Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 10 Oct 2001 14:01:26 -0700
From:      "Crist J. Clark" <cristjc@earthlink.net>
To:        Salvo Bartolotta <bartequi@neomedia.it>
Cc:        Ted Mittelstaedt <tedm@toybox.placo.com>, "P. U. (Uli) Kruppa" <root@pukruppa.de>, freebsd-chat@FreeBSD.ORG
Subject:   Re: Use of the UNIX Trademark
Message-ID:  <20011010140126.M387@blossom.cjclark.org>
In-Reply-To: <1002731960.3bc479b899603@webmail.neomedia.it>; from bartequi@neomedia.it on Wed, Oct 10, 2001 at 06:39:20PM %2B0200
References:  <000601c15084$87edd360$1401a8c0@tedm.placo.com> <1002663600.3bc36eb096ee5@webmail.neomedia.it> <20011009231343.C387@blossom.cjclark.org> <1002731960.3bc479b899603@webmail.neomedia.it>

next in thread | previous in thread | raw e-mail | index | archive | help
Let me preface with IANAL.

On Wed, Oct 10, 2001 at 06:39:20PM +0200, Salvo Bartolotta wrote:
> "Crist J. Clark" <cristjc@earthlink.net> wrote:

[snip]

> <joking>
> The things mentioned above were new when they were discovered.  What if people 
> had "patented" the [more or less trivial] algorithms mentioned the very same 
> day they were created? :-) 
> </joking>

Solving second degree polynomials was not trivial when it was
discovered. Patent protection is a rather recent invention realtive to
mathematics. We tend to think of science being a field where everyone
is just looking for a chance to share all of their thoughts with
everyone else. That's the ideal. This is not how it always worked nor
how it has ever worked. Great mathematicians (e.g. Newton) closely
guarded their discoveries and methods from the rest of the world until
_they_ were ready to release it. You couldn't use Newton's discoveries
for quite a while after he discovered them since he didn't let them
out right away. Knowing Newton's personality, he might have applied
for patents if could have.

> Or, to shift the origin of the time axis, what if people patented each and 
> every new mathematical/informatics/routing/whatever theorem and/or algorithm?

There are many motives for patenting or not patenting. There is no
point in patenting something that you don't think will ever recover
more in license money than it costs to register and maintain the
patent. Most discoveries fall into that category. When speaking of
"routing," a lot of companies spend lots of money developing
technologies and then give away the knowledge in the form of an open
standard and an RFC. In that case, it is in their interest for as many
people to adopt the standard as possible. For someone like a Cisco,
they can spend money on this with no patent return since they plan to
make money off of the hardware and support. Although it is perfectly
reasonable for them to sometimes patent this kind of work if
interoperaility with other products is not something they are after.

> Should/would you have to pay each and every time a theoretical paper/book of 
> yours made [auxiliary] use of a patented theorem/algorithm?

No. Just explaining how a patented process works is not using the
patented process to produce something. For example, back when RSA was
still patented, it was perfectly legal for me to put the algorithm in
a book and sell it.

> And/or each and 
> every time a program of yours made use of the patented algorithm(s)?

But I could not put a CDROM in my book with a working version of the
code without licensing the technology.

> How long 
> [reasonably] for?

Patents have always had expirations. There are very good arguments
that patents for algorithms should have short lifespans realtive to
the current laws.

> What impact would this have on the development of _new_ works and/or [better] 
> programs in the field (or even in other fields)?

It depends on the field. Allowing patents can cause a field to
boom. If people believe they can make money off of a technology and
that they will have patent protection for their discoveries, they may
invest great amounts of money and resources into the field.

> Incidentally, AFAIK, the development of Mathematics, Logic, Informatics, 
> Science at large has occurred fast and furious without patents for the last 
> few centuries[1]. Freedom of development has been (and is) of primal 
> importance.  It is for **this** reason that you are in a position to develop 
> your new nifty ideas _today_.

As I mentioned at the top, it may not be as free as open as you
think. If anyone had patented ideas, it wouldn't matter today anyways
since the patents would have long ago expired. Anyways, patents _do_
allow information to be "free." If you have a patented device or
process you can feel free to share information about how it works, how
you discovered it, etc. and not need to worry about people stealing
your investment. The alternative is people trying to keep all of their
work completely secret to avoid letting other people hear of it and
copy it.

