Subject: Re: Examples needed against Soft Patents
From: Marshall Van Alstyne <marshall@MIT.EDU>
Date: Sun, 19 Dec 2004 14:53:05 -0500

At 04:43 AM 12/18/2004, Bernard Lang wrote:

>- One major argument that was used against me is :
>
>   there have been software patents in the USA and Japan for 20 years
>   (their number ... it is actually more like 10 years, I think)
>   but free software is still doing well in both countries.
>
>    How do you answer that?

They actually make a somewhat faulty argument.  The issue shouldn't be 
whether software is doing OK but is software doing as well as it 
could?  What's the possible optimum and how close are we to getting there?

A colleague of mine did some consulting for a hi tech firm whose sr mgmt 
refused to listen on the basis that they were growing just fine.  But 
competitors grew faster and the dot bomb fell and the rest is history...


>  > - I definitely need a detailed commented list of free projects,
>  >   everywhere, that have been stopped because of patents.

Here are two places not for examples, but for stronger arguments.

The first is an academic repository of papers on diverse subjects at 
http://ssrn.com.  Searching for "software patents" brings up several good 
articles.  One of the most interesting is an empirical investigation by 
Bessen & Hunt who observe that SW patents now represent 15% of all new 
patents but only 5% of them go to software publishers.  The main problem is 
the correlation of SW patents with strategy and NOT with R&D intensity, 
which does not fit the traditional incentive theory of patents.

Note there is a critique of this critique by Hahn & Wallsten who disagree 
with some of the methods.

The second is some modeling we're doing on a related but distinct 
topic.  We're trying to develop a formal analytic model of innovation, 
combining the best of incentive theories and free / open source 
methods.  It's not complete, but one result we do have is that, even in the 
context of profit motivated developers, the social optimum for protection 
is *never* arbitrarily long.  In fact, it must be less than the useful 
economic life of the software.  We're only distributing the working draft 
to "friends & family" so to speak, but I'm happy to share a copy 
privately.  We'll post a public copy as soon as we've vetted results.

MVA