Lenovo seeks to render Nokia's H.264 patents unenforceable | Hacker News Reader

Lenovo seeks to render Nokia's H.264 patents unenforceable

theregister.com

226 pointssohkamyung111 comments

The public interest is that interoperability standards should not require patented technology at all to implement. The point of an interoperability standard is to allow a broad ecosystem with many producers and many consumers. Here many producers of videos, many producers of video-processing and video-displaying technology, many consumers of these videos and associated technology. The public interest is that the ecosystem should be vibrant, with easy access to new participants and low barriers to entry.

So, in the public interest, go Lenovo! Especially considering the slime-ball tactics that Nokia apparently used here.

Indeed, give them the rubber boot!

It’s not a very good system but in theory this is what RAND licensing schemes in standards bodies are for.

> looks at HDMI[a]

But you’re absolutely correct. Something like H.264 which is used by practically everything shouldn’t cost money to just look at the standard.

[a]: I think it’s like a US$10k “membership fee” just to have the privilege of looking at the standard

> Something like H.264 which is used by practically everything shouldn’t cost money to just look at the standard.

While I generally agree, it sounds a bit like "if you're successful, you're not allowed to earn money". It's a piece of work, still, and the creators should be allowed to earn money with it.

I'd personally prefer if all standards are free, since especially with an obscure one you're SOL when it's locked behind a paywall; with H.264, you can probably find a way around. But its hard to combine that with intellectual property. Limiting the patent timeframe to five or ten years might be a good middle ground.

That is a very valid argument. I think the big problem (for me at least) is that these patents are (essentially) on math, just “done on a computer”. Sure, it’s math that would be impractical to do by hand, but it’s still just math.

That’s the rub for me.

You don't think mathematical ideas should be patentable?

>While I generally agree, it sounds a bit like "if you're successful, you're not allowed to earn money".

No, it's "if you're successful by NORMALLY ILLEGAL COLLUSION, than the ways in which you're allowed to earn money may be restricted." That's what "encumbered standards" are at the end of the day, explicitly established horizontal monopolies. Those mathematics patents covering H.264 wouldn't be worth remotely the same if there were a hundred different video formats going around of which only 1 or 2 used them. Or if everyone had just stuck with MPEG-2. Of course, the world would also overall be worse off in those cases too, with an even higher risk of vertical monopolies, wasted hardware/bandwidth, etc. So it makes sense to have a middle ground, to allow "standards" to grow the total pie non-linearly larger than it would have been otherwise. But at the same time that needs to be combined with much more stringent controls to avoid all the normal obvious risks of monopolies.

Actors like Nokia and Qualcomm have been trying to have their cake and eat it too. They like the enormous monopoly lock-in value the standards give their patents, but then they also want to treat them like normal individual patents they can cut deals around. That's not how it should work. The reward for a Standards Essential Patent should be a reliable, fair fixed cut of a very, very big and dependable pie for the life of the patents/pool. Companies have the choice to go it alone if they'd like instead and not join the standard, which can then seek to work around them (or if that's impossible, disband and give up and at least not form any monopoly there). But if they have patents in a standard, they should face more restrictions due to the extra monopoly. It should be one or the other, not both.

The reward is that if you are successful, you get to interoperate with other people. That is the value that is unlocked, and it is a pretty huge value.

The problem with patents and standards is that people are trying to charge for that value (the value of the network effects), and not the value of the technology. The big ideas of video compression are mostly more than 30 years old now, but the new patents are mostly on slight tweaks or additions (or things like header flags and other nonsense). No one would pay money for them except for the fact that they are part of the standard.

SDI is just so much better than HDMI though, single cable which you can make yourself, really cheaply, to be 36cm, or 3m97, or 28m14, or whatever.

What does HDMI offer than SDI doesn't?

You’re not wrong, but the biggest thing is wider support. I’ve only seen SDI available on expensive camera gear (but I don’t look that hard), but HDMI is ubiquitous. It’s sad DisplayPort didn’t end up winning instead, but I guess TV manufacturers are members, so they’d put it in the TVs. DisplayPort still has a stupid “membership fee” but at least it’s royalty free.

SDI doesn’t have the built in encryption so you’ll never see it on something that supports playback.

DRM really does ruin everything, eh?

