IEEE after the patent policy update

A number of technology-contributing IEEE members have now refused to be bound by the IEEE's new patent policy adopted in February 2015. The question is, what does that mean for the IEEE as a Standards Setting Organization (SSO)?

In my experience, the strength of the international competitive-collaborative style of telecommunications standardization - largely pioneered by ETSI - lies in the fact that individual technical contributions compete purely on technical merit. This means that the type of competence needed for the effort can be highly specialized; basically some of the very best research engineers in the field, from all over the world, get together to create these complex standards. This ensures cutting-edge, future proof, global standards.

This in turn is possible when the SSO's patent policy has been generally agreed by all members on a high level, and patent policy issues can therefore effectively be "removed" from the daily standardization work itself. Those specialist engineers can focus all their efforts on inventing new clever solutions and arguing about the technical pros and cons of each other's proposals. And importantly, they can spend zero time thinking about patent policy.

But in the IEEE today we have a fragmented situation where a subset of its members have publicly stated that they will not adhere to the IEEE's new patent policy. Here some readers may of course wonder why those members aren't then simply "thrown out" of the IEEE's SSO work? But actually, in order to make the work of an SSO - not only the IEEE - as inclusive as possible, contributors are allowed to declare on a case by case basis which patent policy their patented contribution would be licensed under, which in fact does not have to be the policy adopted by the SSO. It's just that the norm has typically been that all members have agreed with the SSO's policy. But in the IEEE we now have an anomaly. The existence of these "disagreeing members" creates some distinct challenges to the IEEE's standardization effort going forward, at least within 802.11 ("WiFi") standardization.

Prior to the new policy's acceptance in Feb 2015, disagreeing members had submitted - and the IEEE had accepted - certain so-called Letters of Assurance (LoA) endorsing the old IEEE policy for certain 802.11 standards. LoAs are documents describing under which conditions the submitter agrees to license its potential SEPs for a specified standard, i.e. essentially whether the existing IEEE patent policy is agreeable to the submitter or if another policy will be used instead. Now, with the assumption that accepted adherence to a given IEEE patent policy cannot be changed retroactively, SEPs relating to the IEEE standard(s) specified in those pre-Feb-2015 LoAs will therefore be subject to two different patent policies - the old and new IEEE patent policy - depending on who owns the SEP in question.

Some of those standards are already finalized, and in those cases the impact will largely be limited to the licensing and dispute resolution spheres. However, some standards are still being worked on, meaning that there can be an impact within the SSO work itself. There could be an increased risk of proposals not being selected purely on technical merit, but potentially also based on whether the submitter supports the new or old IEEE patent policy. Thankfully, standardization/research engineers' natural behaviour is to focus on technical merit, but one still can't rule out that their management could convince them to behave differently. 

Furthermore, regarding potential SEPs for IEEE standards not specified in those pre-Feb-2015 LoAs, new LoAs would have to be submitted. But a disagreeing member submitting a new LoA would - by definition - tick the box stating its disagreement with the (new) IEEE policy. So at least if the disagreeing member's proposed solution would technically outsmart all other proposals, there's a potential issue to resolve. Assuming that technical merit continues to be the yardstick, that proposal should be selected for inclusion into the IEEE standard in question. But in this case the IEEE would instead have to examine the LoA and decide whether it would be acceptable or not. This would generally have to be done on a case-by-case basis. And if the LoA would not be acceptable to the IEEE's members, the technical proposal itself can't be accepted. Or in other words, the admittedly best technical solution would have to be actively de-selected from the standard (!). 

There has been some speculation as to whether Qualcomm - one of the disagreeing members - could benefit from its recent acquisition of Cambridge Silicone Radio (CSR) in the sense that CSR had made a "broader than usual" pre-Feb-2015 blanket LoA covering all IEEE 802.11 (WiFi) standards. Qualcomm could then allegedly utilize the CSR LoA to license also its future SEPs according to the old IEEE policy at least for all future versions of WiFi. If so, it would at least save Qualcomm from having to make those new "potentially difficult" LoAs for IEEE 802.11. While that could reduce the probability of its proposals being rejected due to origin, the risk will still be there given that Qualcomm, after all, doesn't adhere to the new IEEE policy. 

