I know it would be more acceptable to many of
> the open source projects out there than a "non-commercial use" kind of
> license.
>

One problem of your covenant for IC's business is that embedded
network applications like XML appliances are a natural area
where high-performance text handling matters.    Proprietary
but patent-free competitors would get a free ride.

Also, commercial use of open source is OK with us, provided any
code modifications are given back.   It is actual commerce in
infringing systems that matters.

[Thomas Lord]

I'm confused about something:

Mr. Cameron says:

  Rob> Also, commercial use of open source is OK with us,
  Rob> provided any code modifications are given back.  It is
  Rob> actual commerce in infringing systems that matters.


but earlier, speaking of IC's covenant, Mr. Rosen says:

  Stephen>> Patent law is not copyright law, it cover this
  Stephen>> situation, and IC made it clear that if the "medium"
  Stephen>> is actually hardware (computer+memory) then you need
  Stephen>> a license.

  Lawrence> Stephen is right. We consider this hardware. If
  Lawrence> software is preinstalled on hardware, then it is a
  Lawrence> combination. The issue is not whether it is "bit for
  Lawrence> bit identical" (that's a copyright issue) but
  Lawrence> whether the product embodies our patent claims.


Aren't you contradicting one another?

Consider a network appliance vendor who has received a GPLed
program embodying the invention.  And suppose that IC has never
itself distributed this program.   And the vendor sells these
boxes to the public, satisfying the GPL source requirements.

According to Mr. Cameron, that's sufficient for the vendor to be
protected by the IC covenant.  According to Mr. Rosen, no, the
vendor needs a license.

Which is it?

(A possible answer is simply "neither": that the conditions of
the GPL weren't satisfied when the vendor got a copy of the
program and consequently he has no rights at all to use the
program.)

-t

[Brian Behlendorf]

On Fri, 29 Sep 2006, Rob Cameron wrote:
> This is a very interesting idea.   It has a nice kind of ironical
> twist to it.   It reflects more of an anti-patent stance than an open
> source stance, though.

Is it anti-patent to ask that those who wish to benefit from the monopoly
rents granted by government via the patent system also be the ones to bear
the costs of patents as well?

> I still think there is merit in trying to create an equation between
> FOSS and systems that promote the "progress of science and
> useful arts."

Clearly, there has been no lack of innovation from Open Source despite the
lack of patents to protect those innovations.  There are a myriad of
reasons one might innovate; a patent is just one possibility.

> One problem of your covenant for IC's business is that embedded
> network applications like XML appliances are a natural area
> where high-performance text handling matters.    Proprietary
> but patent-free competitors would get a free ride.

Those competitors in the marketplace would have a tough time without
partnering with larger companies with distribution channels who are more
likely to have patents they wish to protect.  Perhaps we could define
"entity" to include those other partners.  For example, we could bind not
just the entity creating the product implementing your patent, but also
any owners of that entity whose share is equal to or greater than 25%.
That way if, say, Dell, Fujistu, and Flextronics did a joint venture where
each controlled a third of the entity that created some XML processing
appliance, where the JV itself held no patents, we could bind those three
companies to the covenant... meaning they would likely choose commercial
licensing from IC.

  Brian

[Jamie Lokier]

Thomas Lord wrote:
> I'm confused about something:
>
>  Rob> Also, commercial use of open source is OK with us,
>  Rob> provided any code modifications are given back.  It is
>  Rob> actual commerce in infringing systems that matters.

I'm pretty confused by that paragraph alone.  Is there a intended to
be a difference between "commercial use of open source" (which is OK)
and "actual commerce"?

-- Jamie

[Stephen J. Turnbull]

Brian Behlendorf writes:

 > Clearly, there has been no lack of innovation from Open Source despite the
 > lack of patents to protect those innovations.  There are a myriad of
 > reasons one might innovate; a patent is just one possibility.

That is true.  But what you are not saying, and may not even
recognize, is that some of those reasons apply to some fields of
innovation sufficiently to generate innovation, and not to others.
It's quite possible that vast swaths of "innovatables" are out there
that fall into that incentive void.  (It's also possible that it's an
incentive crack and we don't care.)

If it is in fact a large incentive void, then patents are a useful
policy tool.

[Norbert Bollow]

<stephen@...> wrote:

> If it is in fact a large incentive void, then patents are a useful
> policy tool.

That is non sequitur.

Even if there is a large "incentive void", the patentability of
software ideas can still be much more harmful than beneficial.

If an "incentive void" exists, the first step should be to
identify as exactly as possible what categories of innovations
are part of that "incentive void".