The DDC interface, which is an I2C bus that allows for bidirectional communications to read EDID data, so you know what you are plugging into. Hot plug detection via the HPD line, so you can tell if you are plugged in or not. Embedded 5V power, very useful for conversion dongles, ethernet, probably more things I forget. HDMI has a lot of pins, which makes DIY cables difficult, however they are useful.

You (your hardware) can tell if SDI is plugged in if the line is terminated with 75 ohms (SDI allows looping through to multiple drops)

I can't think of any time I've ever used anything on a HDMI that uses power -- my HDMI to SDI converters (or vice versa) have separate power so it can't be very reliable and/or powerful

I guess EDID data can be useful in the computing space for feeding back from the monitor.

H.264 is publicly available for free from ITU's website: https://www.itu.int/rec/T-REC-H.264

It's annoying because multiple standards bodies all publish the same spec, which is why H.264 is also called "MPEG-4 Part 10" because the spec is part of the MPEG-4 spec published in ISO/IEC 14496 (part 10 for AVC). The ISO specs aren't free (a couple hundred USD for the AVC spec), which is the most annoying part.

You can get the H.263/H.264/H.265 specs for free form ITU, which covers a lot of the details you need for writing an encoder/decoder. Unfortunately sometimes you need the ISO specs for other things related to these codecs (e.g., ISO BMFF stuff).

(edit: I'm a dork and originally misread the comment as saying it was $10k for the H.264 spec, when parent meant the HDMI spec; I removed parts of my comment that were related to that misreading).

ISO/IEC 14496-10 is actually one of the free standards:

https://standards.iso.org/ittf/PubliclyAvailableStandards/

Presumably this is because H.264 is public. Your complaint is well founded, though: if you've built an AVC codec using public documents you may be frustrated to learn that ISO/IEC 14496-14 (describing the MP4 file format) and ISO/IEC 14496-15 (describing the AVC elementary stream and how to encode it as an MP4) are not public.

It's also worth noting that ISO removes public standards. You may find references to public standards, like the link at https://ffmpeg.org/doxygen/2.7/webmdashenc_8c_source.html – but neither the linked ISO/IEC 23009-1:2014 nor any other version of the MPEG-DASH standard is available to the public at this time. Pay up!

You can of course find a lot of those paid standards in the usual places... ;-)

IMHO the generally lax attitude towards IP is one of the reasons responsible for China's strong technological growth.

AV1 is an open codec that I hope will be the industry standard soon. It already has the support of the ecosystem (apple, google, qualcomm, netflix, etc). Unfortunately, it just takes a while to build that momentum

So there is light at the end of the tunnel

> (apple, google, qualcomm, netflix, etc).

And out them, only Apple, and Qcom are hardware makers, and both have dropped the ball on AV1.

Qcom removed AV1 from latest snapdragon, and Apple haven't put it into M1 despite being one of first in the AV1 crowd.

The momentum is lost, and thus is the war. H265 is already a new de-facto standard by the virtue of it getting first to the hardware, while AV1 crowd was apparently arguing over whose logo design would AV1 bear, and exact legalese wording.

Both the new AMD and NVidia graphics cards and the intel CPUs support hardware decoding of AV1. I wouldn't count it out just yet. x265 has been released for many, many years and yet it's not fully adopted by a long shot. Transitioning to new technologies takes time and a step back doesn't proclude a step forward again, in the case of Qcom.

Especially considering AV1 is free, more efficient than x265, and pushed forward by powerful actors.

Hardware decoding is nothing more than a nice to have on desktops. On mobile, it’s mandatory. HEVC is the only format you can be assured will work there.

And AVC. And VP9.

VP9 (and VP8) are only hardware accelerated on Android. On iOS, you only have AVC and HEVC.

And Windows, macOS and Linux.

No, definitely not on macOS. It uses the same VideoToolbox framework iOS does. You could probably write a kernel extension to add VPx support yourself tho.

macOS Big Sur supports VP9 hardware decoding through VideoToolbox.

TIL

> And out them, only Apple, and Qcom are hardware makers, and both have dropped the ball on AV1.

Dropped the ball implies that they made a mistake. Are you sure it wasn't premeditated?

“Drop the ball” also just means an announcement. In colloquial usage, it has a negative connotation, so it’s usage here to mean “announce removal” makes sense.

I'm a bit confused.

My point was that "drop the ball" means that they made a mistake. I don't think they did, it was intentional and I'd say even malicious.

