Hacker News new | past | comments | ask | show | jobs | submit login
Life inside MPEG (chiariglione.org)
77 points by clouddrover 7 months ago | hide | past | web | favorite | 54 comments



See also this thread https://news.ycombinator.com/item?id=18207478 for what that Genomics 'standard' they are trying to push is actually like.

MPEG is a rent-seeking organization that doesn't actually contribute anything useful to society. It just tries to lock up technology and standards behind patents, and then extorts money for use of those standards.


Arguably MPEG isn't the problem, it's just a symptom. You will not find many people as disgusted by rent-seeking as I am, but the real target in my scope would be US Patent law here.

MPEG is the rent-seeking arm of several organizations, some of which actually do excellent technical work that is arguably beneficial to society. They don't have a monopoly on doing excellent technical work, even the kind that is beneficial to society, and back when you could count on MPEG LA being the only contact you needed to license a set of patents, it wasn't the worst state of affairs in the world. Not ideal, but definitely not the worst that an army of lawyers could come up with.


> MPEG is a rent-seeking organization that doesn't actually contribute anything useful to society. It just tries to lock up technology and standards behind patents, and then extorts money for use of those standards.

You're thinking about MPEG-LA, and even then your description is debatable.

MPEG itself is a standardisation organisms (technically a joint subgroup of ISO and IEC), defining standards out of technologies both patented and not.

MPEG-LA is a separate and completely unaffiliated entity which licenses patent pools: MPEG-LA does not create or hold patents itself, they get agreements from patent holders and sell "bundled" licenses for all patents essential to implement a relevant technology (or so they claim e.g. MPEG-LA created a patent pool for DisplayPort independently from VESA, which disputes this necessity and asserts the standard is completely royalty-free).


> > MPEG is a rent-seeking organization that doesn't actually contribute anything useful to society. It just tries to lock up technology and standards behind patents, and then extorts money for use of those standards.

> You're thinking about MPEG-LA, and even then your description is debatable.

As we say in Germany: "Wer mit den Hunden schläft, braucht sich nicht zu wundern, wenn er am nächsten Tag mit Flöhen aufwacht." ("Who sleeps with dogs should not wonder if next days he wakes up with fleas").

In other words: if the MPEG does not clearly distance itself against the business conduct of the MPEG-LA, it should not wonder if the MPEG gets the same hate as the MPEG-LA.


> standardisation organisms (technically a joint subgroup of ISO and IEC

Guessing you wanted "standards organisation" and technically MPEG is a Working Group of a Subcommittee of the Joint Technical Committee between ISO and IEC.

Patents are sad because they create an incentive to argue for a technically inferior choice that you "own" over a superior choice that you don't own.

CA/B Forum (which isn't an SDO but bear with me) ran into trouble when setting up the Ten Blessed Methods (our name, not theirs) for validating control over an FQDN because it turns out some ways to implement these methods were patented by members. Although CA/B isn't an SDO, the effect of it making rules about how things are done is to set a standard, since all significant vendors in this space are members of CA/B, and all major trust stores require you to obey CA/B rules even if you're not a member.

The patents didn't cover any of the good ways to implement the Ten Blessed Methods. Literally in every case the patent described the most unsatisfactory half-arsed way to achieve something you could imagine, the existence of these patents was causing vendors to prefer to do something bad (which they'd patented) rather than something good (which they had not). But there was a reluctance to just say "OK, we patented all this stinking garbage, but forget about that" and get on with it, this stalled things for many months. After all patents aren't free, who wants to admit they bought something worthless? Ugh.


Is there any evidence of MPEG showing a bias in favor or patents? The different standards they define seem to have fairly well defined goals and then, to my casual observer's eye, they've had fairly concrete technical reasons for their decisions.

As it's constructed and the different patent pools involved, I'm not sure that a patent holder could benefit more than others by having MPEG use their patent.

That being said, patents are still annoying when you bump in to them. MPEG is remarkable in terms of the amount of interop though, you can watch video on your phone, laptop or your 4k TV all made by different vendors with different parts.


Okay, so MPEG-LA is clearly a rent-seeking organization, we agree on that.

However, MPEG enables the rent-seeking behavior by MPEG-LA, quite deliberately. I don't think it is unreasonable to say that MPEG are complicit and a big part of the problem. People are always making excuses for MPEG.