This alone will increase the chances of someone working on it:
There are people who want to work on some useful innovation without
caring much whether they will personally benefit from their work.
If these people are provided with information on the "incentive
void", that will significantly increase the likelihood of them
choosing to work on a problem which is part of the "incentive
void".

Secondly, once part of such an "incentive void" has been
identified, we can discuss what's the best way of addressing
the problem.

For example, I'm sure that the overall cost of the software
patents system exceeds $10million/year.  How about awarding, on
a yearly basis, ten "Nobel prize like" prizes of $1million in cash
each, plus the honor, to ten people who have made most impressive
innovative contributions in the area which was previously the
"incentive void"?

Wouldn't that be likely to take care of the problem quite nicely,
without the really ugly side-effects of the software patents system?

Greetings,
Norbert.


--
Norbert Bollow <nb@...>                       http://Norbert.ch
President of the Swiss Internet User Group SIUG        http://SIUG.ch

[Stephen J. Turnbull]

Norbert Bollow writes:
 > <stephen@...> wrote:
 >
 > > If it is in fact a large incentive void, then patents are a useful
 > > policy tool.
 >
 > That is non sequitur.

Sure, but I really meant "large enough", which makes it trivial.

 > If an "incentive void" exists, the first step should be to
 > identify as exactly as possible what categories of innovations
 > are part of that "incentive void".

First, in the U.S. and Japan, the status quo is the other way 'round.

Second, I don't think it will be at all easy to specify at levels more
precise than "software vs. pharmaceuticals".  Rob Cameron's invention
is an example of something that has been latent in the field for
several years, at least according to Tom Lord and Jamie Lokier, but
nobody's bothered to do, despite the fact that XML is one of the most
popular buzzwords in the corners of free software I know about, not to
mention software in general.  Suppose that Cameron's invention did
plunk into that void.  Tell me, how would you distinguish it from all
the XML technology that everybody and his sister is working on?

 > There are people who want to work on some useful innovation without
 > caring much whether they will personally benefit from their work.
 > If these people are provided with information on the "incentive
 > void", that will significantly increase the likelihood of them
 > choosing to work on a problem which is part of the "incentive
 > void".

I think you vastly overestimate the amount of manpower of that kind
available, compared to the number of redundant solutions that never
get published because they're boring and have no obvious audience, and
problems that remain unsolved because no one business alone can profit
enough from solving it to bother.

 > For example, I'm sure that the overall cost of the software
 > patents system exceeds $10million/year.  How about awarding, on
 > a yearly basis, ten "Nobel prize like" prizes of $1million in cash
 > each, plus the honor, to ten people who have made most impressive
 > innovative contributions in the area which was previously the
 > "incentive void"?

We already have that.  In the U.S. it's called the NSF, in Japan the
JSPS.  In Europe I'm sure you have one too.

 > Wouldn't that be likely to take care of the problem quite nicely,
 > without the really ugly side-effects of the software patents
 > system?

Almost certainly not; *impressive* contributions are pretty much by
definition not in the void that I'm talking about.  This system cannot
reward the efforts to promote (as opposed to develop) technology---but
that's where at least 50% of "innovation" lies.

That's the whole difficulty here.  Patents are not really about
rewarding world-class innovations; as we have known for generations,
those are their own reward, in fame, in fortune, or both.  The point
is rather to provide incentive for documenting, accumulating, and
promoting a large mass of small innovations.  Of course no incentive
is provided by rewarding innovations that have already happened, or
that are *immediately* obvious to any practitioner.  But once you've
cleared those bars, the market created by the patent gets the
direction right ---the innovations that most hurt when you have to pay
a high license fee are precisely the ones making the biggest
contribution to society.

It's still a monopoly, and therefore still pernicious.  But the patent
system does, in theory, address a set of problems that no other
proposal I've seen does.  These need to be compared to the costs.

[Craig Brozefsky]

<stephen@...> writes:

> That's the whole difficulty here.  Patents are not really about
> rewarding world-class innovations; as we have known for generations,
> those are their own reward, in fame, in fortune, or both.  The point
> is rather to provide incentive for documenting, accumulating, and
> promoting a large mass of small innovations.

That is an new insight for me, thanx.

> Of course no incentive is provided by rewarding innovations that
> have already happened, or that are *immediately* obvious to any
> practitioner.  But once you've cleared those bars, the market
> created by the patent gets the direction right ---the innovations
> that most hurt when you have to pay a high license fee are precisely
> the ones making the biggest contribution to society.

However, I could not parse the last sentence.  Could you restate it
perhaps?