Why do you think that this was not the case?

Qualcomm is no supporter of AV1 so they shouldn't have been included in that list. That being said, there's plenty of other hardware makers and licensors in the Alliance.

> Qcom removed AV1 from latest snapdragon

From what I can tell, they haven't had AV1 support in any chip yet. They said it didn't make the cut this time which probably means they just haven't finished and verified yet. Likewise, I'd guess Apples engineering efforts are more focused on upcoming, larger designs so they can finish moving off x86. Once those ship, they can refocus (plus it's another reason to upgrade your machine).

Rollout for most codecs is slow. I remember people saying no vp8/9 hardware decoders was deliberate too.

In truth, streaming companies want to save billions in data and licensing costs. Modern mobile SoCs can decode in software reasonably well, but with worse battery life. Netflix isn't paying for that, so they aren't incentivized to care past the most atrocious battery hogging behavior (it's not like you'll be streaming their exclusives from elsewhere). This leaves Qualcomm to implement it or get worse performance reviews.

Disagree. Most large app devs directly care about battery life these days.

>In truth, streaming companies want to save billions in data and licensing costs.

They dont, the cost of transferring those bit are minimal. They are much more likely to be in the very low end of millions.

Internet Content Distribution Licensing are Zero from all known HEVC licenses.

Implantation of hardware codec takes time especially for VP8/9 and AV1 which tends to have changes after their so called finalised version 1. Not something I remember happening with H.26x

There are specific power budget requirement for hardware decoding in Mobile SoC. Unlike Desktop / Laptop GPU and iGPU where you can afford to have 2W to decode a video.

Qualcomm is still the last important company who doesn't join AOmedia.

> Apple haven't put it into M1 despite being one of first in the AV1 crowd.

Except they weren't. They just paid more money to be considered a "founding" members - assuming that's where you got the idea.

You seem to muse as though AV1 were on the same development track as H265 but delayed. This also doesn't reflect reality. It started well after and had a relatively quick turnaround (as far as codec development goes).

Right I remember for a long time early in the development efforts for AV1 Apple was conspicuously absent.

H265 is an older standard than AV1. It's closest relative is VP9.

The reason H265 has momentum now is because it's been around for longer. AV1 already has a higher efficiency than H265.

The standard that is competing with AV1 is VVC or H.266. As it stands, AV1 has much more momentum than H.266.

Nope. AV1 has demand due to the companies behind it. I’m 95% confident it’ll start happening within the next couple of years.

There’s so much engineering knowledge that has to be built up behind how to make these things fast in hardware since the overlap with other mpeg codes is smaller. There’s also a market timing issue. AV1 will be more battery intensive so the consumer demand for AV1 has to make sense (right now the content ecosystem isn’t there). That means Google and Netflix have to transcode their entire libraries. That doesn’t happen often and I wouldn’t be surprised if there was some investigation into how to do that more cheaply as the current encoders are unusably slow at the moment (ie they may need to put video encoder hardware in their clouds which I don’t think I’ve heard of these companies ever doing, making the undertaking even more expensive)

I find it funny how people try to declare H265 a winner when it will most likely never land in Chrome or Firefox. Safari has its user base, especially on mobile, sure. But it will never gain any momentum on the web until Chrome supports it. Which will never happen, as Google fully stands behind AV1.

> …it will never gain any momentum on the web until Chrome supports it. Which will never happen…

That's what everyone said about Chrome and H.264, too. As it stands, Chrome has supported H.264 for over a decade.

https://caniuse.com/mpeg4

Happily, Google's ability to simply declare new standards only goes so far.

Only the brave and rich companies can use H.265 now, because of the patent. There are two major patent holders and there are some companies not belonging to either of them. Using H.265 to anything other than experiment and research in my country(as the patent law stated these for the exception of patent use) has a risk we cannot take.

I guess rest of us, honest patent-abiding citizen and low profit buziness have to wait until the middle of 2030s to use H.265 safely.

Confused why you listed Qualcomm as I don't believe they have anything to do with AV1.

Indeed they don't appear on the member list: http://aomedia.org/membership/members/

ARM is listed as a founding member, though.

> (apple, google, qualcomm, netflix, etc).

Has Apple made any investment into AV1 at all? Meanwhile they actively use and deploy h.265 encoders/decoders & content.

