Tuesday 20 September 2011

Software Patents: a Convenient Misnomer for those who Seek to Expropriate IP

In this, the seventhin his series of posts for IP Finance, Keith Mallinson (WiseHarbor) reviews the recent history of software patent protection and the challenges made against it, concluding that the patent system is here to encourage investment in innovation by helping enable inventors to make a return on their risky investments and arguing that there is no evidence that patent systems are stifling innovation where inventions are implemented in software.

You can follow Keith on Twitter @WiseHarbor.
Software Patents – a Convenient Misnomer for those who Seek to Expropriate IP 
It makes no sense to disqualify innovative technologies from patentability or limit the rights and remedies associated with those patents on the basis they can be implemented in software on general purpose processors rather than only on dedicated hardware. The “software patent” debate is largely a battle of ideology and business models between those who develop patented technologies that can be implemented in software and implementers who would rather not pay for the privilege of using others’ IP. I focus exclusively on technologies in this article because a large and rising proportion of manufactured products are increasingly software defined. Patentability for “business methods”, such as financial trading algorithms, while also contentious, is an entirely different matter. 
Generosity at others’ expense 
Google has made itself popular from the promise of free software with its Android smartphone operating system (OS) and WebM project with VP8 coder-decoder (codec) for video and Vorbis codec for audio. This promise is as in free beer (i.e., something for no payment) rather than merely free speech (i.e., being allowed to say what you like). The proposition obviously seems very appealing to many implementers, including software developers and device manufacturers, who like the idea of getting something for nothing. 
However, this proposition is tricky because many software programs infringe the unencumbered rights of IP owners who justifiably do not wish to give away the fruits of their labour for nothing. In Free and Open Source Software (FOSS) the “free” refers to the freedom to copy and re-use the software, rather than to the price of the software. A fundamental requirement with Open-Source Software (OSS) is that “licenses shall not require a royalty or other fee”. 
Whereas these licences generally require licensees to contribute their patented and copyrightable works royalty free, that is far from sufficient to ensure (F)OSS implementations will actually be completely free of charge to licensees. FOSS licenses are private contractual orderings that have no impact on the obligations of those IP holders outside any given contract’s reach.  Many IP owners decide not to sign away their rights in (F)OSS licenses and others may be oblivious for a long time that specific (F)OSS software programs are infringing their rights. Despite efforts to prevent (F)OSS programs infringing un-liberated IP (that is, IP held by third parties outside the reach of the FOSS license), it is impossible to ensure this will not occur – particularly with respect to patents. 
 (F)OSS licensees may be found by courts of law or agencies such as the U.S. International Trade Commission (ITC) to be wilfully or otherwise infringing IP, with resulting legal costs, financial damages awards and even injunctions or exclusion orders preventing them from selling their products. Some of these licensees might not have expected this due to misleading statements from (F)OSS proponents and given that patent infringement was typically not a problem with packaged software, sold under license from the likes of Microsoft, that has prevailed for decades on PCs and elsewhere. Indemnities – derived from cross-licensing among various IP owners and commonly provided to licensees of proprietary software – are rarely available or as extensive with (F)OSS.  In fact, attempts by either IP owners or FOSS distributors to enter into license agreements with third party IP holders have often been deemed antithetical to the FOSS movement (or event in conflict with the terms of FOSS licenses) and so they have, until recent months, been the exception rather than the rule. 
 Until very recently, Google appears to have provided little or nothing more than rhetorical support for its beleaguered Android licensees who are signing patent licenses or being sued for infringement or by proprietary software providers Apple with its iOS, Microsoft with Windows Phone and others. On the receiving end of the onslaught are HTC, Samsung and others implementing this open source OS. Perhaps Google will assist in various counter-suits following its recent purchase of 1,000 patents from IBM and acquisition of Motorola Mobility with a trove of 17,000 patents. 
 Free riders infringe 
Tensions are running high between IP owners and those who shun paying patent fees for anything implemented in software including standards-based technologies. Already 12 patent owners have joined discussions to create a pool to collect royalties from those that implement the VP8 video codec standard. VP8 is based on technology developed by the Google acquisition On2 for its WebM project. This is purported to be completely royalty free (5th September 2011):

