Re: [tlug] Patents gone wrong?

[<stephen@example.com>]

Josh Glover writes:

 > On 25/08/06, Evan Monroig <evan.monroig@example.com> wrote:

 > > I don't really know much about patents, but I have the feeling that
 > > something is really wrong with the system as it is now (and not only in
 > > the US, which is the focus here).

 > Yes, the patent system in the US is very broken, at least as it
 > relates to software. I personally consider software patents a Bad
 > Thing.

 > Steve, as an economist, is there an argument for software patents
 > being necessary? Is it just the current implementation that is broken,
 > or the design, or the whole idea of software patents?

In economics, nothing is "necessary".  The question is "what is the
tradeoff between real resources spent and real benefits received?"
Fortunately, most of the time you can substitute $$$ for "real
resources", and often enough for "real benefits", too, but that's a
computational hack.  It works in this case, though.

Patents, in a frictionless world, would be a GoodThang[tm].  The
motive to benefit humanity, while pretty strong, is unreliable and
typically extremely biased.  I misdoubt, for example, that it would
ever have occurred to rms to create window-manager "skins" (and maybe
even window managers).  There's a good reason why the Linux kernel is
best-of-breed---and was so long before IBM et al started injecting
huge amounts of money---and AbiWord an unacceptably bad substitute for
the real thing, with OpenOffice.org barely acceptable despite the
infusion of tens, maybe hundreds, of millions of dollars of Sun
money.  Developing a good wordprocesser is unfun; developing a great
kernel is High Art and great sport.

In the end, if you can't make it yourself, you want to option to pay
somebody to do it.  If you're Bill Gates, you can afford that, no
matter what "it" is (except maybe an Enterprise-class aircraft
carrier).  But most of us want to share the burden with tens or
hundreds of millions like us.  If it's a matter of a price that's 10
times higher than the cost of getting the work done, vs. a personal
share that is 1/1,000,000 vs. 1/10,000 (Phil Zimmerman's 1990 estimate
of downloads and personally distributed copies of PGP shareware
vs. registrations), is that a fuckin' great deal or what?  "90% off"
is not "or what", I can tell you that much!  (Hey, if you think Disney
is greedy, try adding up the petty greed of all the Napster users, who
don't want to pay that 10%, even.  Disney is a piker.)

All a patent is, is a device to accomplish that tradeoff.  The factor
of ten overpricing is implicit in the whole idea; nobody expects a
monopolist to undercharge, especially if he's explicitly exempted from
antitrust!

In software, there's a second reason for wanting a system of patent
registration:  to encourage reuse.  Go look at freshmeat, and look at
the number of text editors.  Or just to the firefox plugin page, and
see the number of dumb, incompetently written variations on some of
the more popular basically trivial themes.  Suppose you could get all
those fluff-brained kittens to work together?  "Patent as cat-herd."

And a third, standard interfaces, which is a variation on the theme of
reuse but sufficiently important as to deserve separate mention.

What goes wrong?  Monopoly pricing isn't it, and one manufacturer
paying royalties to another isn't it.  That's the whole point!

No, the problem is that enormous resources are required merely to find
out that a patent exists.  Far more than the cost of developing the
thing from scratch, including doing requirements studies,
architecture, design, and the relevant testing, yourself.  Since
ignorance of a patent is no excuse, this has a strong chilling effect
on innovation---precisely the reverse of the intent.  And that's the
economic argument against the current patent system in a nutshell.
Higher costs, and no net benefits to anybody but the monopolists.
"Cuff 'im, Danno."

With respect to the Creative menuing patent, I'm sorry, but while it's
not in the technical class of say RSA, a good UI for selecting one of
thousands of items based on a button or two is not easy.  Maybe y'all
are too young to remember the "blinking Jan. 1, 1970, 00:00:00" VTR
urban legend, but anybody whose been around since the popularization
of VTRs knows that manus are hard to do.  Compare ease of use of a Mac
with a one button mouse to a Windows box or GNOME with two or even
three!

What really needs to be done with patents IMO is

1.  Immediately raise the bar, specifically by introducing a
reasonable period (say 6 months) of provisional protection after
administrative grant of a patent during which time (a) royalties would
be escrowed to guarantee their return if the patent is nullified and
(b) the patent is published for public comment, including presentation
of invalidating prior art.  Or possibly a temporary suspension of
grants of software patents.

2.  Longer term: turn the patent registration system into a genuine
repository of technology, replete with a search system sufficiently
capable that a patent would be invalidated if an experienced
practitioner can't find your patent.  (I'm not sure how to implement
the inevitable arbitration process involved in such a rule of reason,
but I'm confident the courts can handle it with experience and a body
of precedent.  It'll take time, though, thus "long term".)  This would
be funded by registration fees, and thus they would have to be set
pretty high.

I'm not sure whether software patents should exist at all in that long
term.  It's arguable that menus are expressive works.  In fact, I
would go farther, and say that all UI ("look-and-feel") is an
expressive work, and therefore is more appropriately covered by
copyright than by patent.  But Apple would still be in the dock: can
you seriously believe they didn't check out the competition's UIs?


Steve