Last week's big announcement by the UK government was principally about procurement, detailing the new rules that will apply when government departments acquire software. Naturally, then, it concentrated on the details of that approach, and how it would be deployed and enforced. A key part of that was using open standards to create a level playing field for all companies, regardless of whether they offered open source or proprietary code.
As I explained in my post last week, the critical issue then became what exactly "open standards" meant, and, specifically, how standards that might be encumbered by patents would be dealt with. As I've noted many times before, the only way open source can implement general interoperability standards is if any claimed patents are licensed under royalty/restriction-free (RF) terms. Although that's the preferred mode for key Internet organisations like the W3C, it stands in contrast to the older approach, which was based on "fair, reasonable and non-discriminatory" - FRAND.
The problem is that licensing can be perfectly fair, reasonable and non-discriminating, but nonetheless incompatible with open source code. Typically, FRAND requires a per-copy payment, but for free software, which can be shared any number of times, it's simply not possible to keep tabs on just how many copies are out there.
Some supporters of FRAND licensing have suggested that there are examples where open source has been reconciled with non-RF approaches. I've explained before why those aren't relevant, and do not contradict the general point that FRAND is not compatible with free software. Interestingly, one of the ancillary documents published last week by the UK government, specifically addresses this point:
We recognise that adopting a policy which mandates RF standards could reduce choice, as the vast majority of Standard Setting Organisations have at least one option that allows patent holders to charge on a FRAND basis. We appreciate that this could potentially limit the pool of standards available to us however, the evidence from the consultation showed few examples of what we would be excluded from doing by adopting an open standards policy based on a royalty free definition.
Of the two well known cases cited about FRAND licensing, one case (Canonical and MPEG-LA) relates to hardware which is out of scope of this consultation.
The other involved Red Hat and Firestar and was settled by the payment of a royalty of an undisclosed amount by Red Hat to Firestar as part of the settlement of infringement claims. This is a solution we do not consider to be feasible in most scenarios, as there are too many unknowns around achieving lump sum deals and having an organisation which would be willing to make a payment on behalf of other users. It might also have the effect of locking out companies.
The UK government's preference for RF rather than FRAND is not only a great victory for open source, since it means that procurement will henceforth be on a fair basis, but is also part of a larger movement away from FRAND to RF in the field of software.
One reason for this shift is that however attractive FRAND might sound – who could possible be against something that was "fair" and "reasonable"? - in practice it's not so simple. That's evident in the recently-dismissed litigation between Apple and Motorola, which hinged on what exactly "fair" and "reasonable" means when it comes to licensing technologies in standards. Because, of course, one company's "fair" is another company's price gouging.
The fact that FRAND is so ill-defined means that in an age of increasing litigiousness, it gives too much scope for argument. You don't need a crystal ball to predict that there will be more and more court cases centring around what exactly "fair" and "reasonable" means, as companies increasingly seek to gain competitive advantage in the courts, exploiting any loopholes that they can find.
Against this background, only royalty/restriction-free licensing can offer any real security. If the promise is that no rights will be asserted against anyone using a particular technology that is used in a standard, then everyone knows where they stand. There are no loopholes, and no ways in which clever lawyers can seek to wriggle out of respective commitments.
That's one pragmatic reason why FRAND is likely to fade away, and RF to take over. Another is more profound, and concerns the current shift from traditional software to open source.
Where the former is generally licensed on a per-copy basis, and is therefore compatible with FRAND-type fees that are calculated in precisely this way, the widespread move to software that can be shared freely means that this is less relevant. Moreover, the rise of cloud computing has weakened the idea of a well-defined number of copies further, and with it that of per-copy licensing.
In this respect, the passage from a per-copy regime for patent licensing to one where technology is made available freely, parallels what is happening in the world of copyright. There, too, the old idea of a well-defined and readily-identifiable copy of a book, or of a CD, no longer has much meaning in a world where digital files are shared and multiplied endlessly. Keeping track of these is probably impossible, and certainly not sensible. Instead, new business models need to be found that use the fact that works are being shared freely to generate revenue, just as RF-licensing in standards still allows revenue to be generated, but in new ways.
This underlines why it is beneficial to grant RF licensing of technologies that are included in open standards. Even though the company developing that technology may not receive traditional royalties, it does enjoy other advantages. One is that its technologies are likely to be widely used, which means that it will find itself right at the centre of the ecosystem that develops as a result. As well as enhancing its influence there, it is also more likely to develop ancillary products first, since it knows the underlying technologies better than anyone.
As we know, in a networked world, this first-mover advantage is important, since in a given sector there is typically only one dominant product, and the company that establishes itself early has a better chance at preventing subsequent entrants from achieving significant market share.
Once traditional supporters of FRAND licensing realise that they may well benefit from RF, I expect that there will be a rapid shift from FRAND to RF as the default licensing for most computing standards. In this respect, the UK government's recent move may well prove not only to be great news for open source and UK citizens, but also to be in the vanguard of a much larger global trend.