Secondly: in the case of MPEG-G (the genomics 'standard' that is being developed by MPEG), the very same people who are working on MPEG-G, as part of the MPEG working group, are the people who are filing the patents. See the hackernews thread on MPEG-G that I linked above.

That means the rent-seeking behavior is not only the domain of the 'separate and completely unaffiliated entity' MPEG-LA. MPEG is just as bad as MPEG-LA.


You keep saying rent seeking, but it doesn’t mean what you think it means. If you’re building a standard and charging people money to use it, that’s not rent seeking by definition because you’re creating value. Rent seeking would be if MPEG-G folks tried to use their patents to extract royalties for things people were already doing.


I know what rent-seeking means, and I understand your point, but I disagree with it. I think you are giving the MPEG-G folks too much credit here.

Unfortunately, the (software) patent system does not distinguish between people actually innovating and rent-seekers. Patents very explicitly give their owners broad rights to shut down any use of their patented inventions through injunctions.

If you think that the MPEG-G patents are not going to be enforced in case someone was already doing what they were doing before the patent was granted, well, have I got a bridge to sell you...

As for MPEG-G creating value, I think that is highly debatable. See this post: https://datageekdom.blogspot.com/2018/09/, which summarizes the MPEG-G situation as "Patents + extensive high profile marketing + lack of evidence for benefit".

That's not a valuable contribution to society. It's simply a tax on the bio-informatics community.


> Unfortunately, the (software) patent system does not distinguish between people actually innovating and rent-seekers. Patents very explicitly give their owners broad rights to shut down any use of their patented inventions through injunctions.

That's sometimes the case for software patents in general, but not for patents tied to a standard like MPEG-G, where the point is to get people to pay for being compatible. I've looked at a couple of the MPEG-G patents, and if you're infringing the MPEG-G patent it's probably because you're actually using MPEG-G without paying for it, not because you happened to independently recreate the same thing.


> MPEG itself is a standardisation organisms (technically a joint subgroup of ISO and IEC), defining standards out of technologies both patented and not.

One could argue that the MPEG could take a stronger stance by requiring a RF grant from its participants like some other SDOs do.


One could definitely argue that, but that's a far cry from "a rent-seeking organization that doesn't actually contribute anything useful to society".


> MPEG is a rent-seeking organization that doesn't actually contribute anything useful to society. It just tries to lock up technology and standards behind patents, and then extorts money for use of those standards.

It’s not “extortion” to charge money for people to use technology you developed, and its a better business model for society than bankrolling that technology through advertising (which is the FB/Google alternative to R&D).

MPEG-LA is not locking up technology that already existed or other people developed. Its members are the folks who originally developed this technology. It’s notable that no open-source entity developed something like h.264 first.


> It’s not “extortion” to charge money for people to use technology you developed

They aren't charging for use of their technology, they're charging for use of their algorithms/patents.

The extortion part comes from making it an industry standard, we should never allow proprietary protocols to become (official or de-facto) standards.


MPEG-LA members built the algorithms, at great expense. These aren’t Amazon’s “one click shopping” patents. And what’s the alternative to bankrolling protocol development with patents? It has worked just fine for decades, including for really fundamental things like memory interfaces. What’s the alternative monetization model? The web? Where the pretense of standardization has been dropped and the “standard” is de facto driven by compatibility with a browser that is developed for the sole purpose of spying on people and peddling advertising?


The alternatives to patents are trade secrets, open source style collaboration and academic research. All 3 have been pursued quite successfully to develop compression algorithms. 20 year monopolies are not necessary.


Open source collaboration is usually bankrolled by something else. E.g. the top two corporate contributors to the Linux kernel each contribute more than all the independent contibutors together. Sometimes the something else is benign (Redhat funds Linux development by selling support), sometimes not (Google funds Blink development through advertising).[1] Trade secrets wouldn’t work for a video codec. Which leaves academic research, which is bankrolled by taxes and tuition. Should tax dollars be used to subsidize development of consumer products primarily used for entertainment? I think the 20 year monopoly is relatively innocuous in comparison, or at least the alternatives are not obviously less problematic.

[1] 802.11fb, now with opposition research acquisition at the MAC layer!


> Should tax dollars be used to subsidize development of consumer products primarily used for entertainment?

If this means abolishing patents: then yes.

