I’ve been doing a lot of research recently into the patent system. The
impetus is the history-making
Peer to Patent project,
and specifically an article I recently finished for the Economist magazine. (“A patent improvement” in the September
8, 2007 issue, Technology Quarterly section.)
This blog is not about the Economist article or Peer to Patent, but
about how several months of immersion in the subject of patents has
affected my understanding of related policy issues, and how my
opinions differ from many of my friends in the software field.
Patents are necessary to promote invention
One of the fuzzier notions making its rounds in the software field
condemns patents as backward and oppressive. Not just software
patents–some people criticize the whole notion of patents.
What you have to remember is that every major innovation–even in free
software–tends to be driven by a highly-motivated visionary. Just
because a company may fund the project (or a community may provide bug
fixes and enhancements, in the case of free software) doesn’t mean the
role of the wild-eyed genius is eliminated.
Sara Boettiger shows in a
First Monday article
that patents are by no means incompatible with open source
development; in fact, they can protect openness. What if you create an
open source technology and someone patents an improvement to it? If
you have a patent on the base technology, you can bargain with the
person who patented the improvement, because he or she needs your
patent license to use the improved version of your software. That’s
one reason the
Open Invention Network
Open source is great, but it has worked so far in areas where people
can innovate sitting comfortably at a computer. They don’t generally
have to writhe around on the floor trying to insert a bolt in a
hard-to-reach place or mix dangerous concoctions of expensive
chemicals, which is how a lot of significant innovation takes place in
other fields. (However, the locus of innovation is shifting, an
observation I’ll take up in a moment.)
A good idea is a precious find, and not all of us are fated to make
the discovery. The inventors need both a guarantee of recognition and
the hope of financial reward (which sometimes has to be massive) to
continue the amazing roll-out of new inventions we’ve seen over the
past several decades.
One can speculate endlessly on what drives inventors to put in their
effort, but it appears they are spurred on by the opportunity to
own their invention, both in the sense of being recognized
publicly and in opportunities to make money from it. Indeed, the
patent system is valuable because it offers recognition and financial
reward while persuading inventors to give up some “ownership” and
share the invention with the public.
Patents that are improperly granted or exercised with too much
reckless force can cause harm and lead to high legal costs. Similar
things could be said about police who act improperly. As with patents,
some people (particularly among racial minorities) say they’d rather
not have police at all. But the abuses just remind us that powerful
forces need to be carefully regulated, and abusers have to be
Buying and selling of patents is required
Not everybody can take an idea from conception to market. Do you know
an inventor? Now do you know an inventor who can assemble the means to
produce the invention on a massive scale at high quality? If you do,
can that inventor also set up a company that’s expert at branding,
marketing, and distributing the invention?
All but a handful of independent inventors need to assign their ideas
to other companies in order to bring them to fruition. Even if a
company owns an invention, the managers may realize they’re not the
best company to manufacture or market it. Furthermore, innovation is
based on the recognition that an inventor deserves to be compensated
for his or her creativity even when others bring the idea to market.
So technology transfer is a fundamental part of the robust ecosystem
that allows inventions to reach the public in the form of
cost-effective products. And given that patents have an economic role
in encouraging inventions, I see a role even for the companies that
buy patents with no intention of manufacturing the invention. They may
deserve the accusation of trolling if they strike a manufacturer
strategically when it’s vulnerable and has to agree to loathsome
terms, but if they simply demand the license fees that would normally
be due the original inventor, they’re part of the same capitalist game
as everyone else.
It’s easier to understand licensing if you think of innovation in the
form of risk. Inventors may risk several years of their lives
developing inventions, and manufacturers risk money in developing and
marketing the resulting product. Assigning a patent to a patent
holding company is also risky, because the patent may be worth a lot
more than what it’s sold for. But it also may be worth a lot less, so
patent holding company is also taking a risk and deserves a comparable
The one risk inventors can’t tolerate is for someone else to get a
patent on something that’s obvious or already in widespread use.
That’s where we should focus on improving the patent system.