> It's still a monopoly, and therefore still pernicious.  But the patent
> system does, in theory, address a set of problems that no other
> proposal I've seen does.  These need to be compared to the costs.

You admonished me to calculate the costs in another thread.  I
attempted to describe a model of the dynamics -- which is what I was
basing my original statement that "so much friction, so little benefit
to so narrow a class".  I would like to see if we could come to some
agreement about how to calculate that cost.


--
Sincerely, Craig Brozefsky <craig@...>
"Do you want to live forever?"       -- Valeria

[Norbert Bollow]

<stephen@...> wrote:
> Norbert Bollow writes:
>  > <stephen@...> wrote:
>  >
>  > > If it is in fact a large incentive void, then patents are a useful
>  > > policy tool.
>  >
>  > That is non sequitur.
>
> Sure, but I really meant "large enough", which makes it trivial.

Ok, but in that case your statement might well be trivially true
in a very misleading way:  If it turns out (like I and other software
patent opponents conjecture is the truth) that enforced software
patents always cause more overall harm than benefits, then even if the
"incentive void" turns out to be 99% or more of all possible, useful
software innovations, then that is still not "large enough" in the
required sense for making the statement trivial.

>  > If an "incentive void" exists, the first step should be to
>  > identify as exactly as possible what categories of innovations
>  > are part of that "incentive void".
>
> First, in the U.S. and Japan, the status quo is the other way 'round.

Not really:  There is not currently a functioning (in the sense of
allowing those people who work on innovations, or who finance such
work, to get paid, in the case of success, an amount of money which
is sufficiently large for providing an effective incentive, while
ensuring that the total of this revenue and transaction costs does
not exceed the aggregated benefits of the innovation) patents
system.  This could change if e.g. IC's business model, and the
community patent review process they're participating in, work out
and they have a lot of imitators.  But currently the status quo is
that a software patents system exists but it doesn't achieve its
stated objectives.

> Second, I don't think it will be at all easy to specify at levels more
> precise than "software vs. pharmaceuticals".  Rob Cameron's invention
> is an example of something that has been latent in the field for
> several years, at least according to Tom Lord and Jamie Lokier, but
> nobody's bothered to do, despite the fact that XML is one of the most
> popular buzzwords in the corners of free software I know about, not to
> mention software in general.  Suppose that Cameron's invention did
> plunk into that void.  Tell me, how would you distinguish it from all
> the XML technology that everybody and his sister is working on?

Ok, I'll try, but of course, when starting from arbitrary assumptions,
it will be a perfectly acceptable result to arrive at a contradiction.
(In that case it may be inferred that at least one of the assumptions
was wrong.  This could be one of the explicit assumptions which you
have suggested, or one of my own implicit assumptions.)

I think that it's not very smart to use XML in areas where efficient
performance really matters.

Unicode has been popularized by XML, but it makes good sense to use
unicode also independently of XML, including in performance-sensitive
contexts.

This is only now starting to be popular.

Hence only now a market for the kind of innovation which IC has
patented is emerging.

Therefore, the distinction which I find is between features for which
there has been a market for quite some time now (cool, non performance
sensitive, XML-based features) and a feature for which the market is
only now developing (efficient unicode character conversion).

Generalizing this finding, there are many software patents of this
kind:  Not-extremely-big innovative steps addressing problems where
the market for solutions to the problem is only now emerging.

Looking at the history of software development before software ideas
became patentable, I would say that this kind of innovative work of
finding solutions to problems where a market for solutions to the
problem is emerging, and where the problem can be solved by means of
not-extremely-big innovative steps, this is not part of the "incentive
void" that we have been talking about.

Alas this conclusion contradicts one of the assumptions that you
wanted me to make.  Sorry about that.

>  > For example, I'm sure that the overall cost of the software
>  > patents system exceeds $10million/year.  How about awarding, on
>  > a yearly basis, ten "Nobel prize like" prizes of $1million in cash
>  > each, plus the honor, to ten people who have made most impressive
>  > innovative contributions in the area which was previously the
>  > "incentive void"?
>
> We already have that.  In the U.S. it's called the NSF, in Japan the
> JSPS.  In Europe I'm sure you have one too.

Please provide details of why you believe that NSF and JSPS are
already doing what I'm proposing.

>  > Wouldn't that be likely to take care of the problem quite nicely,
>  > without the really ugly side-effects of the software patents
>  > system?
>
> Almost certainly not; *impressive* contributions are pretty much by
> definition not in the void that I'm talking about.  This system cannot
> reward the efforts to promote (as opposed to develop) technology---but
> that's where at least 50% of "innovation" lies.
>
> That's the whole difficulty here.  Patents are not really about
> rewarding world-class innovations; as we have known for generations,
> those are their own reward, in fame, in fortune, or both.  The point
> is rather to provide incentive for documenting, accumulating, and
> promoting a large mass of small innovations.