Clearly, scenarios as described above threaten to remove the "pure technical bliss" of the classic international competitive-collaborative standardization effort within the IEEE. There may now have to be more patent engineers and patent attorneys involved in the SSO work processes. The efficiency of the IEEE as an SSO will likely suffer, and alarmingly, the IEEE's future standards run the risk of not containing the best technology out there. Even worse, the best technology may have been made available to the IEEE, who then actively rejected it. If such fears materialize, some companies might very well leave the IEEE and instead support - or perhaps even create - other competing standards through other means.

The new IEEE patent policy's apparent disconnect with the law is problematic. Only last week in its CSIRO vs Cisco appeal decision, the US Federal Circuit squarely rejected one of the key principles of the new policy, namely the mandatory chipset royalty base. In the same breath, the Federal Circuit re-emphasized the importance of evidence in the form of comparable licenses, also in stark contrast to the new policy. We also know that the new policy's rule on injunctive relief in practice also differs from Federal Circuit opinion. 

The IEEE, and ultimately the general public, would surely benefit the most if the IEEE stays strong and relevant, capable of efficiently creating top-notch standards going forward. So perhaps it's time for a pragmatic and open discussion among IEEE members with the aim to consolidate the different viewpoints in the light of the law. Another update of the policy may very well be needed to put the IEEE's standardization work back on track again.

Cellular FRAND royalty levels

What's a fair and reasonable royalty for a cellular Standards Essential Patent (SEP) portfolio license, and how is it determined? This sounds like a simple enough question, but actually it's not. In this post I'll share some of my experiences and thoughts in relation to this much debated subject.

To tackle this subject, I believe one should start by assessing values. Let's first look at the value of the cellular standard to a mobile phone. To say that the cellular standard itself is absolutely crucial for a mobile phone is hardly controversial. In fact, the cellular capability defines the phone as a commercial product: an iPhone wouldn't be an iPhone without cellular capability. Apple is able to sell its iPhone at a price of several hundred US$ more (!) than the price of its iPod, a very similar product but without cellular capability. This price difference demonstrates the tremendous value of adding standardized cellular technology to an otherwise similar device.

So yes, the cellular standard is definitely fundamental to a mobile phone. But what about a cellular SEP portfolio, what's the value of that? Before looking at a given SEP portfolio though, let's start by considering the cumulative. What's the license value of all SEP portfolios taken together. After all, that's the ultimate concern of the Original Equipment Manufacturer (OEM) selling the product. How is this "total mass" of cellular SEP portfolios related to the standard?

To address those questions, let's first assume that we'd like to create a brand new cellular 5G standard as a successor to 4G, but - for whatever reason - we want it to be completely void of SEPs. What could that correspond to in practice? Well, since patents live for up to 20 years, only standardized technical solutions older than 20 years would be completely "safe" to include. So we might actually end up selecting the first (1990) version of GSM (2G) as our 5G standard. This certainly feels like a very bad idea indeed. But why exactly? Because a 2G standard can't possibly fulfil the technical requirements placed on a 5G standard. The example might seem a bit extreme, but I want to illustrate an important point. The value of a standard strongly relates to new pieces of technology added to it. This is because a new cellular standard is subject to more harsh technical requirements - with respect to supported data rates, delay, capacity, flexibility, new services, efficient data compression and so on -  than previous standards. Having own experience from standardization, I know that these new requirements are there to make the new standard as future-proof as possible. While many consumers may accept to change phones quite often, mobile network operators make tremendous investments in infrastructure and will hardly be amused about having to replace it after only a couple of years because it's gone obsolete. In order to fulfil these tougher requirements of a new standard, brand new technological innovations are very often required, and these are also typically patented. So, while all cellular SEP portfolios together may not be strictly equivalent to the standard itself, they represent the value of the standard since they include the technologies that allow the standard to fulfil its specific requirements.

So how about the license value of all cellular SEP portfolios in terms of actual money? And who decides it for that matter? Here I'd first like to point out that the consequences of "wrong" pricing can be tremendous. If too high, then consumer demand will suffer. If too low, then incentives for standards innovation will suffer. But in fact, we can all see mobile device sales and usage expanding strongly and new state-of-the-art standards continuing to be created. So clearly, the market must have somehow managed to settle at a balanced pricing. But at what level? In my experience from working with cellular SEP licensing for quite some time, I can see an established cumulative royalty cost for all cellular standards up to 4G somewhere around 10% of the OEM selling price of a mobile phone. Note that this is not my opinion of what it "should" be. This is the reality that I see.