“Some video codecs require content distributors and manufacturers to pay patent royalties to use the intellectual property within the codec. WebM and the codecs it supports (VP8 video and Vorbis audio) require no royalty payments of any kind. You can do whatever you want with the WebM code without owing money to anybody.”
Whereas there is no reason to prevent VP8 being developed free of any copyright or patent fees to any of its developers who agree to such terms, the codec is most likely infringing the patents of these 12 and many others. Non-assert provisions in VP8 licensing anticipate that Google has essential patents --and licensees might too. Different, independently developed, programs will likely not infringe software copyrights, where code is not copied, but all codecs implementing a given standard will infringe the same set of patents that are essential to that standard. Software developers, by definition, cannot design around essential patents when implementing a standard. Similar (or “competing”) standards may well have common technologies among them which are also covered by the same patents. This is particularly the case in Codec algorithms, which represent cumulative technological developments made over many years, including many players and at substantial costs. Different codec standards setting organisations (SSOs) can try to design around patents in formulating their standards. While this is possible to some extent, it is difficult, and impossible to eliminate all infringements while also seeking to achieve high-performance functionality exploiting latest technologies. In some cases, SSOs might not even be aware of some patents their standards are infringing.
 MPEG LA licenses the H.264 video codec extensively. More than one thousand licensees have agreed royalty terms compensating 28 different licensors through a patent pool. These fees are due even where the software program implementing the codec is subject to royalty free copyright licensing, as is the case with the x264 – “a free software library and application for encoding video streams into the H.264/MPEG-4 AVC format, and is released under the terms of the GNU GPL [a royalty free licensing agreement]”.
 With other codecs reading on hundreds of patents and significant similarities among codecs, it is also most likely VP8 infringes some of the patents that are also infringed by other video standards including H.264. The question is simply how many patents and which of them are infringed? 
 Changing the rules  
Meanwhile, the patentability of any technologies and algorithms implemented in software are being significantly challenged with lobbying to policy makers around the world. 
Those who argue against “software patents”, including some absurd and unsubstantiated claims, seek to invalidate issued and pending patents associated with, for example, smartphone features and video codec standards. Others have suggested that the perceived problems with “software patents” could be remedied by requiring that those patents be licensed on a royalty free basis in certain contexts (i.e., in standards). The fact that many standards-essential and other technologies implemented in software infringe numerous different patents, rather than typically just a few patents in a drug or simple mechanical device, is no justification to deny any patent rights at all. A combination of bilateral (i.e., cross licensing) and multilateral arrangements (i.e., with patent pools) can be used to negotiate rates and collect payments efficiently. The average aggregate royalty for video codecs on a DVD player is just a few dollars and aggregate standards-essential patent licensing on mobile phones rarely costs more than 10% of the wholesale product price. Moreover, the unsubstantiated claim that FOSS developers are prohibited by the terms of FOSS licenses from paying these royalties has been debunked and shown to be little more than an attempt by certain implementers to gain business model advantage. 
Processors and software in everything  
The products and services we all use every day are increasingly software defined and computer-intensive as microprocessors are included in many different manufactured items. Software predominantly implements the innovative algorithms for a wide variety of technological functions; from touch screen scrolling and bar code reading to turn-by-turn navigation. Just a few of numerous and varied examples also include anti-lock brakes, eco-friendly air conditioners, medical equipment, programmable lathes and toys. 
The existence of microprocessors and computers over the last 30 years has fostered a marketplace for downstream development of computer programs performing a wide variety of functions with relatively low barriers to entry. For example, there are thousands and thousands of smartphone application developers.  Many of these set themselves up with just a computer and a few software tools in their sitting rooms or dormitories. 
Computer technologies with general purpose processors are increasingly substituting for application-specific designs. In some cases, state-of-the-art general processors make it possible to implement technologies (e.g., radio interference reduction, video compression or touch screen gesture recognition) significantly in software, in comparison to the more hardware-specific implementations such as with Application-Specific Integrated Circuits (ASICs) that were once required.  Mobile communications protocols including GSM, HSPA and LTE can now be implemented in Software Defined Radios (SDR)s. SDRs are already commonplace in network equipment and increasingly in terminal devices such as phones and dongles. Similarly, whereas older codec implementations were significantly in hardware with dedicated signal processors and hardware accelerators, it is now possible to implement these in general processors with customised hardware and accelerators being used mostly for high-end devices. 