But isn't "documenting, accumulating, and promoting a large mass of
small innovations" precisely what Microsoft, and all the companies
which try to compete with Microsoft, are doing?

Microsoft has been doing this with great commercial success for a long
time totally independently of software patents.

I don't agree with your assertion that an "incentive void" exists in
this area.

Greetings,
Norbert.


--
Norbert Bollow <nb@...>                       http://Norbert.ch
President of the Swiss Internet User Group SIUG        http://SIUG.ch

[Stephen J. Turnbull]

Craig Brozefsky writes:

 > <stephen@...> writes:

 > > But once you've cleared those bars, the market
 > > created by the patent gets the direction right ---the innovations
 > > that most hurt when you have to pay a high license fee are precisely
 > > the ones making the biggest contribution to society.

 > However, I could not parse the last sentence.  Could you restate it
 > perhaps?

If the license fee hurts more than the value to you, you simply won't
pay it.  A high license fee that is actually paid signifies high value
to the payer.  Because of "downward sloping demand", it's quite
possible that even at a very high fee, the value to some payers is a
multiple of the fee.  That adds up to "big contribution".

Nevertheless, if you could (or already did!) invent it for yourself at
much lower cost, that fee is really really gonna hurt.

In other words, it's a very rough estimate, but if there isn't much
value to the innovation, you just can't charge enough for it to hurt.

I'm going to leave your question about "agreeing on how to measure"
without a proper answer for a while.  But the basic answer is "not
soon", because answering these questions is always a matter of
successive approximation.

[Stephen J. Turnbull]

Norbert Bollow writes:

 > >  > If an "incentive void" exists, the first step should be to
 > >  > identify as exactly as possible what categories of innovations
 > >  > are part of that "incentive void".
 > >
 > > First, in the U.S. and Japan, the status quo is the other way 'round.
 >
 > Not really:  There is not currently a functioning

Yes, really, in the sense of the politics necessary to get any changes
made.  Malfunctioning by your standards is not a showstopper from the
point of view of the big players; they believe it can be reformed to
their benefit, as far as I can see.

 > But currently the status quo is that a software patents system
 > exists but it doesn't achieve its stated objectives.

Don't you mean that you think that it doesn't achieve the objectives
you think it should serve?

For example, in the U.S. the software patents system was created by a
court which decided that there was no legal ground for distinguishing
between hardware and software, and issued a decision requiring the
USPTO to abolish that distinction.  The court's objective---to
rationalize this wart in the application of the law---has been
achieved, without doubt.

You can say that's more dumb than rational, and I tend to agree.  But
that is the way that the American justice system tends to view these
matters, and in that sense the patent system is "successful".

So what are these "stated objectives" that the system doesn't achieve,
and by what criterion did you determine that failure?

 > > plunk into that void.  Tell me, how would you distinguish it from all
 > > the XML technology that everybody and his sister is working on?
 >
 > Therefore, the distinction which I find is between features for which
 > there has been a market for quite some time now (cool, non performance
 > sensitive, XML-based features) and a feature for which the market is
 > only now developing (efficient unicode character conversion).

As far as I can see this reduces to the assertions that (a)
demand-driven innovations, not demand-inducing, innovations are the
ones that matter, and (b) demand will bring forth the corresponding
innovation in a timely way; there's no point in having it in advance.
I think both assertions are quite questionable, (a) more so than (b).

 > > We already have that.  In the U.S. it's called the NSF, in Japan the
 > > JSPS.  In Europe I'm sure you have one too.

 > Please provide details of why you believe that NSF and JSPS are
 > already doing what I'm proposing.

Basic research has been alleged to be underfunded by commerce for
millenia, most recently on this list by simo.  Thus the government
research foundations clearly are aiming at an incentive void.
Although they claim that they fund interesting proposals, in practice
the good ones do not; they fund researchers with a track record of
producing interesting results.  This isn't exactly what you proposed,
but it's similar enough that it seems likely that the results would be
the same: developers would be attracted to big wins "interesting to the
government referees", and distracted from getting product to market.

 > > That's the whole difficulty here.  Patents are not really about
 > > rewarding world-class innovations; as we have known for generations,
 > > those are their own reward, in fame, in fortune, or both.  The point
 > > is rather to provide incentive for documenting, accumulating, and
 > > promoting a large mass of small innovations.
 >
 > But isn't "documenting, accumulating, and promoting a large mass of
 > small innovations" precisely what Microsoft, and all the companies
 > which try to compete with Microsoft, are doing?

