Subject: Re: US Software Patents Hit Record High (fwd)
From: <>
Date: Tue, 26 Sep 2006 14:06:38 +0900

Brian Behlendorf writes:
 > On Sat, 23 Sep 2006, wrote:
 > > Brian Behlendorf writes:
 > >
 > > > And the $64,000 question: if software could not be patented, how much of
 > > > the work done that resulted in the patents would not have been
 > > > done?
 > >
 > > > My guess: it would have made no difference.
 > >
 > > My first guess is that a lot of the work would not have been done.
 > > Specifically, a lot of documentation would not have been written.
 > > Much of that documentation isn't useful for real work, of course, but
 > > I bet the authors could turn out real documentation pretty quickly if
 > > license revenues were on the line.
 > I'm confused - are you saying that patents essentially form useful 
 > documentation to the ideas that arise from solving problems,

No, I'm saying that the exercise of writing a patent probably makes it
possible to write useful documentation more cheaply.

 > > My second guess is that a lot of work would have been redirected from
 > > products where software ends up in the hands of users doing data
 > > processing to products where data ends up in the hands of ASPs.  Not a
 > > pleasant bias.
 > Because software that exists on the client desktop can be decompiled, and 
 > someone's "secret sauce" would become plain?  I suppose, but I can't 
 > imagine there being a *faster* migration to ASP-style software than we are 
 > currently seeing today.

That's an empirical question.  I will accept that your judgment is
better than mine, and (until I have empirical evidence to the
contrary) call that a "potential" bias that doesn't seem to be important.

 > > My third guess is that a lot of the work that was inhibited by fear of
 > > submarine patents would have been justified in that fear by the
 > > existence of prior work---and therefore (in a frictionless world)
 > > shouldn't have been done anyway.
 > If we had no software patents, how could the fear of submarine patents 
 > been justified at all?

You're missing the point, which is the clause after the dash.  See my
reply to Tom Lord, who paints the rosiest possible picture of the
social benefits of patents while complaining bitterly of how
ungenerous International Characters is.

 > > My fourth guess is that "friction" is substantial enough that any
 > > reasonable attempt to measure these effects will show that by far the
 > > most important effect is that many products that could have been
 > > brought to market never get there, while very few additional products
 > > (including "patent licenses") are generated by the patent system, and
 > > the effects of redirection of work are relatively minor, too.
 > Sounds pretty damning.

Not until the measurements are done.

It's also damning primarily in the context of the immediate past
practice, where even a cursory search for possibly relevant patents
could cost thousands of dollars.  Suppose (in a Brooksian gendanken
experiment, not as a practical matter) we could reduce the cost of
finding relevant patents to *zero*?  How would that change this
equation, not to mention the whole practice of software engineering?