I'd like to make two important comments in relation to this 10% figure:

1) High-end phones often have advanced screens, cameras etc. that may be argued to be unrelated to the cellular standard. According to such argument, the percentage-principle leads to an unfairly high royalty for these products. But in reality, those attributes actually increase the usage of the cellular standard. In fact, the 3G and 4G standards were even specifically developed with the performance of such devices in mind. So the value of at least 3G and subsequent standards - and hence of all SEP portfolios relevant to those - is higher for high-end phones. Nevertheless, high-end phones are still often subject to "royalty caps" in order to address the perceived concern. Conversely, at the other end of the spectrum, the percentage-principle may cause very-low-end phones to devalue SEP portfolios. In such cases "royalty floors" may be used as well.

2) Most major SEP-holders are practicing entities, i.e. they sell standards-compliant products themselves. When such an SEP-holder offers a license to its SEP portfolio to an OEM, it typically also needs a license to the OEM's SEP portfolio. This is known as a "grant-back" license. In these cases, there will be a "royalty-netting" effect. One party pays a net royalty to the other based on their respective SEP portfolio strengths and net product sales. Therefore, SEP-holding OEMs ultimately pay a lower net cumulative royalty rate than non-SEP-holding OEMs. Importantly, this is not a violation of the Non-Discriminatory ("ND") part of FRAND since the value of a license is transferred back in place of some royalties. An OEM either makes large investments in basic R&D - resulting in innovations and patents - or pays full license fees. After all, there are no free lunches. 

Some OEMs new to the market have recently started to argue that the cumulative royalty should instead be no more than a fraction of the cost of a "baseband chip". This is a component inside the mobile phone that includes cellular connectivity functions. Such a change would actually make the cumulative royalty about an order of magnitude lower than what it is today. This argument doesn't make sense to me because it doesn't go to the value of using the cellular connectivity, but rather to the cost of a component necessary for that connectivity. Patent damages, i.e. the royalties determined by a court, are based on the value for the use made of the invention (e.g. in the US according to 35 USC 284).

As a parallel, one could look at a movie DVD. The entire component and manufacturing cost is probably no more than US$0.50. Yet nobody would seriously argue that the IPR value for the product as a whole, including the movie script, production, direction, screenplay, music, acting and so on, should be a fraction of that amount. It's the value of the IPR to the end product that's important. Arguing for a 90% de-valuation of cellular SEPs when there's been an established and well-working SEP valuation regime in place for decades seems irrational to me and would certainly disrupt the market balance, and for sure not in favour of innovation.

Having discussed the cumulative royalty at some length, I can now return to the original question: What's a fair and reasonable royalty for a given cellular SEP portfolio license, and how is it determined? Or in other words, how much should a given SEP-holder obtain within the cumulative budget and how is that fraction actually decided?

Most market participants agree that there has to be some form of proportionality at play here, in keeping with general principles of fairness and reasonableness. SEP-holders with "stronger" portfolios are entitled to larger royalty fractions. But using what metric? Counting SEPs? Valuing the SEP portfolios in terms of "innovation value"? This is not an easy question. Still, the actors in the market - the SEP-holders and established OEMs - over time "get to know" the overall value of each other's portfolios. They do this through license negotiations as well as by gaining knowledge of each others' R&D, standardization activities and patenting quality. The main challenge arises for someone who's inexperienced with this market.

Take as an example a willing licensee OEM new to the market and without standardization involvement or SEPs. It needs to allocate around 10% of its product price to cellular SEP license royalties, but how does it determine the fractions? Are there ways to find at least estimates of the distribution of those 10%? Well there are indeed some commonly discussed methods, which I'll briefly explain here:

1) ETSI IPR database declarations
Due to its apparent simplicity, it's very tempting to look at various patent holders' SEP declarations on ETSI's IPR database and to estimate royalty fractions based on them. However, such results are actually totally unreliable, since neither the declarer nor ETSI has any obligation to verify essentiality. The original purpose of the database was to ensure that there would be no hidden surprises in terms of unknown patents turning up at a later stage as SEPs. For this reason, patents "believed" to be essential could be declared, even though they might turn out not to be. As a result, significant over-declaration regularly takes place. This leads to gross over-estimates of the total SEP holdings and distorts the distribution of SEPs. So at best, this kind of study may be used to get a picture of potential SEP-holders rather than estimates of actual SEP holders' relative fractions.