These standards aren’t just “plug C is 0.25 inches wide.” They are intimately tied to the compression technology itself. Developing new standards involves developing new compression technology, building and testing reference implementations, iterating, etc. All that costs money to develop. That will get paid for one way or the other. What way that happens affects the public interest analysis.

Look what happened with web standards. Developing and testing new web standards requires building reference implementations in browsers. Browsers are expensive to develop, so the standards are controlled by a handful of companies like Google and Apple. But the standards are free to implement. Is that better for the public? Maybe, but the public interest analysis is more complex than “free to implement.”

There is no such thing as “free.” R&D will be monetized somehow—the question is how. Different monetization approaches have different and non-obvious impacts on the public interest. Consider for example the shift away from paying for software to giving it away “free.” You used to pay Symbian $5-10 bucks per phone for an OS. Android you can get for free. But Symbian didn’t build an advertising platform into the core of each phone. Monetization matters.

I'd be more sympathetic if the money were mostly going to the engineers and researchers who did the work, not a bunch of corporate fat cats.

My point has nothing to do with sympathy. I’m talking about the impact on the public of monetizing R&D one way versus another.

Additionally, the companies pay the salaries of engineers and researchers who did the work. If your point is that engineers and researchers don’t get paid enough for the value they create, you’re talking about a completely distinct issue that has more to do with the structure of corporations generally.

It takes more than engineers to make a standard. The organizational infrastructure, investment in those engineers, associated risks if the investment doesn't bear fruit, engagement with the capital markets necessary to provide the financial infrastructure to keep the blood flow of money pumping... There's a heck of a lot more going to make any moderate R&D project happen. Why is the engineer more important than the many people required to make the organization function in a way that makes the engineer's work possible in the first place?

Maybe people at the top get too much of the compensation, but what about the network technicians and IT departments that provided a functional work environment? What about building maintenance? What is harder, the research involved, or learning & navigating knowledge of the legal and financial mechanisms necessary for large complex organizations with a thousand interlinking pieces to other organizations? Who should get more compensation when the absence of any one component, not just the researchers, might have caused failure?

I want to be clear though: I'm not defending astronomical compensation of the very few at the top. I'm saying that if there's inequity here, it doesn't begin or end with the engineers and researchers. Its victims are the sum total of all the people, all of the functions, without which the system fails.

You're making more of a criticism of advertising as a business model than of free software.

AV1 is supported by Google, sure, but also companies like Netflix and Apple. If Google's business model for YouTube was to charge a subscription fee to everyone like Netflix, wouldn't they still have the same incentive to develop AV1?

This is really commoditize your complement. Which works regardless of the revenue source in the primary market.

https://www.gwern.net/Complement

> You're making more of a criticism of advertising as a business model than of free software.

I’m saying that if your business model is free software, it needs to be monetized some other way. Advertising is one way, which has become common. How that happens needs to be part of the public interest analysis.

Well, fully funding developers is one option.

Rather than monetize the software, fund its creators. They, along with everyone else, get great use value out of the software and they are empowered to live, live, do, play, and generally be.

Valve is a great example of this. Aside from just shipping Proton as a single layer to manage the various compatibility libraries involved with playing Windows games on Linux, they've also hired multiple developers that were already working on projects like DXVK and Zink because of how their platform benefits from the underlying enhancements to Vulkan as an ecosystem.

Who pays? And where do those entities make the money?

Users and supporters do. And they make the money to pay however they usually do.

They may also make money using the software to do whatever it is other people will pay for. That is the case for one project I contribute to on Pateron. Use value is real high and a community of users is funding the developer.

I am supporting a couple of other projects that way right now. I've got good developers on Patreon, and I'm happy to know that they're developing the software that I value. That's the end of it.

I'm also happy to know that software is out there for general use too. I don't really care if everyone's paying. Would I do care about is the developers are getting paid.

I think more developers should communicate where they are at in terms of support, and I think a whole lot more of us need to put small amounts of money toward making sure those developers are okay.

> if your business model is free software, it needs to be monetized some other way

Mutual self-interest is a good, solid business model. JPEG was developed long ago by a bunch of companies who needed a good, standard image format. Now AV1 is being developed by a bunch of companies who want to use less bandwidth for the video and will save truckloads of money not paying patent license fees. Not to mention standarization, economies of scale, and interoperability benefits. A company simply not having to pay the fees for something proprietary can be all the funding needed to support a free software version.