Only if you drastically twist the meanings of "document" and "promote".

 > I don't agree with your assertion that an "incentive void" exists in
 > this area.

But that's not my assertion.  My assertion is that I see some evidence
that it does exist, and no conclusive evidence that it doesn't.  It is
therefore premature to jump to the conclusion that abolition is the
best policy for anyone except the FSB lobby.

[Norbert Bollow]

<stephen@...> wrote:

> Norbert Bollow writes:
>  > > We already have that.  In the U.S. it's called the NSF, in Japan the
>  > > JSPS.  In Europe I'm sure you have one too.
>
>  > Please provide details of why you believe that NSF and JSPS are
>  > already doing what I'm proposing.
>
> Basic research has been alleged to be underfunded by commerce for
> millenia, most recently on this list by simo.  Thus the government
> research foundations clearly are aiming at an incentive void.
> Although they claim that they fund interesting proposals, in practice
> the good ones do not; they fund researchers with a track record of
> producing interesting results.  This isn't exactly what you proposed,
> but it's similar enough that it seems likely that the results would be
> the same: developers would be attracted to big wins "interesting to the
> government referees", and distracted from getting product to market.

I would suggest that those who have a breakthrough insight _and_
what it takes to create an marketable product around that insight
will not be so foolish to neglect that business opportunity for
the gamble of trying to win a prize.  People who have what it
takes to succeed in business are not foolish in this way.

However I'm sure that there are people who'd have a decent chance
of succeeding in creating a marketable product and a successful
business around it, if they're given a $1mln prize which they're
free to use as seed capital.

Greetings,
Norbert.


--
Norbert Bollow <nb@...>                       http://Norbert.ch
President of the Swiss Internet User Group SIUG        http://SIUG.ch

[Ben Tilly]

As loathe as I am to see this thread continue, I'll respond to one
point that I found interesting.  (And so it continues...)

On 10/2/06, stephen@... <stephen@...> wrote:
> Norbert Bollow writes:
[...]
>  > But currently the status quo is that a software patents system
>  > exists but it doesn't achieve its stated objectives.
>
> Don't you mean that you think that it doesn't achieve the objectives
> you think it should serve?

I believe that Norbert means that it doesn't achieve its stated
objectives.  The stated objective of the US patent system is, "To
promote the Progress of Science and useful Arts, by securing for
limited Times to Authors and Inventors the exclusive Right to their
respective Writings and Discoveries."  At least that is what the US
Constitution says, and that is where Congress gets its authority to
pass laws involving intellectual property.

[...]
> So what are these "stated objectives" that the system doesn't achieve,
> and by what criterion did you determine that failure?

As I stated above, the stated objective is as outlined in the US
Constitution.  I suspect that Norbert, like myself, believes that on
balance the patent system does not achieve that objective.  However
quantitatively demonstrating it is not easy.

[...]

>  > Therefore, the distinction which I find is between features for which
>  > there has been a market for quite some time now (cool, non performance
>  > sensitive, XML-based features) and a feature for which the market is
>  > only now developing (efficient unicode character conversion).
>
> As far as I can see this reduces to the assertions that (a)
> demand-driven innovations, not demand-inducing, innovations are the
> ones that matter, and (b) demand will bring forth the corresponding
> innovation in a timely way; there's no point in having it in advance.
> I think both assertions are quite questionable, (a) more so than (b).

I do not think that (a) is involved.  However he has a very good point with (b).

When a market is just developing that makes certain lines of
innovation natural, you're going to see lots of people begin
developing along those lines of innovation.  We don't need patents to
motivate this work, and introducing patents into that situation is
likely to create more friction than useful additional innovation.
Also given how young the market is, obviousness becomes hard to
determine.

However if someone comes up with a new idea in a mature market, the
fact that nobody has done it before pretty much demonstrates that the
idea was not obvious.  And it is very arguable that patents are needed
for a new entrant to survive in that market.

It is hard to distinguish innovations that are being driven by current
market conditions from ones that come out of the blue.  But one proxy
is to look at the dependencies for the patent.  I'd like the patent
system a lot more if it, for instance, said that you cannot patent
anything that depends on a technology which is less than 10 years old.
 (I'd also like the patent system more if it wasn't insanely broken,
but we're talking theory now, not practice.)  For instance this would
have blocked the insane number of "do X on the web" patents that we
saw during the dot com era.  Sorry, but most of those were obvious.
And the fact that nobody claimed it yet just means that nobody claimed
it yet.

