20 May 2012

Bickering About Software Patents Misses the Point About Innovation

Software windowsOn Thursday 16 May 2012, IP Australia held an ‘IP Forum’ event in Sydney, on the subject of ‘software patents’.

Regrettably, being located in Melbourne, Patentology was unable to attend in person.  However, past events have been recorded and made available via YouTube, so we are hopeful that this will also be the case for the software patent forum.  Especially because, by all accounts, it was a fairly spirited occasion.

In the pro-software-patent corner was Australian inventor Ric Richardson, whose company Uniloc has recently reached a substantial settlement with Microsoft over Richardson’s patent covering anti-piracy technology, including Microsoft’s online activation system.  The history of Uniloc’s battle with Microsoft has been covered by two episodes of the ABC television program Australian Story: ‘A Done Deal’ in April 2012, and ‘The Big Deal’ in August 2009.

For the anti-software-patent side, IP Australia invited Ben Sturmfels, principal of free software developer Sturm.  Sturmfels is also the Melbourne discussion group organiser for the Australian Free Software Association, which in 2010 gathered over 1000 signatures on a letter to (then) Innovation Minister Senator Kim Carr calling for the abolition of software patents in Australia.

IP Australia was represented by Deputy Commissioner Phil Spann, who is perhaps not especially averse to software patents, unless they happen to involve ‘business methods’.  His most recent decisions as a Hearing Officer include Network Solutions, First Principles, Iowa Lottery, and Invention Pathways.

So the stage was certainly set for a lively discussion.  But is all this energy really being well-directed?  What actual impact are software patents having on innovation?  And is the question of whether of not software should be patentable really the most important issue to which we should be drawing politicians’ attention, or are there greater barriers to the successful commercialisation of innovative technologies on which to be focussing?

THE PARTICIPANTS’ POSITIONS

The IP Forum has been well-reported by ITNews.com.au (IP Australia debates software patents), ZDNet Australia (The world needs patents: Uniloc founder), CIO.com.au (Do software patents stifle innovation?) and by Ric Richardson himself, on his blog.
Phil Spann
As reported, Deputy Commissioner Spann had the task of delivering facts and figures.  According to Spann:
  1. computing-related patent applications have increased moderately since 2003;
  2. this increase has been consistent with an increase in patent applications in general;
  3. computing-related applications account for six per cent of Australian patent applications; and
  4. this compares with four per cent in Germany, six per cent in China and Japan, and ten per cent in the US.
These figures are broadly consistent with the approach to the patenting of software in these various jurisdictions.  Europe and China, in particular, limit patentable computer-implemented inventions to those which use technical means to solve a technical problem.  This effectively excludes much software which addresses, e.g., business or social needs, rather than a specifically technical requirement, such as making a device or network function more effectively. 

Japan is a little more liberal, requiring a patentable invention to be an application of a law of nature

The US, as many are aware, has perhaps the world’s most liberal approach to patenting of computer-implemented inventions, as well as being a large market for the associated products.  It is therefore not surprising that the largest proportion of software-related patent applications are filed in that country.
Ric Richardson
Software patents – one in particular – have made Ric Richardson a wealthy man.  Unsurprisingly, then, Richardson spoke in favour of software patents.  Based on the various reports of the event, his points included:
  1. that he did not see cases of software patents being used by large companies to repress innovation by smaller companies;
  2. on the contrary, a patent may be the only protection the ‘little guy’ has against ‘open slather’ on ideas being stolen by large companies with the means to put them into large-scale production and distribution;
  3. furthermore, patents add value to a small company which makes it a more valuable acquisition target for larger companies; and
  4. regarding copying, Richardson is quoted by CIO.com.au as saying ‘no one who’s a fair thinker thinks that’s an honest way to treat somebody who has come up with a good idea ... Looking back at that whole process, I am so grateful for the patent system.’
Ben Sturmfels
Ben Sturmfels apparently took more of a ‘social policy’ view of the patent system, saying (according to CIO.com.au):

The patent system isn’t to give individual wealth to inventors – it might seem like it is, but that’s misinformation. It’s actually to benefit society, so we need to evaluate the patent system on whether it is actually benefiting society overall in whether it’s a success or not.

