Subject: Re: termless copyright and patents
From: <stephen@xemacs.org>
Date: Sat, 23 Sep 2006 17:59:02 +0900

Thomas Lord writes:

 > Not that any of that earlier work got practiced in anything
 > close to what Joe did.

Whether it was practiced or not doesn't matter.

 > Not that when that more broad patent was filed people had anything
 > like what Joe did in mind.  Well, fine.  That doesn't invalidate
 > Joe's claim in the slightest.

That's just an assumption though.  And your whole scheme is assuming
that Joe *cannot know* because he hasn't done the homework.

 > In short,  this is a partial answer to the "follow-on
 > innovation" problem.

I don't see how.  The follow-on innovation problem is that a
patent-holder has a nearly absolute right to prohibit practice of
claims.  This creates uncertainty for would-be licensees, *especially*
in Joe's case.  There's no reason to suppose that IBM will settle for
less than covering its legal expenses of negotiating the license, and
that could be more than Joe is asking.  You still have an anti-commons
problem.

 > Joe shouldn't use these patent options for a claim where that is a
 > serious possibility.

In Joe's case, it's always a serious possibility; it's pure arrogance
to think "nobody thought of this before" if you haven't researched.
You know that the PTO is going to do a poor job of researching it.

 > Conversely, if Joe's patent is about a clever way to manage
 > bookmarks in a WWW browser, and it shows up in IE and/or Mozilla
 > after Joe gets some hackers interested in it (regardless of whether or not
 > he can show a direct line of transmission) then it shouldn't matter if
 > something vaguely kinda-sorta analogous happens to have been described
 > re Xanadu.

What do you mean "shouldn't matter"?  As Larry Rosen would say, you
put the Xanadu-related patent on the table next to Joe's patent, and
if it invalidates Joe's claim, Joe's claim is pig fodder and the
royalties go to the owner of the X-r-p.  This is somewhat unlikely
because the PTO *does* do a patent search in its check for prior art,
but we all know it doesn't do a great job.

 > It's a bit like the patent equivalent of small claims court (but
 > 50x larger stakes or so).

That's going to require substantial legislation to establish.

 > Hey, it's a darn sight better than some proposals that have been
 > seriously made like "Let's legislate that software patents have a
 > shorter term than others."

Why?  I agree that giving either the PTO or the Congress another
control to tweak is something we should think twice about, but your
"small patent court" isn't merely a tweak, it's a whole new setup, and
quite a bit harder to figure out the implications in advance.  So,
what makes all this better?

On the contrary, there's plenty of theoretical economic research
showing that the optimal length of patent grant depends on a number of
factors which are likely to vary by industry.  So it comes down to
"does different terms by field give more opportunity for manipulation
and confusion than the benefits of optimizing the term for each field
justify?"