Free software development generally only needs about 1/10th as much monetization to begin with. You may get to start with the works of other people; others using your work builds the user base quickly without paying to advertise and sell your product; you'll get a lot of free testing/debugging/code feedback; etc.

>> There is no such thing as “free.” R&D will be monetized somehow—the question is how.

There is such a thing as free. People who write Free software are doing R&D without any plans for monetization.

Codecs are software. Software has about zero manufacturing or distribution cost. Development does take effort. Very much effort for modern CODECs, but it's one thing to cover development costs and another to collect rent from - err monetize - a "standard".

I'm not sure that FLOSS developers have no plans to monetize. First, many of those devs offer paid support. Or their employability gets better. Etc. If people could not get any reward whatsoever out of FLOSS, it would be a much less active space.

Even with the Linux kernel, almost all development (over 90%) is done by people paid to work on Linux: https://thenewstack.io/contributes-linux-kernel/

This serves as evidence that 90% of all codec development could be done by people paid to work on the codec, and yet still available permissively and for free.

Sure. You can also make web browsers for free. But how do those companies make their money, and what are the public interest implications of that?

This technology can be valuable, but getting a patent made standards essential doesn't just compensate you for your invention-- it creates a unearned windfall. Practitioners regularly us anticompetitive and dishonest moves to get their techniques made into mandatory parts of formats, regularly these techniques offer little value over non-patented alternatives-- in some cases they actually hurt coding performance over completely naive techniques.

Xiph.Org's 2011's letter to the FTC directly discussions "industry standard" unethical conduct like effectively concealing what would later be claimed to be a patent encumbrance from a standard setting organization, exactly as Nokia did in this case.

https://www.xiph.org/press/2011/ftc/ftc-comments-20110614.pd...

How come it's an unearned windfall and not return on investment of sorts?

Developing the technology that has become the standard takes a lot of resources; you have spent them.

Standard bodies usually compare several technologies and pick one; you have risked that your development effort would not make it into the standard, and would be a dead end.

It looks a bit like bidding for a military contract. Did General Dynamics receive unearned windfall when USAF accepted their bid for a fighter jet, and kept paying for each F-16 they ordered? Surely the price of each included a part of the R&D costs.

Building a valuable bridge that people voluntarily pay you to use is an investment.

Planting a flag next to General Dynamics driveway so that you can collect a percentage off every aircraft that goes buy is rent seeking.

If patent holders were required to specifically identify and price the technology they encumber standards with such that the standards setters can pick and choose which were worth it-- that would be a different matter, but it isn't how it works. Note that I pointed out that the specific patented technologies often have negligible or even negative benefits.

At the very least there should be a global irrevocable license to implement the standard exactly. I imagine this could still allow the patent holder strong rights if any other party were to deviate or iterate on the standard. But it’s insane to me that standardization bodies don’t insist on this.

There's a whole body of law that's grown up around compulsory licensing for standard-essential patents.

https://en.wikipedia.org/wiki/Reasonable_and_non-discriminat...

The issue here is that Nokia allegedly has patents that are infringed by the standard, and didn't disclose them to the standards body.

Some of them do, at least now. The Alliance for Open Media, for one, has terms that revoke your license to relevant patents if you sue anyone for implementing the spec.

Ignoring the Lenovo vs Nokia case here because it really has nothing to do with video patents other than two company having disagreement.

>The public interest is that interoperability standards should not require patented technology at all to implement.

That is what EVC [1] Baseline Profile is. Basically bringing in all the tools from MPEG-2 and MPEG-4 / H.264 that have patnets expired or soon to be expired ( by the time finalised ). It is expected to be somewhat better than AVC High Profile but not as good as HEVC. ( Which I should remind everyone this claim seems too good to be true and requires 3rd party testing to verify )

It is interesting because a lot of the original purposed tools for H.264 were deemed far too complex for hardware at the time and were not accepted into the standard. Now 20 years later those tools are being put into good use.

[1] https://en.wikipedia.org/wiki/Essential_Video_Coding

IMO patents (the idea) are not the problem, they are a reasonable system for funding the work. The problem is that while patents (the implementation) are about money they don't apply to money, they apply to the freedom to utilize the patented work.

Patents should not limit freedom, they should funnel some of the money made from the patented work back to the people who did the work. If money isn't involved the patent shouldn't come up. IE. if you aren't making money from a patent you should be free to use it as you want. To put it another way, we shouldn't put arbitrary restrictions on freedom in exchange for money.