On the other hand if there is no prior art for your idea and it
depends on technologies that are at least a decade old, the odds are
much better that it is not obvious and market incentives are not
sufficient to get people to find it.

[...]

Cheers,
Ben

[Stephen J. Turnbull]

Norbert Bollow writes:

 > I would suggest that those who have a breakthrough insight _and_
 > what it takes to create an marketable product around that insight
 > will not be so foolish to neglect that business opportunity for
 > the gamble of trying to win a prize.

But patents are not about "breakthrough insights."  They are about
useful improvements over current practice.

Nor are they about "creating marketable products".  They're about
following through and actually doing the marketing.

 > However I'm sure that there are people who'd have a decent chance
 > of succeeding in creating a marketable product and a successful
 > business around it, if they're given a $1mln prize which they're
 > free to use as seed capital.

They're also free to burn it or to create a non-marketable product,
because it's their money.  The point about patents is that they do not
pay off on burning money, nor on non-marketable products.

[Stephen J. Turnbull]

Ben Tilly writes:

 > I believe that Norbert means that it doesn't achieve its stated
 > objectives.  The stated objective of the US patent system is, "To
 > promote the Progress of Science and useful Arts, [...]"

Sure, that's the objective of the patent system.  But most people do
not doubt that the patent system does to some extent promote progress,
at least in some fields.  Now, Norbert specifically said the "stated
objectives of the *software patent system*".  If you applied that kind
of logic to IT, there wouldn't be any IT in the sense of CIO (how many
IT departments post *any* revenue at all to the accounts? they're all
so deep in the red, fire 'em all right now!)

I presented reasonable evidence that the stated objective of the
*software* patent system might be mere Emersonian consistency.  For
better or worse, the law in the U.S. is generally built on principles
of consistency.  So the horrible software patent system gets a free
ride vis-a-vis general principles on the (assumed for the purpose of
this paragraph, only) success of the general patent system.

The law and the economy are not software (as anybody who has done
requirements work knows!)  You cannot arbitrarily modularize them, and
fix the modules individually, nor will the political system treat
demands that it do so with respect.

 > I suspect that Norbert, like myself, believes that on
 > balance the patent system does not achieve that objective.  However
 > quantitatively demonstrating it is not easy.

But that's what I ask for, is a quantitative measurement.  Nor do you
have to show "on balance" in my case---as a member of the Cult of the
Invisible Hand, all you need to do is show that the net benefit is not
so big.  Now, there's no a priori reason why the burden shouldn't be
on the pro-patent lobby to provide it, but the politics and status quo
are such that it's on the abolitionists.

Can't we try for an achievable reform instead of the acid trip of
abolition?  The need for reform is easy to document, in fact,
politically you probably don't even have to (except to push IBM &cie
further in the desirable direction).

 > > As far as I can see this reduces to the assertions that (a)
 > > demand-driven innovations, not demand-inducing, innovations are the
 > > ones that matter, and (b) demand will bring forth the corresponding
 > > innovation in a timely way; there's no point in having it in advance.
 > > I think both assertions are quite questionable, (a) more so than (b).
 >
 > I do not think that (a) is involved.

You don't believe in demand-inducing innovations?  How do you classify
the (3M variety) Post-It note, or spreadsheets, or the web browser
(yeah, right, J. Couch Potato is going to get his giggles by down-
loading 10MB DNA visualizations and CERN preprints over his 1200 baud
line), or the general idea of killer app?

 > However he has a very good point with (b).

Well, I admitted that, and of course you think so!  (b) is the ideal
workflow for hackers with itches, so in an environment of free
software, it's likely that you're going to observe innovations that
address current needs, not innovations that elicit needs that people
didn't know yet.  Eg, the Cameron technology, which we've been told
several times is oh-so-obvious and has been forever.  Consider your
dependency criterion: is SIMD ten years old yet? Unicode is, and XML
is getting close.

 > is to look at the dependencies for the patent.  I'd like the patent
 > system a lot more if it, for instance, said that you cannot patent
 > anything that depends on a technology which is less than 10 years
 > old.  For instance this would have blocked the insane number of "do
 > X on the web" patents that we saw during the dot com era.

*guffaw* It's harder than that to unbreak an egg.  Here, just cite Ted
Nelson and you've got a couple decades leeway.

 > And the fact that nobody claimed it yet just means that nobody
 > claimed it yet.