Substituting software for hardware implementations of a given radio or codec technology is a design decision driven by considerations on feature performance, power consumption, heat dissipation, semiconductor die size, time-to-market and fixed versus variable manufacturing cost structure.
The speed, ease and low costs of coding in software— rather than having to design and fabricate dedicated hardware— does not negate the innovative steps, substantial costs and risks entailed in developing new ideas and technologies, regardless of their means of implementation. For example, development of anti-lock brakes requires lab work and drive testing under various conditions and medical instrumentation techniques (e.g., measurement of oxygen saturation in the blood) requires lab work and extensive clinical trials. Algorithms are first conceived, then modified and refined to improve performance, reliability and safety on the basis of this work. Software just happens to be an efficient and effective way to implement.
What is patentable?  
So-called “software patents” do not actually depict software per se: instead they describe algorithms and processes that can be performed by a programmed computer. It is such computer-implemented techniques— not the software itself—that can be eligible for patent protection. 
 In Information and Communications Technology (ICT), it is the underlying useful, novel and non-obvious techniques that can be implemented in hardware or software to perform real world functionality—such as in radio communication, audio noise reduction, video encoding, and touch screen operation—to name just a few possibilities —that are potentially patentable. To be patent-eligible in the U.S., generally, a claimed method must involve a machine or a transformation of an article—that is, it must describe a series of steps that use physical means to produce a result or effect in the physical world. All the above examples and many other technical processes do just that – whether they are, or could be, implemented in hardware or software. 
  In 2002, the European Commission proposed a Directive on the patentability of computer-implemented inventions, but the European Parliament rejected the final draft with the result that national laws were not harmonised. The European Patent Office, which generally adapts its regulations to new EU law, has no reason or incentive to modify its practice of granting patents on certain computer-implemented inventions, according to its interpretation of the European Patent Convention and its implementing regulations. 
 Copyrights protect software owners from having their programs duplicated, but this does not prevent reverse-engineering of software-implemented innovations. Similarly, it is increasingly possible to implement previously hardware-based functions such as radio modems and video codecs on more general processors such as SDRs and with software-based rather than hardware-based graphics accelerators. It would be nonsensical to disqualify patented innovations from protection, simply because independent advances in processor and software technology make the former implementable on general purpose processors as well as dedicated hardware. 
 Openness and patents in standards 
Whereas some assert that open standards should be royalty free, the International Telecommunications Union defines open standards, among other factors, as follows: 
"Open Standards" are standards made available to the general public and are developed (or approved) and maintained via a collaborative and consensus driven process. "Open Standards" facilitate interoperability and data exchange among different products or services and are intended for widespread adoption. 
Intellectual property rights (IPRs) – IPRs essential to implement the standard to be licensed to all applicants on a worldwide, non-discriminatory basis, either (1) for free and under other reasonable terms and conditions or (2) on reasonable terms and conditions (which may include monetary compensation). Negotiations are left to the parties concerned and are performed outside the SDO [standard- development organisation]. 
There are numerous open standards. However, IP policies differ widely among standards-setting organisations (SSOs). A relatively small number of SSOs have IPR policies that require participants to license essential patent claims on a royalty-free basis, but this can only bind those who elect to join those organisations and so standards implementers can be exposed to IP infringement claims by non-members. Most SSOs including those for mobile communications, video and audio codecs accept that patent owners can license their IP on a (Fair), Reasonable and Non-Discriminatory basis, including a royalty. 
For example, H.264 is open in the sense that the specifications are freely available from a copyright perspective. One can distribute an implementation of H.264 freely as long as one abides by certain terms. However, implementers of the H.264 standard are required to pay patent royalties. 
Software is no exception 
There is no good reason to abandon the widespread practice of allowing patents on technologies that are implemented in software. The patent system is there to encourage investment in innovation by helping enable inventors to make a return on their risky investments. There is no evidence that patent systems are stifling innovation where inventions are implemented in software. On the contrary, innovation continues apace in ICT, as illustrated by the rapid development and extensive adoption of smartphones and video encoding technologies, to name just two from among numerous examples, as I have explained in my previous articles with IP Finance.


Anonymous said...

The European Patent Office, which generally adapts its regulations to new EU law

It does? Just curious, when did EPO do such a thing?

Anonymous said...

"FOSS developers are prohibited by the terms of FOSS licenses from paying these royalties has been debunked"

Freedom of copying and royalties are not compatible with each other.

If you give freedom of copying to your neighboors, you cannot count the number of copies they might make.

Bastian Best said...

Great article!