Wednesday, November 16, 2022

Beware of pseudoscience: actual numbers of narrowband IoT-related standard-essential patents are hard to come up with, and keyword searches are inherently unreliable in this context

The Internet of Things (IoT) is a tremendous growth area in multiple respects: for product makers, for standard-essential patent (SEP) holders, and for patent analytics companies looking to acquire new categories of customers. But in the midst of a gold rush, not all that glitters is really gold.

Last week there were two IoT SEP-related announcements: Sisvel's new narrowband IoT pool, which started with 20 licensors, most notably also Ericsson; and just one day earlier, IPlytics released a report on who is supposedly winning the IoT SEP race. As an IP and antitrust commentator, I wish to help my readers avoid being misled. That's why I criticized a ranking of German patent litigation firms last month, and suffice it to say I received a lot of positive feedback from people who appreciated it. And there are serious issues with that IPlytics report, which I believe someone has to call out.

Prior to this one, I publicly criticized IPlytics only once, and that was when Tim Pohlmann interviewed the president of an Apple astroturfing operation claiming to represent small app developers and IoT companies while actually working against the interests of small innovators, especially in the App Store antitrust context. When that interview took place, there was enough information out there already to know that ACT is not a legitimate representative of whom the claim to speak for, and IPlytics should (or must) have known that.

Now IPlytics has been acquired by RELX (congratulations!), a company known for such services as LexisNexis. I have no problem with RELX, and I have no position at this stage on whether WIRED is right that RELX's practices represent a threat to data privacy.

That recent IPlytics report on IoT patent holdings is nothing that decision makers, whether in the public sector (such as competition enforcers and policy makers) or in companies (patent holders, pool administrators, or implementers), should rely on. I have problems with its methodology and with some of the results.

If it wasn't free, I'd have to say: caveat emptor!

First, the methodology makes it a clear Daubert case (those familiar with U.S. litigation know what I mean, and the others can figure).

The report mentions a "keyword approach" three times. On the penultimate page, there is a disclaimer:

"We used a keyword approach without additional filters for the active or granted status of a patent or patent family. We want to highlight that other keywords or additional selected filters might result in other ranking positions and shares. Further, we refrain from making assessments of the technical relevance of patent portfolios."

A "keyword approach" cannot make up for a problem facing anyone (not just IPlytics) who will undertake to evaluate the narrowband IoT patent space: there are no databases of NB-IoT- or LTE-M-specific declarations. The reason being that NB-IoT and LTE-M are simply true subsets of the wider 4G/LTE standard, and that's where the declarations and concomitant FRAND pledges were made. Those narrowband standard-setting efforts started when 4G/LTE was already far along, and what they did was to select those parts of LTE that they deemed suitable to task.

Keywords are a poor indicator, and actually they are no indicator when many patents were filed before it was possible to know or even just predict the relevant keywords. Many 4G/LTE patents read on narrowband IoT, but when the applications were drafted, the keywords simply weren't floating around yet.

The sad truth is that one would have to get down almost to the level of claim charts to really find out who owns how many narrowband IoT SEPs. That's obviously not going to happen.

Second, there are some implausible results. And some of those implausible results are plausibly attributable to the methodology issues I just outlined.

I don't have a plausibility problem with Huawei and Qualcomm coming out on top here. Those two companies have powerful portfolios and long-standing strategic interests related to IoT. It's hard to think of any wireless technology where they wouldn't be very strong. My problems are mostly at the level right below Huawei and Qualcomm:

I would normally assume that companies like Ericsson, Nokia, Samsung, and LG are roughly on an equal footing (Ericsson probably being the number one among them)--to give you a ballpark figure, I would expect each of them to hold around 10% of the relevant patents. Those are companies that clearly have not only a general interest in IoT but also participate quite actively in the related standard-setting. But in IPlytics' narrowband IoT ranking, Nokia is roughly at a level with a company like Lenovo, which has a rather different focus and business model.

Below companies like Ericsson and Nokia, I would see players like Sony and NTT DoCoMo, yet they presumably own more relevant patents than a number of companies listed ahead of them.

Maybe IPlytics did its best and it's not good enough (yet). So what are the alternatives?

The first alternative in a context like this is simply the Socratic approach. In his Apology, Greek philosopher Plato praised the key quality in Socrates not to think he knows what he doesn't know. "I know that I know nothing."

It would be terrific to have a set of numbers that would enable us to get an idea of what the licensing costs for the full stack might be. That would enable top-down royalty determinations, which are so very popular. But we may have to live with the fact that, at least at this stage, we just don't have the data at hand for that. And when data is unreliable, it's better to just accept that fact. Things get worse--not better--if we act in defiance of that realization.

There are bogus medications for incurable diseases. They raise false hopes and have adverse effects--as do unreliable rankings, even those produced with the best intentions.

I wouldn't underestimate the ability of the market to figure out solutions and royalty rates, especially in narrowband IoT, where different standards are competing with each other for "design wins."

And I would strongly recommend to use common sense. While one can't assume that companies' contributions to a standard-setting process always correlate to a given player's portfolio strength, we should at least look at who really participated in standardization and take that into account, as we all know that those who sit at the standard-setting table will also ensure that some of their technologies are included and will strategically file patents so they read on the relevant standard.

Is more transparency needed in the narrowband IoT context? Absolutely. It should be a priority for the industry to come up with something better, ideally in 2023.

To be clear, I am not taking a position here on any other IPlytics report than the one I specifically mentioned, nor on their raw data or anything else they offer.