That's elegant enough to be worth saying.  In fact, I've quoted you so
everybody will see it again. :-)

[Norbert Bollow]

<stephen@...> wrote:
> Norbert Bollow writes:
>
>  > I would suggest that those who have a breakthrough insight _and_
>  > what it takes to create an marketable product around that insight
>  > will not be so foolish to neglect that business opportunity for
>  > the gamble of trying to win a prize.
>
> But patents are not about "breakthrough insights."  They are about
> useful improvements over current practice.

With "breakthrough insights" I mean very non-obvious insights on how
to improve existing practice.  The contrast I intend to make is with
the kind of improvements over current practice where is the current
system at least, patents are granted, but where the inventive step
isn't really that non-obvious to world-class professionals in the
field.

I feel that making and publishing breakthrough insights is an
acitivity that ought to be rewarded in some way.  And I've been
getting the impression that among those who think that software
patents somehow are, or could be, beneficial is some way, the
popular reason why these people think that software patents are
good is that they expect those who make breakthrough insights
to be able to get a significant reward precisly because they can
get a patent.

That's why I'm suggesting an alternative way in which such
"breakthrough insights" could be rewarded.

> Nor are they about "creating marketable products".  They're about
> following through and actually doing the marketing.

No. Patents are about demanding royalties from competitors (and
sometimes even preventing them from selling their products), totally
independently of the quality of your products and your marketing.

Greetings,
Norbert.


--
Norbert Bollow <nb@...>                       http://Norbert.ch
President of the Swiss Internet User Group SIUG        http://SIUG.ch

[Stephen J. Turnbull]

Norbert Bollow writes:

 > With "breakthrough insights" I mean very non-obvious insights on how
 > to improve existing practice.

I *know* what you mean by "breakthrough insights."  My point is that
nobody except the anti-patent lobby believes that patents should be
limited to such (eg, very few drugs are break-through insights;
rather, most are discovered by trawling through thousands of tonnes of
Amazonian muck and dropping droplets on Petri dishes full of
bacteria), and there's absolutely no foundation in U.S. or Japanese
law for such a policy.  Dunno about the EU.

 > > Nor are they about "creating marketable products".  They're about
 > > following through and actually doing the marketing.
 >
 > No. Patents are about demanding royalties from competitors (and
 > sometimes even preventing them from selling their products), totally
 > independently of the quality of your products and your marketing.

Exactly.  And they're *still* about following through and actually
doing the marketing.  What you're missing is that one patent does not
a product make.  If your products and marketing suck, your competitors
can use an inefficient workaround to provide minimal functionality,
and whup your sorry butt into bankruptcy on the basis of overall
superiority.  Then they pick up your patents at distress-sale prices.

The only time "working around" is impractible is when the patent
pertains to a genuine breakthrough insight.  But even many FS
advocates will concede that they could swallow patents for those, if
they could be limited to those.

Your prize suggestion is theoretically reasonable.  Nonetheless, both
theory and long experience suggest that markets are far more accurate
at evaluating marketables than prize committees are.  The patent
system proposes to take advantage of this accuracy, at the expense of
granting monopolies as you describe them.  There is no theorem that
says that one outweighs the other.

[Ben Tilly]

On 10/3/06, stephen@... <stephen@...> wrote:
> Ben Tilly writes:
[...]

>  > > As far as I can see this reduces to the assertions that (a)
>  > > demand-driven innovations, not demand-inducing, innovations are the
>  > > ones that matter, and (b) demand will bring forth the corresponding
>  > > innovation in a timely way; there's no point in having it in advance.
>  > > I think both assertions are quite questionable, (a) more so than (b).
>  >
>  > I do not think that (a) is involved.
>
> You don't believe in demand-inducing innovations?  How do you classify
> the (3M variety) Post-It note, or spreadsheets, or the web browser
> (yeah, right, J. Couch Potato is going to get his giggles by down-
> loading 10MB DNA visualizations and CERN preprints over his 1200 baud
> line), or the general idea of killer app?

What I meant is that I do not think that (a) was involved in Norbert's
statement.  I'm not saying that demand-inducing innovations are not
important.  Just that they are orthogonal to what Norbert was saying.

>  > However he has a very good point with (b).
>
> Well, I admitted that, and of course you think so!  (b) is the ideal
> workflow for hackers with itches, so in an environment of free
> software, it's likely that you're going to observe innovations that
> address current needs, not innovations that elicit needs that people
> didn't know yet.  Eg, the Cameron technology, which we've been told
> several times is oh-so-obvious and has been forever.  Consider your
> dependency criterion: is SIMD ten years old yet? Unicode is, and XML
> is getting close.

That is not why I thought so.