By the way: video codecs have a lot more applications than entertainment. I just want to mention teleconferencing.


Exactly. VP9 development is funded by YouTube, mostly. Works great.


I mean I think this is just where we disagree. Its better to just buy a device and have a couple of bucks go to h.264 development than rely on something like Youtube which makes money, among other things, peddling advertising to children and teenagers, convincing them to buy crap they don’t need. When I was young I thought Microsoft dominating everything was bad, but at least they had a business model where you paid them money for a product. Buy a Pocket PC and Microsoft gets $15 or whatever. I think this new trend of everything being funded by ads and data mining is destroying computing. I’d love to go back to the “bad old days” of companies having to pay Symbian instead of getting Android “for free” or having to buy Office instead of getting Google Docs “for free,” or buy a Garmin instead of getting Google Maps “for free.” This is not how I saw computing turning out, I don’t think this ends well, and I certainly don’t want more of the that replacing models that worked perfectly well before. What’s next, Facebook bankrolling DDR5?


The problem with h.264 was, that it dictated the business model for everything that embedded it. In this regard, it was exactly the same "cancer" as Microsoft once said about GPL.

Today, purchasing Windows doesn't guarantee you ad- and data mininig free system, and soon it will be difficult to purchase Office at all, you will be renting. In this regard, I'm happy that there are different business models, not just one that would slowly pick the worst traits of everything else.


> The problem with h.264 was, that it dictated the business model for everything that embedded it. In this regard, it was exactly the same "cancer" as Microsoft once said about GPL.

To add on this argument: another huge problem is that the license terms made open source development/distribution really hard (i.e. a patent minefield). Similarly it could lead to legal trouble for people who just hosted their videos.


Exactly, as a consumer I don't mind paying $1 more for VVC technology. And that together is nearly $2B of patents fee yearly assuming no cap. From Mobile Phones, PC, Tablet, and many others.

The biggest problem is those companies within the group don't agree on the percentage spilt between them. Basically they are just being bloody greedy. And the reason why we have MPEG-LA, HEVC Advance, Velos Group and others.

The good things is that most companies realise how much of a poor job MPEG-LA did ( My guess is that they only cares about their commission any way ), they started MC-IF [1], which includes most of the original HEVC MPEG-LA Group members, HEVC Advance Members and one Velos members as well. They are basically most of the Japanese players... and ...... Qualcomm.

I really wish VVC succeed, and push the technical limits of Video Compression as well as replacement of JPEG, without all the dramas and patents problems.

[1] http://www.mc-if.org/our-members


The problem with patents is that it's not good enough to simply ignore other patents and build your own thing - you can still end up infringing, despite never having heard of the company with the patent, let alone the way they implemented something that's obvious.


> MPEG-LA members built the algorithms, at great expense.

At least MP3 was created partially with taxpayer money as a cooperation between the the Fraunhofer Institute and the Friedrich Alexander University at Nuremberg. The Fraunhofer Institute is largely funded by commercial research, but ~30% of its funding come from the German federal state and the individual German states.


MPEG-LA is not associated with the MPEG. It’s a company that holds patents which are crucial for implementing MPEG videos.


> What’s the alternative monetization model?

There's the heart of the issue, the belief that everything can and should have a monetization model. The fact that we had to invent a whole new artificial class of property to handle the existing model should set off some alarm bells here.

A lot of organisations make a lot of money from these algorithms, is having them chip into a non-profit to develop free and open standards too much? How about having some of that research from universities and government standards bodies chip in too?


I don’t think it’s beneficial for society to have standards that are covered by patents.

I agree that inventors should be compensated, but this 20y patented-standard model looks too one-sided to me.


In patents, 20 years seems oppressive. But in copyright, it is something to dream of.


Compression has always been a patent minefield. Even for something as simple as gzip, it was a lot of legal work. For video compression, it is much worse. I am suprise that VLC does even exist despite such a minefield.


Usually, free software get a royalty free license. In the case of HEVC, software decoders are free as long as they are not bundled with hardware. They explicitly said that they are OK with VLC.

Worth noting that the patent conditions still apply. It may be a problem for some free software distribution, and as a result software like VLC isn't always available by default.


> Usually, free software get a royalty free license. In the case of HEVC, software decoders are free as long as they are not bundled with hardware.