If policy changes tightened requirements for getting a patent,
facilitated the overturning of improperly granted patents, and reduced
the inflated damages that are awarded for infringement in many cases,
we’d reduce the specific “trolling” problem of companies that ask for
license fees greater than the invention is worth.
Software represents the current stage of innovation
This may be the most controversial observation in this article, and
the one whose implications I am least sure about. Technological
innovation has always represented the embodiment of a creative spark
in a machine or process–you might say that what people are patenting
is a mental process for manipulating some physical or
But technologists have explored the physical worlds pretty thoroughly.
Certainly we’ll see more new compounds–and the world of biology is
wide open for exploration–but it’s natural for a lot of innovation to
have moved to a higher level. Now that we’ve invented a lot of
physical or chemical embodiments of mental processes, innovation
continues among those mental processes themselves. And insofar as
mental processes are embodied in software, one can argue that patents
on software will promote further inventions.
I don’t buy this argument, even though I’ve tried to state it in a way
that’s fair to proponents. Correspondingly, many of the arguments
against software patents are flawed.
The complaint that a lot of non-novel processes were being patented
because software patents were new was a powerful argument thirty years
ago, but it’s getting old as the repository of software patents piles
The complaint that software is complex is unsustainable because lots
of non-software machines and processes are complex too. Is most
software more complex than an oil refinery?
The argument against software patents that I find most persuasive is
the fundamental unity of a program. An inventor can always
identify the particular hinge she has added to a machine, or the
particular molecule she has added to a compound, making it unique.
There is no such construct in software. Everything a piece of software
does draws on data and code scattered throughout a program.
One can strip a function out of a well-designed program and substitute
another, but that doesn’t mean one can identify what makes one
function novel in the sense of being patentable. The difficulties of
designing modular software (well documented in dozens of books on
software engineering over the past four decades) show that in fact
software is a unified entity that is not merely the sum of its parts.
The damage caused by the enforcement of overly broad patents is often
called a “boundary” problem. (See, for instance, work by
and Michael J. Meurer.)
I consider the unity of a software program to be the ultimate
expression of this problem.
Another way to state this point is that an idea is of little value,
and that the real innovation lies in the embodiment of the idea in
code. Protection of software by copyright is therefore sufficient to
protecting an inventor’s rights, and patenting software is overkill.
Tim O’Reilly, my employer, looks back toward the original social
contract behind patents: that the patent holder gets the exclusive
rights to develop a product in exchange for teaching the public about
the new technology. He interprets this tradition as demanding that
software not be patented unless the patent-holder publishes a working,
open-source implementation. (But the patent would still be restrictive
because it would rule out the release of other implementations, which
might be better or address different application domains.)
The fundamental dilemma in innovation is that it requires such a high
investment of time and money, with such a tenuous hope of reward, that
it’s amazing for people to undertake the effort at all. A legal and
economic system that offers enormous financial rewards to the inventor
who has benefited the public is a success. But such a system also
increases income disparities. The money spent by people on new
medicines and other patented technologies is one factor in a complex
equation that seems to funnel wealth to a tiny segment of the
However, it should be possible to tweak the system without reducing
the rewards for innovation. One can hope that a more equitable
distribution of wealth will create more people with the time and
education to pursue inventions. There’s nothing wrong with a patent
system that acts as a tax whose proceeds go to inventors, so long as
undeserved patents are kept to a minimum and the public can afford the
I’ll finish with one comment about Peer to Patent. The most common
objection to it is, “They won’t get people to sign up and
participate.” I think the project has broken through this barrier to
feasibility, having gotten over 1,300 people to sign up during the
first two months and having received 57 suggestions for prior art
covering 14 patent applications (as of an email I got on August
22). This success may, of course, reflect the first flush of
excitement over a new idea, as well as the choice of computer-related
patents for the pilot.
However, I could very well see participation in patent review becoming
one of the volunteer activities that research communities come to
expect of their members, just as they now review journal articles, sit
on the boards of non-profits, and perform all sorts of
university-related duties unrelated to their fields. Most people who
achieve notable success in a particular field lead public lives and
take on public responsibilities, and the whole message of Peer to
Patent (enthusiastically picked up by the US Patent and Trademark
Office) is that the public can take on some responsibility for