Also, Ric’s success generally isn’t feasible for software companies. Large companies have a standard defence against this. They have a truckload of patents and they can bury anyone in endless litigation about this sort of stuff.

According to Sturmfels:
  1. building on other people’s ideas is best practice, and software patents stifle this;
  2. smaller companies cannot compete with big companies which have large patent portfolios, and thus have the capacity to bury the competition in patents;
  3. small and medium companies are the ‘losers’ in the system; and
  4. the software patent ‘minefield’ is restricting the willingness of investors to become involved with small software companies.

COMMENT

The first point to note is that the relatively low percentages of computing-related patent filings (between four and ten percent) reported by Spann suggest that the level of controversy which tends to surround software patents on the internet (i.e. among those with a particular personal interest in the issue) may be somewhat disproportionate to the actual issue.

Patents create monopolies in all fields of technology, but there are few in which objections are raised quite so vocally as in the domain of computer-implemented inventions.
But are software patents really so controversial?
We would hazard a guess that the level of actual concern amongst the wider population is far greater in relation to the controversial issue of gene patents than it is with respect to software.  The gene patent debate has had a great deal of mainstream media coverage, whereas the software patent debate is almost entirely confined to the technology press and various online sites and forums.  And if you were to pick a person at random on the streets, the chance that they would have an opinion about whether or not genes should be patentable would be very much greater than the possibility that they are even aware there is a debate around software patents.

Furthermore, as with genetic technologies, those who actually have experience and expertise in development and commercialisation of software are far from being in agreement about the merits, or evils, of software patents.  If all software developers were opposed to patents – as Sturmfels suggests they should be – patent attorneys would not be contracted multiple times per week by people enquiring about how they can protect their computer-implemented innovations against copying by larger competitors.  Any attorney working in the field will tell you that there are numerous small companies in this country, and elsewhere, which see patents as the only viable protection they have against some larger established player simply ripping off their idea as soon as the web site goes live!
Perhaps it’s just political differences?
As for the diametrically-opposed views of Sturmfels and Richardson, it would be tempting to suggest a division along political lines.  Perhaps Richardson is simply the arch-capitalist, supporting a system that can help the individual get rich, while Sturmfels is of a more socialist bent, concerning himself with the betterment of society as a whole?

In fact, however, the patent system (and intellectual property more broadly) is one area in which division rarely occurs along party lines.  The Raising the Bar Act passed in Australia with bipartisan support in both houses of Parliament, as did the America Invents Act in the US last year.

There is ambivalence about intellectual property on both sides of politics.  Capitalists abhor a monopoly, but wish to ensure that those individuals willing to take risks and create businesses are able to succeed.  Socialists have fewer problems with monopolies as a general proposition, if they produce social benefit, but are less keen on the idea that they should be controlled by private individuals or corporations.

So IP laws are inevitably a compromise, on whichever side of the political fence you sit.
But aren’t patent monopolies supposed to provide social benefit?
In a sense, both Richardson and Sturmfels are correct.  The patent system was created – as Sturmfels essentially asserts – in the belief that it would provide a net benefit to society, by ensuring that there is an incentive for innovation and investment in new ideas and technology in the form of a temporary, limited monopoly to enable a return on that investment to be obtained. 

However, the way in which the net benefit is achieved is – as Richardson argues – by rewarding individuals for their efforts and risk-taking.  The specific wider benefits of this are new products and services for all consumers, and the development and growth of businesses which employ people and help grow the national economy.