The problem is that "bundling with hardware" is very vague. For example, if some OEM sells their PCs/laptops with preinstalled GNU/Linux (containing VLC), does this count as bundling. Or let us consider, say, a singleboard computer that can be both used as a PC and a set-top box just by installing a different firmware (think GNU/Linux distribution) that is provided for free (as beer) at the vendor's website - what is bundling then?


The PC is a bundle and the OEM (not VLC) needs to pay for the license. I think the second case is also a bundle: there are specific clauses to close such loopholes.

Worth noting that the license is less than a dollar per device and the manufacturer is only required to pay if he sells like 100k devices per year.


The PC is obviously a bundle.


In some countries, software patents are not enforceable.


And is codec development improving at a much more rapid pace there compared to the US?


> MPEG-LA is not locking up technology that already existed or other people developed. Its members are the folks who originally developed this technology. It’s notable that no open-source entity developed something like h.264 first.

Note that its claims are not necessarily founded on anything e.g. MPEG-LA created a patent pool for DisplayPort, VESA claims the tech is completely royalty-free.


MPEG-LA is unrelated to MPEG.


Yes? Nothing I said links MPEG-LA to MPEG, and my other comments specifically make the point that they're unrelated.


Software patents protection racket is a morally bankrupt "business model". It only exists because of the broken legal system that allows such parasites to slow down progress.


They're not slowing down progress AT ALL - their patent is remarkably easy to understand, their reference implementation is available, and if you wanted to reimplement it, they will send you a complete spec by email! They are using patents for their intended purpose - they're making money by licensing the method of doing something they figured out how to do better than anybody else.


If you don't know how software patents are slowing down progress, there are a lot of materials that explain it in detail. Some might be proponents of that approach (for the sake of profiteering), but claiming that it's helping progress is hypocrisy.


While software patents may be broken in many instances, I don’t think video codecs are one of them. It’s not like MPEG-LA is suing people for using independently-developed designs that just happen to infringe. If you’re infringing the MPEG patents it’s likely because you’re purposefully building on their work.

And that’s okay in my book. I much preferred the old days when everyone just paid Sony and Panasonic for a license than the current model where “free” codecs are bankrolled by spying on everyone’s email and helping Russians fix elections.


> It’s not like MPEG-LA is suing people for using independently-developed designs that just happen to infringe

I think you missed the whole FUD campaign against VPx, where they tried to persuade the public that VPx does infringe patents in their pool.


The problem comes in when patents have the effect of walling off an entire subfield of computer science from the perspective of the commercial R&D world.

Classic cases in point are IBM's patent on arithmetic encoding, which rendered that entire class of algorithms radioactive in the 1980s-1990s era, and the original Lempel-Ziv patent claims dating back to 1978.

We are most certainly not better off as a society because we granted decade+ monopolies on those claims, and I see no reason to think any differently of MPEG.


MPEG-LA are actively trying to impede development of open codecs, or slow them down with FUD and threats. They are acting like a classic racketeer thug. I'm surprised some are trying to whitewash their behavior. I guess glorified mafia is the way to "move things into the future" according to MPEG-LA supporters.


Commercial development of codecs has put a lot of resources towards codec development, the ability to be compensated for that work enables continued investment, and the ability of standards to incorporate patents means that this work that can be used in broad-based standards. All that society has gotten out of it is cheap, ubiquitous video on a lot of devices that wouldn't have it otherwise without consumers having to pay attention to what video formats they're paying for using.



All these people do more harm than good. Their formats should be abandoned in favor of royalty free formats, for example like AV1.


> All these people do more harm than good. Their formats should be abandoned in favor of royalty free formats, for example like AV1.

Know your enemy. :-)


The MPEG-2 patents expired earlier this year; are H.264 patents expiring very soon, or do they just want to bring in more royalties by putting out H.265 and already working on what comes after it?


Of course they want this. This is how they make money - by developing novel video specifications, usually highly competent ones (H.264 was so much better than WMV9 it was silly, and Thora came much later). H.265 isn't quite as good as VP10 in picture quality, so they're developing a new format to fight for marketshare.


Ends in 2027 / 2028. So quite a long time to go still.


MPEG hit the unavoidable dead end by pushing the broken patent system to extreme. Their patents aren't driving progress forward, they are slowing it down, and MPEG now reap what they sowed.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: