Subject: Re: Is free software innovative ?
From: "Jonathan S. Shapiro" <>
Date: Wed, 13 Dec 2000 10:47:45 -0500

Apropos Bernard's quest for patent info...

I was interviewed extensively for an article in Johns Hopkins Magazine that
will shortly be appearing. When it does I'll let people know where to find
it. The particular context of the article is, of course, the impact of
patents on academic research.

How much of my comments will survive into the article is unclear, but here
are some points I made to the interviewer:

1. The concept of patent exists to advance the interests of the public, not
the corporation or inventor.

2. Whether you believe in patents or not, the patent lifecycle should be in
line with the product lifecycle. A patent exists to allow the inventor a
reasonable chance to follow through on their invention, not an unbounded
opportunity to hold back the tide or hold the public for ransom. If the
inventor cannot make good within one product lifecycle (as measured by the
appropriate industry), they never will. There are always outliers, but
patent duration must be decided on the "greatest good for the greatest
number" theory.

3. In software, the lifecycle duration is at most two years. The reverse
engineering cycle is also about two years. Patents are unnecessary.

6. In software, the public's interests are advanced by allowing ideas to be
integrated. The entire purpose of patents is to balkanize ideas. Patents do
not advance the interests of the public at all. Rather, they allow large,
conservative companies to hold back progress in the industry by preventing
others from introducing disruptive change.

7. We've seen a number of examples where holders of software patents have
used them specifically to hold up the interests of the public. There is a
long list of software applications that started integrating cryptography in
October, because that's when the RSA patent ran out.Unisys made an
incredible mess out of GIF files when they started abusing their patent on
Lempel-Ziv compression, and *that* patent isn't even valid! Mark Wegman
(IBM) filed a prior patent on the same techniques.

8. The patent office is unable to assess the quality of the applications, so
they basically tend to trust the filer that the patent is innovative. The
results are predictable. IBM, for example, holds a patent on table lookup.
Demonstrating in court that one of these patents is bad costs the defendant
a minimum of $250,000. This is enough that people settle rather than go to
court. If you're a small company, this may be more than you have. The end
result is that software patents have become a form of legally-supported
extortion. The financial burden is on the wrong people. Combine that with
inadequate patent evaluation and you have a real problem. Which we do.

A couple of questions the reporter asked are potentially of interest. I've
reconstructed the discussion, so the comments may not exactly capture the
interview wording, but I think I've preserved the sense of the exchange.

Q: What about drug patents? The drug companies have argued that these are
absolutely essential to their profitability.

[Me:] Let's test that. The drug companies are claiming that without patents
they wouldn't make any money, right?
[Reporter:] yes.
[Me:] When were drug patents introduced?
[Reporter:] around 1980 -- I'll check the year.
[Me:] Were there profitable drug companies before 1980?
[Reporter:] Yes.
[Me:] And they made money without patents, right? There were a number of
large, successful drug companies in the 1960's and 1970's.
[Reporter:] Yes
[Me:] So if those companies want patents, it's not about being profitable.
That's an excuse. It's about being able to restrict advances in medications.
Patents exist to protect the public. Are drug patents protecting the public?
[Reporter:] But before patents, they used trade secrets...
[Me:] Trade secrets aren't bad. What one person can discover and use,
another can too, so long as repressive laws do not stop them.

Q. So how long should software patents run?
A. If we have to have them, I'ld personally argue for 18 months, but I'ld
settle for two years.
[Reporter:] But the lawyers are saying that it takes three years to fully
file a patent, so they argue that anything less than three years doesn't
make sense.
[Me:]You really have to be careful when somebody with a vested interest --
like a lawyer, or a drug company, or a Hopkins professor -- is talking. You
need to ask "What is in it for them?" A patent lawyer isn't interested in a
fair patent system. They are interested in higher legal fees and getting the
best for their client. So here they are arguing "the beaureaucracy can't
handle reality, so lets change reality." Does that make sense?


Jonathan S. Shapiro
Assistant Professor, Department of Computer Science
Johns Hopkins University