>standards should not require patented technology at all to implement

Do we really want every standard to rely only on ideas that are so old that the patents have all expired? FRAND licensing seems like a far better solution.

Someone should review what happened in Microsoft v Motorola here: if Nokia knowingly withheld information about patents they owned that will affect H.264 licensing and Nokia is on the H.264 WG, then what will happen is that the patent is not invalidated but not enforced (effectively enforced royalty-free licensing) if a manufacturer wants to make a conforming H.264 decoder (as H.264, H.265, VP9 and AV1 is defined only for decoding and the main reason that encoders may vary in their quality and performance). For encoding, any patent that is not necessary to make a conforming H.264 stream (in other words, to enhance H.264 quality and performance) is negotiated outside of FRAND conditions (but in common H.264 practice is also available to licensing through MPEG LA). If the patents are solely used to enhance H.264 encoding quality or performance, then Lenovo doesn't have any chance here (as these falls out of scope in the WG contract Nokia signed with ITU/ISO), but if the patents pertains to the decoding side (like Microsoft v Motorola), then Nokia will not be able to enforce the patents, at least for the H.264 standard.

Edit: Personal opinion: this is H.265ing H.264 and I hate it.

Edit edit: Lenovo is trying to pass some contested patents that was properly declared by Nokia before ITU/ISO as not properly declared: https://news.ycombinator.com/item?id=25398403

Final edit: I gave up reviewing Nokia's declarations after finding matches in the filing. Apparently, the so-called contested patents (at least those that I have checked) were called differently when Nokia submitted the declaration (Lenovo's legal reviewed the patent numbers but did not looked if they matched the filing number). I have looked at some contested patents and it seems that Nokia did declare them properly as far back as 2002, so I genuinely don't know what Lenovo is trying to claim here. Accordingly, I let others complete the review here.

Sadly patents are mostly used to protect a _discovery_ rather than _invention_. Who finds something first gets the cake and so on. For example when I was playing with digital synthesis, it felt natural for me to modulate the phase within the grain of sample and then I learned it has been patented by Casio. Then recently I read about so called TMT technology, where company essentially patented adding tolerance to components when simulating circuits... I mean d'oh? If you want to do realistic simulation, that's one of the things to take into account, but now that it is patented, good luck. This whole thing needs to be scrapped.

Just because it ‘felt natural’ doesn’t imply that it’s a discovery rather than invention. Generally in IP the word discovery is in regard to a natural phenomenon, something found in nature. Natural laws are discovered, everything else is invented. So what you describe sounds like an invention rather than a discovery.

I would disagree. For example phase is one of the parameters of the grain, so to modulate the phase how come it an invention, but to modulate a filter cut off is not? The same tolerance is inherent part of analogue component, as it is very difficult to achieve exact value of e.g. capacitance, so components are manufactured with different levels of tolerance, so if you simply acknowledge this fact of life in your simulation how is that an invention?

At least US has a carve-out for obviousness to people in the field. It has been chipped away at for a long time, so there might not be anything left of it.

The TMT I was talking about has gone through fairly recently. Please take a look https://patents.google.com/patent/US20170060527A1/en It's crazy.

What they are describing is literally a vanilla Monte Carlo simulation! It's the bread and butter of applied mathematicians in thousands of fields.

It reminds me of a patent for fire. Yes, fire. It was worded carefully to hide this, and it was approved. Something like: "A method by which a self-sustaining avalanche of thermal chemical reactions is initiated by an externally applied heat source."

In other words: We lit it on fire with a match.

Fun fact: there was an attempt by the MPEG working group to design a non-patent-encumbered internet video format.

It was killed late in the process by several companies making patent claims on the finished standard, without saying which patents were infringed, therefore making it impossible for the standard to be reworked.

I'm not doubting your story, I'm just curious how is that legally possible? I thought in the US you're innocent until proven guilty.

First, that's a principle of the penal system, not the civil courts which would decide such matters. Second, this didn't involve any legal action at all, and was instead just as a threat, basically saying that if they went forward with the standard, they would sue companies that invested into the standard, but by that point the companies would be in a bad situation as they'd have invested tremendous resources into the codec.

The moment you make any kind of non trivial program (say 20k lines and above) you almost guaranteed violate some patent so they were probably even right with their claims.