>  > is to look at the dependencies for the patent.  I'd like the patent
>  > system a lot more if it, for instance, said that you cannot patent
>  > anything that depends on a technology which is less than 10 years
>  > old.  For instance this would have blocked the insane number of "do
>  > X on the web" patents that we saw during the dot com era.
>
> *guffaw* It's harder than that to unbreak an egg.  Here, just cite Ted
> Nelson and you've got a couple decades leeway.

According to Wikipedia, the first software release for project Xanadu
was in 1998, so that egg remains broken.  Besides, all you need to do
is say that "depends on" in this case are dependencies for the actual
implementation.

>  > And the fact that nobody claimed it yet just means that nobody
>  > claimed it yet.
>
> That's elegant enough to be worth saying.  In fact, I've quoted you so
> everybody will see it again. :-)

I like tautologies because they are always true. :-)

But I did have a point.  Which is that there is a world of difference
between ideas that nobody has claimed because nobody has gotten around
to it, and ideas that nobody has claimed because they really aren't
obvious.  The patent system is supposed to reward the latter and not
the former.  The problem is that the two can be hard to distinguish.

Cheers,
Ben

[Stephen J. Turnbull]

Ben Tilly writes:

 > On 10/3/06, stephen@... <stephen@...> wrote:
 > > Ben Tilly writes:
 > [...]
 > >  > > As far as I can see this reduces to the assertions that (a)
 > >  > > demand-driven innovations, not demand-inducing, innovations are the
 > >  > > ones that matter, and (b) demand will bring forth the corresponding
 > >  > > innovation in a timely way; there's no point in having it in advance.
 > >  > > I think both assertions are quite questionable, (a) more so than (b).
 > >  >
 > >  > I do not think that (a) is involved.
 > >
 > > You don't believe in demand-inducing innovations?
 >
 > What I meant is that I do not think that (a) was involved in Norbert's
 > statement.  I'm not saying that demand-inducing innovations are not
 > important.  Just that they are orthogonal to what Norbert was
 > saying.

Ie, he doesn't believe they're important.  I really don't think
throwing away what is potentially the biggest benefit of a patent
system (each innovation being made earlier) is kosher.

Asking the right questions is almost always the hard part.  IMO, doing
so before anybody else does is what deserves the patent, not the
comparatively straightforward work that goes into answering them.
Doing so years before anybody else was going to even more so, but
unfortunately predicting futures that no longer *can* happen is even
harder than predicting futures that *will* happen, so we can't use
that as a criterion.  We're left with "first-past-the-post".

 > > didn't know yet.  Eg, the Cameron technology, which we've been told
 > > several times is oh-so-obvious and has been forever.  Consider your
 > > dependency criterion: is SIMD ten years old yet? Unicode is, and XML
 > > is getting close.
 >
 > That is not why I thought so.

I know that is not why you thought so; you thought so because reading
the description you find it easy to believe that asked for that
technology you'd come up with it.  But you didn't, nor did Tom, nor
Jamie.  So let's not talk about hindsight; it *is* orthogonal to the
issues.

From the foresight point of view, you proposed a ten year criterion.
The Cameron technology passes.

 > > *guffaw* It's harder than that to unbreak an egg.  Here, just cite Ted
 > > Nelson and you've got a couple decades leeway.
 >
 > According to Wikipedia, the first software release for project Xanadu
 > was in 1998, so that egg remains broken.  Besides, all you need to do
 > is say that "depends on" in this case are dependencies for the actual
 > implementation.

You *really* don't want to go there.  That is the biggest breakage in
the current patent system: the fact that only patents are considered
prior art.  That's for pragmatic reasons: they're easier to search
than the whole CS literature for the examiners.  But you want to
enshrine even worse breakage in the defined procedure?!  Do you really
want the only thing that counts as prior art to be art that's actually
released?  You do understand that to get the interpretation you want,
you'd have to concede that in fairness?

There is no shame in just wanting patents to go away.  That's an
honest, and perhaps the most correct, point of view.  Why not stick to
that?

 > But I did have a point.  Which is that there is a world of difference
 > between ideas that nobody has claimed because nobody has gotten around
 > to it, and ideas that nobody has claimed because they really aren't
 > obvious.  The patent system is supposed to reward the latter and not
 > the former.  The problem is that the two can be hard to distinguish.

That's a structural problem, that we expect patent examiners to be
experts in technology rather than arbitrators among competing
experts.  Adversarial procedures have been proposed for a long time,
but few patenters like them for obvious reasons, and patent opponents
are chasing the will-o-the-wisp of abolition, so there's too little
grass-roots support for it to overcome the status quo.