Subject: Re: "incentive void" (was Re: A different patent covenant...)
From: simo <>
Date: Mon, 23 Oct 2006 11:26:01 -0400

On Mon, 2006-10-23 at 23:33 +0900, wrote:
> simo writes:
>  > On Mon, 2006-10-23 at 14:54 +0900, wrote:
>  > > This market offers real benefits to those who would rather code than
>  > > manage.  Without the patent, they have to take an equity position in
>  > > a firm ... but they have no redress if a rival outperforms their
>  > > managing partner.  With the patent, they can say, "cash, please; no
>  > > shares, no options."  And the frog must jump.
>  > 
>  > I don't get this point, with copyrights it is the same, you can keep
>  > them or ask for a price,
> Absolutely.
>  > where's the difference? 
> The area of the monopoly provided.  In general, if you need to protect
> a valuable technology whose working is visible to the public, a patent
> works, copyright does not.  Cf. the GNU Project, whose primary goal in
> the early days was circumventing copyright protection.  They were
> pretty darn successful at it, too!

I see your point, but consider this: how long did it take from start to
an effectively competing product? I'd say at least 10-15 years.
So copyright was damn effective imo.
Single components were much more fast, some took just a few years, but
then I think that 20 years monopoly in the software field is non-sense,
so to me copyright was effective even in the fastest cases.

>  > Oh, but in the field of software there is no such possibility except in
>  > very rare cases. Copyrights block free riders as well,
> Nonsense, unless you insist that "free" means "exactly zero cost".  Go
> reread the GNU Manifesto.  RMS is quite explicit about the fact that
> he intended to free ride[1] on the design and any network
> externalities available to Unix-compatible systems.  Some at least of
> these could surely have been subject to software and/or business
> method patents, which would have been much harder to circumvent.  I
> mean, howcum I can't download GNU DeCSS? ;-)

It's non-sense only if you consider "zero" the cost of rewriting
software, which is usually the costly part of the game[1], and that's
why I say that copyright is effective.
After all the patent system has been invented to grant the inventor
return of investment, if copyright can do the same I don't see why we
should grant another form of protection on top of it.
And I'd say there is enough evidence that copyright has been effective
that discussing on the patent system is a kind of stretch. I'd say that
any economic theory can easily recognize that too much protection is
bad, and if we can conclude that copyright grants return of investment
then adding the patent system on top of it falls necessarily in the "too
much protection" category.

>  > So, in the field of software, what's the point of patents at this point?
> The same as in any other field: providing a monopoly as compensation
> for the social value of innovation that otherwise would not occur.

The problem is that you have double protection, and this is surely not
good imo, and I'd like to see this point taken into account when you
discuss about the patent system applied anywhere there are other forms
of protection that overlaps, or the theory can't be serious IMO.

> How much value there is (including the value of inducing software
> engineers to learn better reuse practices and inducing improved
> indexing of knowledge, both of which have clearly higher value under a
> system that grants patents), and whether it is larger than the costs
> of the monopoly (including the chilling effect on future innovation
> and the pure frictional costs of patent sharking), is an empirical
> question.

Yes it is, but it is _the_ question you have to answer _before_ allowing
such system (now it is late for some countries and that's a very big

> Since that necessarily indirect measurement has not been done to
> everyone's satisfaction yet, it's rather important to get the theory
> right, otherwise we will not be able to interpret the direct
> observation correctly.

Getting the right data to validate any theory seem a very difficult
problem today. Too many interests involved :-/


[1] The exceptions I mentioned are those fields where serious
mathematical research is needed to solve a problem (but then that's
mathematics to me, and it should not be patentable anyway).