If your last statement is true, how come there are companies working on software every day and not getting sued constantly?

There’s threats of lawsuits all the time. Run an e-commerce site and you’ll get demand letters for cash from patent trolls.

They are actually getting sued, maybe not constantly, but if you are a valuable target, often enough. It has just become the cost of doing business.

They are. Quibi was under lawsuit by Elliot Management.

I'm pretty sure that patent contests are the fanfare of Europe (not EU or anything, Europe). Since ISO and the now-disbanded MPEG group is based in Europe... (If I remembered correctly the contesting group were mainly European and Asian companies. I forgot if Nokia opposed this.)

It's very difficult to preemptively establish that something is not encumbered by a patent, which is what the group releasing the standard would have to do to make the standard adoptable.

If someone successfully claims the standard infringes their patent after it is released, it failed to accomplish the goal.

It was just a vague threat that MPEG decided to heed, they could have decided to ignore it too.

The problem of course being that if anyone tried to implement the standard without having a license agreement with those companies, they'd get a threatening letter.

One question that's been bugging more with regard to these older encumbered standards as we get into the 2020s: does anyone know if there has been an exhaustive search of when they go patent free? H.264 Version 1 was 2003, and Version 3 (which added the high profiles, so it's probably by far the most common by this point) was 2005. So worst-case scenario, we're 3-5 years from any possible patent being expired. But in most cases standards patents have priority dates at least a year or two before, because negotiations happen well before standard official finalization (first draft in this case was sometime in 1999 or 2000 IIRC) and patent holders who try to keep it a secret and apply at the last moment risk getting cut out or having someone else come up with it and neutralize them.

I remember a lot of people were watching for when MP3 went patent free, and it'd be interesting if there was a site keeping track of when some of these other still commonly used standards did the same. Patents last relatively way too long in the tech world, but even here it's not forever.

https://www.osnews.com/story/24954/us-patent-expiration-for-...

> H.264 is a newer video codec. The standard first came out in 2003, but continues to evolve. An automatically generated patent expiration list is available at H.264 Patent List based on the MPEG-LA patent list. The last expiration is US 7826532 on 29 nov 2027 ( note that 7835443 is divisional, but the automated program missed that). US 7826532 was first filed in 05 sep 2003 and has an impressive 1546 day extension. It will be a while before H.264 is patent free.

The automatically generated list that this quote talks about also lists US 7835443, which expires in 2028...

Surely the publication of the standard 17 years ago constitutes public disclosure and/or prior work.

Using extensions to increase the working lifespan of a patent to 25 years (as done here, apparently) really shouldn’t be allowed.

It’s one thing if the work is under development during filing or something, but we’re talking about an openly published standardized technology.

Patent lifespan should be limited to the minimum of 17 years from filing (or maybe grant date), and 17 years from publication/commercialization.

Patent term should vary based on the field. In IT, let it be like 5 years. Maybe less. 20 feels like an eternity. And no extensions, ever. If you missed, well, congratulations, you failed, now someone else will implement it and make the world better anyway.

I believe H.261, MPEG-1, MPEG-2 (H.262), and possibly H.263 patents have expired now. MPEG-4 (DivX and such) is going to be next to expire before H.264.

I have found that at least one of the patents contested (U.S. Patent No. 6,879,268, formerly filed as 10/054,610) was indeed properly declared by Nokia back in 2002 [1].

1: https://www.itu.int/dms_pub/itu-t/oth/04/07/T040700053E0001P...

Edit: I have found other contested patents that were properly declared but I already said quits and let the courts decide on this.

So, Lenovo, who was not party to the standard definition, is looking for a free ride on properly disclosed IP by claiming that the undisclosed IP makes the properly disclosed IP unenforceable.

Suck it up and pay, Lenovo.

FFII is crowdfunding a lawsuit against the UPC, which will enforce software patents in Europe:

https://ffii.org/bundestag-vote-for-unitary-software-patents...

Nokia is trying the ole Qualcomm strategy, refuse to license patents to chip makers, then try to extract x% of a finished car or a finished laptop from Daimler/Lenovo because somewhere in the $100/$1k product there is a dollar value chip that does H264 decoding.

That won't come to pass, of course.

How's this different than ARM?

That something is cheap to manufacturer says a ton more about the huge advancements in semiconductors than it says about any value of what that chip does.