[snip]

> Theorem/algorithm T/Ai (the ith theorem and the associated algorithm, if any) 
> is a scientific achievement, for i=1,2,...,N.  As such, these theorems must 
> figure in the appropriate book(s).  They constitute knowledge and as such they 
> must be shared.  Not sharing knowledge is a dangerous form of obscurantism.  
> And YES, it should be a national[2] interest to finance researchers rather 
> then make them attempt to *patent* *knowledge*...

They "must be" shared with the world. Hmmm... Again, it is and never
really has been this way. Take one of the most interesting scientific
and engineering efforts of the century, the Manhattan Project. That
research took place amongst a _very_ small but able scientific
community sequestered in the desert. It was quite a while for that to
be shared with the world at large; we all know why. Same story for the
hydrogen bomb or any other technology developments placed under the
cover of national security. The idea that all information "needs to be
free" is a rather naive one.

By the way, I should point out, and this may not be clear to you, that
you just cannot patent information. Say that tomorrow you discover
a proof that (much to everyone's surprise) NP really does not exist
and any mathematical problem can be solved in polynomial time. You
cannot patent that fact or that information. However, if you discover
a method to actually solve some set of problems once believed to be NP
in polynomial time, you can patent that method if you want. If someone
else finds a method to solve some other class of NP problems using
some method of their own, you cannot claim you own it because you
wrote the proof that it could be done.

> It is about this point that I disagree.  In general, IMO Science should be 
> patent-free.

That's just not a reasonable view of how the world works. For example,
the entire field of medicine would come to an almost complete
standstill without patents. HUGE amounts of money are spent developing
drugs and devices to be patented. Again, drug companies and other
research groups freely publish their results in the very rich medical
literature due to the fact that they have patents on the
technologies. If they didn't have patents, they wouldn't publish the
information. (And no, I'm not going to fall into the arguments over
patenting genes or organisms. That is a troubling detail of patent law
that needs serious consideration, but is no cause to throw the baby
out with the bathwater.)

> > (Not to say that some people haven't pulled some things off with the
> > Patent Office. One of the jokes about a former employer of mine was
> > that the company had a patent on least-squares regression. It really
> > did.)
> 
> Urk.

The patent was for applying least-squares regression to a certain
specific type of data in a specific way. It is unlikely it would have
stood up to scrutiny, and I think it's expired by now.

It's not as outrageous as Amazon's "single-click" patent. No where am
I saying patent law is perfect. I don't shop Amazon.

> If Numerical Analysis (Numerical Calculus/Methods in English?) were to become 
> a collection of patents, how [fast] would people go ahead?

I am sure that there are some patents on various methods of numerical
analysis.

> Next, what would prevent all fields of Mathematics from becoming a set of 
> patents?  Would you imagine Functional Analysis (cf Quantum Mechanics) to be a 
> collection of patents? You are most probably aware that Q.M. will have a 
> number of consequences for C.S. and computers at large. 

Again, you cannot patent information. A process you patent can then be
freely shared with the world. You really only need to worry about
getting a license for a patented algorithm if you are selling the
algorithm or using it yourself and selling the products of it.

> There has to be a better form of protection, both of the programmer's work as 
> a form of art and of the programmer's work as science (well, when it is the 
> case :-)), a form of protection avoiding obscurantism.  IMO, FWIW, patenting 
> algorithms is a dangerous way.

But you agree there needs to be protection. If you have a better way,
offer it up. I don't think that the patent system is perfect or that
it is not frequently abused[0]. I do believe patents are the best fit
we have right now for protecting the IP inherent in computer programs.

[0] There are some beauties in the computer and Internet industries
beside's Amazon, Open Market (5,715,314) owns electronic shopping
carts, Priceline.com (5,794,207) owns patents on buyer-driven sales
(and has sued M$ and Expedia, I hate to have to cheer for M$, but...),
and Sightsound.com (5,191,573) owns music downloads (has sued Time
Warner and CDNow, they demand 1% royalties on _all_ online music
sellers).
-- 
Crist J. Clark                           cjclark@alum.mit.edu
                                         cjclark@jhu.edu
                                         cjc@freebsd.org

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-chat" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011010140126.M387>