There is no doubt that patents work more effectively in some areas of endeavour than others.  However, there are no clear lines dividing the ‘better’ fields from the ‘worse’.  Legislative divisions have been considered, but no solution which would provide an improvement in practice has been devised.  The patent system may be a blunt instrument, which provides a net benefit across the whole economy while performing differently in specific sectors, but to date it is the best instrument we have for the purpose.
Do software patents stifle innovation?
There is, to the best of our knowledge, little factual basis for Sturmfels’ contention that patents are stifling innovation and preventing small companies from succeeding.  A recent article on Delimiter – Australia’s IT startup scene: Blooming, not dying – paints a very different picture of the fortunes of new tech businesses in Australia.  The intelligent, informed comments are well-worth reading, too, with one commenter providing a summary of the tech startup scene in Australia and the US (where, remember, the proportion of software patent applications is far higher):
  1. venture capital funding investments in Australia totalling an estimated $13.2 million, in 14 reported funding events spanning the six-month period from November 2011 to April 2012; and
  2. for the same period in the US, total funding of $5.3 billion over 781 deals.
These figures say far more about conservative, risk-averse Australian investors than they do about the impact of software patents on innovation.

At the same time, it certainly seems that open source software continues to exist side-by-side with proprietary products, while both ‘free’ and ‘commercial’ activity benefits from intellectual property protections.  After all, how would the GNU General Public License (GPL) operate to compel developers to make their own contributions free were it not for the copyrights which give legal teeth to the license terms?  Claims that software developers are being prevented from building on other people’s ideas certainly seem to be lacking in substantive evidentiary support.
Conclusion – we need better incentives for investment, not fewer patents!
While the patent system is not perfect, its flaws are not technology-specific, and there is no clear basis for treating software as ‘special’.  As ITNews.com.au reported Richardson saying, ‘everything is digital these days’.  Any attempt to exclude ‘software inventions’ from patentability is therefore likely to have wide-ranging and unintended consequences, all to solve what seems largely to be a non-problem outside the minds and writings of a vocal minority.

It would be good to see the same level of effort going into lobbying the Australian government to legislate to create a more supportive environment for early-stage investment in technology startups.  Per capita, there is around 16 times the level of investment activity in the US as there is in Australia.  Now there is an issue that really needs to be addressed!

9 comments:

Tyrone Berger said...

A robust IP system should be just that, robust but fair and transparent. You're correct Mark, we always need to address such issues via legislative measures.
The jury is still out on whether patents stifle innovation... 

ozvic said...

Two questions Mark:

1. I do not understand your comment on the GPL. What has copyright got to do with patents?
2. Software is just math. What makes one algorithm different from another and therefore patentable? Is math patentable?

Patentology (Mark Summerfield) said...

I am not sure that your questions are genuinely intended, rather than rhetorical.  Both of the issues you raise are rather well-worn in the debate over software patents: the appropriateness of treating a number of different legal rights under the common rubrik of 'intellectual property' principally by Stallman; and the 'software is math' argument most prominently by Knuth.  I assume that you know this, but I shall also take you at face value and assume that you also actually want to know what I think.

I think it is fairly clear that I have raised the GPL as an example of the way in which intellectual property rights can be used to support the objectives of open source proponents.  While patent rights are of a somewhat different nature, patents and copyright have the common feature of 'property' that they can be owned, sold and controlled to the benefit of the holder.

In any event, section 11 of GPLv3 extends the effect of the license to cover patents as well as copyrights.  In principle, an open source patent license could go further, requiring even an independent implementation of the patented technology to be open/free.

On your second point, and with all due respect to the great man, Knuth is simply wrong.  Software is not 'just math'.  Mathematics is an abstract science, generally expressed using an agreed symbolic system (of which there are many).  Software, on the other hand, is a functional entity which causes a hardware-based system to do something, often useful and sometimes commercially valuable.

If software were 'just math', none of us would have any need or desire for it.  Computer programs would execute away inside machines, silently processing abstract mathematical concepts, and never impacting in any manner upon the outside world.  It would be theoretically interesting (to academic computer scientists like Knuth, presumably) but of no practical use whatsoever.  Certainly nobody would be able to build any kind of business based upon it!