2) Third party evaluations
Some instead turn to SEP evaluations performed by third parties. Unfortunately many such evaluations tend to be based on "keyword searches" in patent databases, which make them about as useful as horoscopes. More serious evaluation attempts actually try to look at the patent claims in some detail and compare them to the standard. Although those evaluations could in principle give indications of actual SEP-holder fractions, they also tend to suffer from reliability restrictions. Besides the obvious concern of bias, their accuracy is also limited due to time and cost constraints in combination with the large mass of potential SEPs to go through.

3) Accepted standards contributions
A newer method, originally explored by Ericsson, is not to look at patents at all (!), but instead at "accepted standards contributions" to the 3GPP standardization process. That essentially refers to technical solution proposals accepted into the standard. It's assumed that this metric has a strong correlation with SEP-ownership over time and that significant standards contributors all have similar "patenting rates" and "patenting quality". Since these assumptions are actually quite reasonable and the contribution data is all publicly available, this method is potentially accurate. Interestingly, it also aligns itself with incentivizing technical contributions to the standard. However, with this method becoming more widely used, its accuracy may suffer in the longer term since some standards-participants can be tempted to make numerous proposals on relatively trivial technical solutions solely to increase their "accepted contribution rate".

In practice though, these kinds of estimations are mainly used only as tools to roughly gauge the SEP portfolio landscape. Our new OEM will obtain the most accurate information about the SEP-holders' portfolios and FRAND royalty requests simply by conducting licensing negotiations with different SEP-holders in parallel. In this way the OEM can scrutinize the different SEP portfolios down to whatever detail it sees fit. This may be done through examining and discussing so-called claim-charts detailing the patents' relation to the standard in question. If needed, OEMs sometimes even employ external experts for such evaluations. Over time, these negotiations typically do result in a fair and reasonable royalty partitioning.

In this context I'd also like to point out that OEMs don't always fully appreciate the restrictions that the non-discriminatory ("ND") part of FRAND places on the SEP-holder. Many SEP holders have a multitude of already signed SEP portfolio license agreements in place and simply can't discriminate against existing licensees in favour of a new OEM. It's also inappropriate for a new OEM to compare itself with unwilling licensees - i.e. those essentially not wanting to pay anything - in terms of how much royalty it should pay. Doing so further exacerbates the "patent hold-out" problem discussed in my previous post.

In this post I've tried to describe the realities surrounding cellular SEP royalty levels for mobile phones, based on my own SEP licensing experience. Under the FRAND regime, I believe that the market has managed to strike a balance between innovation value and consumer value and established a cumulative royalty rate at around 10% of the mobile phone OEM selling price for all cellular standards up to 4G. I'm aware that some still believe that cumulative royalty rates are too high, that there's too much litigation going on and that radical changes to the system are urgently needed. I'll end this post by addressing these issues one by one.

Are cumulative royalty rates "too high"? I can't really tell for sure, but at least I don't see any obvious support for that notion. As I mentioned in an earlier post, consumer choice and new future-proof cellular standards hardly seem to be lacking. Also, among the top-10 global mobile phone OEMs of today, virtually none of them had sold a single mobile phone only a decade ago, and some of them didn't even exist a that time. So at least entry barriers are low. Margins may be getting lower for some OEMs, but that can be an effect of fierce global competition thanks to those low entry barriers (!) and perhaps partly even the result of patent hold-outs, i.e.some other OEMs paying  - ironically - too low (i.e. zero) cumulative royalties.

Is there "too much litigation"? Well, with some high-profile litigation cases going on in recent times, one might think that most cellular SEP portfolio license negotiations end up in court. But actually, this is a classic example of an availability error - "if you see it, it must be common". The truth is that most such negotiations are successfully concluded in good faith, far from the limelight, and only a small fraction end up in court. Actually, the most visible litigation case, Apple-Samsung, is fundamentally not about SEPs at all but rather about Apple's attempts to protect its proprietary smartphone user interface functions.

Are changes to "the system" needed? Well no system is perfect, so improvements are always needed. For example, new ideas for improving the transparency of the various fractional ownerships of the total SEPs could be welcomed, especially by new OEMs.  But I can't see an obvious need for any radical changes, since the FRAND technology sharing regime largely works the way it's supposed to work. Policies should continue to encourage successful cellular SEP portfolio licensing through good faith FRAND negotiations, which is still the norm today as it has been for decades. It's important that the cumulative rate and SEP portfolio proportionality principles are respected by all market participants. Any policy adjustments should work within the current framework and in a balanced way continue to make sure that it's difficult for individual SEP-holders to seek too much royalty as well as for OEMs to hold out for too little royalty.