Perhaps rather than assert that I missed the point, maybe it is more accurate to say that I disagree with your interpretation.
I do think I got the point of non-linearity. To my mind, Nokia's assertion is that the value of patents in a patent portfolio is distinctly non-linear. Which means that in any patent portfolio, a very small handfull of patents are truly very valuable, and the rest have negligible individual value, but might, if numerable, add value by virtue of being numerous.
In other words, a portfolio of 1000 patents is not necessarily worth 10x the value of a portfolio of 100 patents, and one single patent may be more valuable than an entire portfolio of hundreds. Furthermore, the reality is that the vast majority of value in any portfolio arises from a scant handful. All it takes is one. But even having an "essential" patent does not mean value. To wit, Qualcomm's "Essential" GSM patents are no longer worth the paper they are printed on.
What I also think Nokia is saying is that when it comes to valuing a handful of patents, it is very difficult. And I happen to agree with this, since it lines up with my own experience.
I also see that Nokia is saying that three patents together may indeed be worth a few percent, but really most of that comes from only one of the patents, some from another, and less from another. That is entirely reasonable.
But I don't see anywhere that Nokia is suggesting that one can charge X% for three patents used by one manufacturer and Y% for the same three patents used by another manufacturer.
Even Qualcomm charges the same price for all of its patents for all of its licensees. If it were true that a collection of patents had different value for different manufacturers, then through the process of negotiation over hundreds of licensees, Qualcomm would have ended up with a distribution of realized royalty rates that reflects this intrinsic inter-manufacturer difference in values. In Qualcomm's case, this distribution is extremely narrow, and indeed Qualcomm points to this narrow distribution as evidence of FRAND pricing "Everybody pays the same price for all of our patents".
I don't see either Qualcomm or Nokia arguing that if a company charges one rate for a set of patents to one company, it can turn around and charge a different rate for the same set of patents to another company. Actually, what evidence I see (above) is more supportive of the opposite.
Therefore, those three hypothetical patents, together when priced in a fair and non discriminatory way are either worth X% or Y% but not both.
Once we have determined that these three patents are worth, say, X%, certainly no two of these patents are worth more than X%, at least not on a "non-discriminatory" basis.
Furthermore, the only way any two of these patents could be worth X% is if the third patent is worth 0%.
Therefore, if we ascribe Z% to this third patent, then clearly, by de-construction, the other two patents together are worth X%-Z%, which is less than X%.
Conversely, if we determine that the two patents together are worth A%, then by de-construction we know that the value of the third patent is Z% = X%-A%.
And so on...
Nothing stops the relationship in value between these three patents from being "non-linear". One of the three patents could be worth 2%, the other could be worth 0.5% and the third could be worth 0.05% and the total portfolio value would be 2.55%, with most of the value coming from the 2% patent.
This same logic applies to buckets of patents, recursively.
In the end, once we start putting values on subsets of patent portfolios, we are forced to expose the internal nonlinearities. Qualcomm might as well paint a bullseye on their patent portfolio for all the problems this begets. |