I imagine that Knuth knows perfectly well that it is one thing to say that software systems can (in principle) be described completely using mathematical formalism, but quite another to say that 'software is math'.  This is not, in fact, what Knuth said.  His particular sleight of hand (or thought) involved the insertion of the concept of an algorithm between the practical software application and the abstract math: 'every algorithm is as mathematical as anything could be. An algorithm is an abstract concept unrelated to physical laws of the universe.'  This may be true, but every practical (and patentable) implementation of an algorithm involves action upon information representing objects or concepts in the real world.

Imagine two toasters, outwardly identical in appearance and operation, but internally different in that one uses conventional electrical circuitry (resistors, capacitors and inductors) for all of its timing and electromechanical functions, while the other employs software executing on an embedded microcontroller.  Why should one be patentable in principle, while the other is not?  Is one a toaster, while the other is 'just math'?  And if Donald Knuth were hungry and in need of breakfast, would it matter to him whether his toaster contains software or not?

You might just as well argue that nothing should be patentable, because we have good reason to believe that mathematical formalism can be used to describe completely any physical process which takes place in the universe.  But practical applications of the 'laws of nature' are patentable.  Indeed, this is the actual legal definition of patentable subject matter in some countries (e.g. Japan).

Purely mathematical algorithms and formulae, divorced from practical application, are not patentable.  By the same token, someone sitting at home manipulating indefinite integrals need have no fear that by doing so they are infringing upon anybody's intellectual property rights!

Patentology (Mark Summerfield) said...

Thanks for your comment, Tyrone.

I think the jury will always be out.  The problem is, we have no 'control' universe in which the patent system was never developed with which we can compare the outcomes.  Now that we have a global patent system, dismantling it is a practical impossibility.

There is a deeper philosphical question, which has from time-to-time carried some weight (most notably, perhaps, in France following the Revolution), which is whether we human beings have a fundamental right to ownership of our intellectual creations.  If so, then laws would be required to protect against intellectual 'theft'.

This is not, however, the policy underpinning the modern IP system.  We do not have IP rights because people are entitled to them as some form of 'human right', but because (to use the language of the US Constitution) they are believed to 'promote the progress of science and the useful arts'.  Outside the US, such lofty ideals are rarely expressed, with IP rights being viewed as one of a number of instruments of economic policy control.

iskryabin said...

I software is just math, than  computer chips are just physics, materials
are just chemistry and genes are just biology. All of them are "just math" when
described using mathematical formulas. 
Creating a  “new state of affairs”
is critical whether or not it was achieved 
by transformation and manipulation of data. 

guest said...

some interesting issues from the article:

1. capitalism vs socialism. I think a true socialist policy would seek to
abolish the patent system altogether not just software patents. The half-arsed
socialist approach from Mr Sturmfels is basically another form of capitalism in
which, innovation from one sector of the economy is suppressed because some
disgruntled interest groups were not receiving a share of the economic rewards
from an inventor’s hard work.



2. Open source vs proprietary software. Why do some
people argue that the coexistence of open source and proprietary software is
not possible? I use both open source and proprietary software and I have never
had any problems in spending money on proprietary software if it has the
functionalities I want or need.




3. Do software patents stifle innovation? Do software
patents prevent researchers and inventors from making improvements or innovate
from granted patents? Software developers have donated software to charities
therefore any innovation from existing software could be offered as donations.
There is no law to prevent software developers from donating their products for
a charitable cause, is there?

Nickmiller said...

This is a great article, thanks, I learned a lot.  (Typo: 'principle of free software developer'.)

Patentology (Mark Summerfield) said...

You're welcome, glad you found it useful.

And thanks for pointing out the error (which I have corrected).  I try to save my typos for blog posts -- they're not a good look in client correspondence!

Flash2011 said...

I know that this link is old but even Jeff Bezos thought 17 year (after issue) patents for software did not make sense (he proposed 3 - 5 years):

http://oreilly.com/news/amazon_patents.html

Post a Comment