It's fine to dislike patents, but I don't think this reasoning is sound.

>How's this different than ARM?

What!? ARM doesn't do any standards essential stuff, they just developed their own ISA. And they have a massive array of options, from their own ready to go cores all the way up to perpetual architecture licenses that someone can do whatever they want with (so long as it's conformant) indefinitely. How is that remotely similar to something like codec or wireless standards? ARM is about as polar opposite from "refuse to license to for chips, try to extract based on patents for the whole device" as it gets. Their entire business is based around licensing to chip makers with nothing to do with the rest of the device. And if they ever tried such a thing, well first of all it wouldn't affect their perpetual licensees at all, and second everyone would just move to RISC-V or whatever.

Software patents are all wrong and should be eliminated anyway, IP value there and development incensitves are covered more than sufficiently by copyright. But it's much worse in what are literally monopoly collusion cases in the good sense, aka "standards". Normally even having a patent doesn't give an entity any special privileges against anti-trust there, they have a monopoly on their own patent but they can't seek to force everyone to use it illegally. Standards though are of great value, and while being horizontal monopolies can be very important in preventing vertical monopolies. Plus they're key when dealing with hard and soft natural monopolies, such as utilization of EM spectrum or content formats.

But the counter to that is supposed to be that while everyone in the pool gets a fair cut, they also give up their rights of discrimination or exclusivity. Anyone who doesn't do that for a SEP should have the book brought down on them hard.

There's a difference between selling flowers in a shop and randomly going around and pointing your gun at people holding flowers and asking them to pay up. The alternative to buying flowers is just spending your money on other things, which is different to the alternatives you have in an armed robbery.

Patent law gives patent owners a monopoly (i.e. the gun) over how an invention is used, but just like there's a difference between self-defense with a gun and using it for armed robbery, there should be - and increasingly is - regulation that forces patent owners to act as market participants offering something valuable rather than street thugs going for protection money.

It's fine to dislike patents, just as it's fine to dislike anti-competitive or outright dishonest practices around patents. Either alone or both in combination are reasonable standpoints.

So does it still work for qualcomm? If so, why wouldn't Nokia try the same?

Qualcomm has slightly better standing since they make the modems. “Your phone isn’t a phone without our IP” is a little easier to argue than “your phone isn’t a phone without our video playback”.

That type of damages (% of finished product) is not really a thing anymore, fwiw.

Patents are a nightmare for innovation and are used by bullies and criminals to crush improved operations. Patents are simply a money grab. No one should be able to own an idea... They can own their business and compete in the free market.

This is how the worst corporations that we all hate stay alive... Patents. Look at Oracle for example.

It sounds like some of the patents were granted after standardization was complete.

That’s incredibly sketchy. For all we know, they incorporate third party IP that Nokia derived from the standardization process.

I imagine Nokia would be hard-pressed to prove this isn’t the case.

I understand the need for patents. But the length of patents is absurd. 20 years for each of, what is basically a single piece of a huge puzzle, is absurd. Especially in tech, where most of the things are obsolete way sooner.

I don't know if I remember this correctly but it sounded to me like these patents are just regurgitation of open source efforts. A bunch of algorithms just stolen from the open source codebases.

And bunch of them just alter constants and/or apply filters or convolutions at different steps. One reads this stuff and will continually ask them selves, "this is patentable?'

It is if you are at the right place and the right time, which is shame and a sham because then the patent system rewards the corps with the deep pockets that s()it on standards bodies.

Is it really open-source though? As I looked at the patents, most were filed in the late '90s (don't ask me how the grant were delayed by ~5 years, I also don't know this). The main question is that the techniques discussed in the patents were prior art before 1999? Because chanting "open source" is not really helpful here unless there are code that exists then that matches the patents.

5 year delay before granting isn't terribly unusual based on my experience with the USPTO. It takes about a year for your patent to get to the examiner, then they might have comments or questions, which it takes some time to address, then it tskes some time to get back to the top of the examiner's queue. If they deny, and you appeal, it doesn't take too much to get to 5 years.

Of course, there were some games you could play to delay almost forever, read about submarine patents, and US patent rules changed in mid 2000s for patents filed after that date, the patent exclusive use period starts when the patent is filed, not when it's issued. Any patents still pending from before that will have an exclusive use period starting on issuance, if issued.

Reply on news.ycombinator.com