Thankfully for the rest of the world, that particular disease is primarily an affliction of the computer industry.
It's not an incentive to make incremental improvements, it's an artifact of the industry's strong need for interoperability in the face of overzealous patent issuing, and seemingly a way for anticompetitive behavior to dodge antitrust scrutiny.
The patent pools come before the standards. If your patent doesn't cover what's in the standard then nobody will use it, and if you're not one of the cool kids then nothing your patent covers will get put in the standard (see also Rambus).
The net effect of those patent pools is that the likes of Apple, Microsoft and Google pay each other amounts that essentially cancel each other out, but anyone wanting to enter the market has to pay a tax to all of their incumbent competitors. Not clear how that benefits innovation.
Most of the patents in the pools also aren't incremental, they're independent patents that cover different parts of the same standard.
> Most of the patents in the pools also aren't incremental, they're independent patents that cover different parts of the same standard.
Is there any analysis that supports to this statement? Having seen many hundreds of patents, many of them in the video encoding area, I wouldn't hesitate to describe most of them as incremental improvements.
Also, the likes of Apple, Microsoft and Google end up paying the MPEG-LA much, much more than what they get paid for their share of the patents in the pool. They certainly do not cancel out, and if they had their way, they'd rather not pay anything at all. Unfortunately, with those codecs enjoying so much popular hardware support, they have little choice but to partake in the pool and pay their dues.
Anyone willing to contribute innovation to the market has a simple option: invent some technology that is valuable enough to be included in the standard, offer their own patents to the pool and get a share of the licensing pie. That way, inventors can do what they are good at (inventing) and vendors can do what they are good at (implementing and commercializing). Classic division of labour.
> Is there any analysis that supports to this statement?
Why are you asking for analysis? Aren't the lists of patents in the pools published? You can look directly at the source materials.
> Having seen many hundreds of patents, many of them in the video encoding area, I wouldn't hesitate to describe most of them as incremental improvements.
Certainly they aren't revolutionary, but you're ignoring the question of whether they're incremental improvements over existing patents or over the public domain. If the question is whether a patent reduces the incentive for others to make incremental improvements, finding an incremental improvement to something that exists in the public domain provides no such evidence.
> Also, the likes of Apple, Microsoft and Google end up paying the MPEG-LA much, much more than what they get paid for their share of the patents in the pool. They certainly do not cancel out
Do you have some evidence of this? For any company whose percentage of patents in the pool is proportional to their market share in the market, basic arithmetic says the royalties should cancel.
> if they had their way, they'd rather not pay anything at all. Unfortunately, with those codecs enjoying so much popular hardware support, they have little choice but to partake in the pool and pay their dues.
These are the companies that make the hardware or who tell the hardware makers what to make. If they collectively wanted to make the patent pool go away they could all just support both for a couple of generations while announcing immediately when support for the old codec will be discontinued so that people can stop using it. For some reason they don't.
> Anyone willing to contribute innovation to the market has a simple option: invent some technology that is valuable enough to be included in the standard, offer their own patents to the pool and get a share of the licensing pie.
But that's not how the standards bodies work. Nobody wants NPE patents in the pool but the NPE, and NPEs don't get a vote in the standards body. There are two primary ways that NPEs get royalties from the pool. The first is patents that cover the standard but nobody knew that when the standard was finalized or they would have fixed it, the classic submarine patent. You can't make money on this except by accident or by dishonesty. The second is patents that are so unreasonably broad they can't be avoided no matter what you do even if you know about them ahead of time. These are a defect in the patent office which shouldn't have issued them. No part of his involves some nice fellows in lab coats making decent wages without doing anything untoward.
> That way, inventors can do what they are good at (inventing) and vendors can do what they are good at (implementing and commercializing). Classic division of labour.
Patent pools have nothing to do with division of labor. They're an evolved response to patent thickets. Division of labor is: you get a patent and you sell it to a practicing entity who uses it to enjoy a competitive advantage. Patent pools only happen when there are so many patents you can't help tripping over them and the industry is trying to make the problem go away as best it can.
> but you're ignoring the question of whether they're incremental improvements over existing patents or over the public domain. If the question is whether a patent reduces the incentive for others to make incremental improvements, finding an incremental improvement to something that exists in the public domain provides no such evidence.
Here's how patents are prosecuted: Applicant files a patent with some broad claims. Examiner usually (statistically speaking) rejects it on the basis of non-novelty and/or obviousness and cites some prior art references, which are usually other patents. Applicant then amends their claims with the minimum amount of limitations needed to avoid the prior art references (because they want a patent as broad as possible) and tries again. This cycle typically continues until the examiner can no longer find a concise enough set of prior art references to cover all the limitations of the claim, at which point the patent is allowed.
This process necessarily means that the vast majority of patents technically introduce only incremental improvements (or, more correctly, differences) over the prior art.
Case in point, randomly selected HVEC-related Apple US patent 7,769,084. It was rejected 4 times by the examiner, the last rejection citing patents 5,986,712 (owned by Thomson Electronics) and 6,310,897 (owned by Sony). Apple prevailed by adding a single 20-word clause to a 3-part claim, which the examiner deemed different enough to allow.
If that is not incremental, I don't know what is. Also, note that prior art owned by two different companies did not prevent Apple from making (and then patenting) an incremental improvement to the art.
Here's Microsoft saying MPEG-LA costs it more than it makes:
I feel like I saw similar statements from Apple and Google but I can't find them offhand. The rough math is easy: your share of the licensing pie is proportional to the fraction of your patents in the whole portfolio. The actual calculation is extremely complex, but you can get a good idea by looking at the lists of patents and their owners.
> ... NPEs don't get a vote in the standards body.
Are you sure? I thought I saw a lot of research institutes and universities on standards committees?
> Patent pools have nothing to do with division of labor.
Patent pools are not directly concerned with division of labor, but patents are. Pools are just a emergent licensing convenience when many fragmented rights are involved.
"The patent pools come before the standards. If your patent doesn't cover what's in the standard then nobody will use it, and if you're not one of the cool kids then nothing your patent covers will get put in the standard"
Also the patents in the standard are often broadly written to cover whole genres of techniques. If your new innovative idea is excluded from the standard and you create a new competing standard, then your new standard is illegal because all the patents in the pool that might affect your new standard are unlicensed.
And the licenses the pool sells strictly apply only to uses of the established standard. There is no license available for your new standard at any price.
That's one of the reasons WebM was crippled for years; it had to design around hundreds of established techniques. Now that Google has spent years and billions on legal fights and patent acquisition MPEG-LA has opened up innovation legally to WebM and future versions might be good.
Your innovation isn't probably backed by Google, so good luck ever getting permission to try it.
And remember that even trying patented techniques in the lab is illegal if you have commercial applications in mind. There is a pure research exception, but if you want to develop a new codec or encryption idea built on older ones, you're skirting the law at best. The incumbents don't especially want to encourage your competition.
http://en.wikipedia.org/wiki/MPEG_LA