This is actually a huge badge of honor for VP8, and a repudiation of the meme that somehow it's just an H.264 knock-off. It's always stunned me that the same community that hates on patent trolls is in love with H264 (aka MPEG4++). MPEG-LA are IMSHO the very definition of trolls -- patenting (and charging royalties for) techniques that were obvious and known art in the video compression community for ages.

The risk for MPEG-LA in not agreeing is that the patent pool would collapse - if Google isn't in then why should anyone else remain in the pool?

I do agree that this agreement seems to be pretty good for Google though. Their ability to sub license VP8 has the potential to undermine royalties for H264 to the patent pool.

BTW, MPEG-LA isn't a patent troll by any conventional definition.

Firstly, MPEG-LA doesn't patent anything itself, they are merely an organisation to pool related patents amongst member organisations.

Secondly, the member organisations are mostly (all?) the opposite of patent trolls, too: they hold patents in the field because they are actively producing software in the field. That is the opposite of the typical non-practising-entity patent troll model.

Thirdly, if you do accept software patents then the patent pool model is actually quite a nice way to manage them. It means people who want to work in the field only have to deal with one entity and gain protection from patent claims from any other company in the field. (Note: I don't accept software patents at all - just pointing out that while they exist this isn't a bad way to do them).

I have no real opinion on VP8 vs H264, but I do know that the patent situation is much more complex than you seem to believe.

I think you have it kind of backwards. This whole thing is inside baseball.

Start with the patent system. It's really easy to get a narrow patent. It's far too easy to get an overly broad and totally obvious patent. It's really easy to get a narrow patent.

So what happens when all the industry players come together to create a standard for a video codec? They all offer up their "technologies" for inclusion in the standard, because if it gets included and they have a patent then they get royalties. Some of these are the ridiculously overbroad and obvious patents everyone is accustomed to complaining about. But here's where the narrow patents come in: Narrower patents tend to be stronger. It's harder to find invalidating prior art if the claims are more specific. Plus, more is better, and narrow patents are easier to get. So what they do is they all file for specific patents on exactly what they intend to put in the standard, so that all the major industry players each have several strong patents that cover the standard they're creating. The incentive that they each have individually to do this is obvious, and the others allow it because they want to do it too and they implicitly expect reciprocity.

It's basically a method for implicit collusion. They each include a bunch of patented stuff in the standard primarily for the purpose of increasing the amount of royalties they get from anyone who uses it. Since each of the major players does the same thing, the royalties between one another largely cancel out, but if there are any minor players or any new competitor attempts to enter the market, now they've got to pay a bunch of royalties to all the incumbents to use the standard whereas the incumbents don't have to pay any royalties to them because the newcomer wasn't around at the time to have put their patented technologies in the standard. It serves as a pretty effective method to thwart competition, and it's really hard for the antitrust authorities to prove anything illegal is going on.

So enter Google/On2 and VP8. They don't like all this. "Promote competition in markets complementary to your own" is business 101. So they specifically design a codec that doesn't infringe the H.264 patents and release it free to the world. Obviously this means one less barrier to competition in those markets -- it's exactly the same strategy as giving away Android for free. You can't allow a cartel to form in a complementary market or eventually they'll find a way to lock up all your customers behind a toll booth and squeeze all the margins out of your market.

This doesn't make MPEG-LA happy. They're in the toll booth business. If people start using VP8 or its successors then the whole antitrust-resistant implicit collusion scheme is going to fall apart. So they try to figure out how to stop it, but what can they do? When you sell hammers, you start seeing nails everywhere. So they threaten to create a patent pool for VP8. The point of the pool was never to actually license patents to people who wanted to use VP8. VP8 is somewhat worse than H.264 on account of having to avoid some of the ridiculously broad and obvious patents. If you have to license patents then you might as well just use H.264. Which was the real point of the pool for VP8. Make sure there doesn't exist a decent royalty-free codec so that people can't use it instead of H.264.

But VP8 was specifically designed to avoid all known patents, so setting up a pool for it turns out to be harder than it sounds. On the other hand, software patents are so broken that you can generally find a few bad ones that maybe possibly read on any given thing, at least to the point that you can tie it up in litigation for years. So that's what I imagine happened here: MPEG-LA probably knew they wouldn't be able to win in the end, but they had the ability to cause trouble, which can be leveraged into a tidy settlement.

Imagine if you are patent-holder-X, present scenario: We want to transpose the MacGuffin, but that is covered by the patents held by ZZZ Corp. But that's ok, because ZZZ Corp is in the MPEG-LA patent pool, and the patent is covered by our license so we'll just do it. Happy days at ZZZ Corp

Scenario if Google didn't pay the license: We want to transpose the MacGuffin, but that is covered by the patents held by ZZZ Corp. So lets see what VP8 does instead... hmm.. we can do that. And now we don't need out MPEG-LA license, so lets drop paying that.

Meanwhile at ZZZ Corp: no we aren't getting any revenue from MPEG-LA for out patent on transposing MacGuffin. But look: Apple and Microsoft and Adobe all transpose the MacGuffin exactly how the patent explains to do it. If we drop out of the patent pool now, and then file a new derivative patent on fast MacGuffin transposing, then everyone will have to pay us fees themselves, or not implement the new, faster MacGuffin transpose.

But there is no pool for VP8. I'm not aware of anyone being able to articulate a specific patent that VP8 is alleged to infringe. The point of a patent pool for VP8 could never be to actually license anyone because the only reason to use it over H.264 is if it's royalty-free. The point of threatening to create the pool is to cause FUD as to its royalty-free status. The "license" Google just bought is them paying MPEG-LA to shove off and stop spreading FUD about VP8.

The same community that hates patent trolls continued to use gif and MP3 for many years, so I'm not sure why use of H264 would surprise you. And are people really 'in love' with it? All I see is people who prefer the clearly technically superior option. Not to mention the fact that many of us live in places where software patents are not relevant, and don't feel inclined to use inferior technology just because of the idiocies of the US legal system.

But since those "idiocies" have had their intended effect of squashing competition you don't know what you are missing. Its like everyone complaining about how bad other office suites are when MSFT controls the evolution of the .doc format and Windows.

You'll only notice if we end up locked into using and paying for H.264 long after it stops being technically current.

Back when the US authorities were trying to restrict exporting strong crypto, was the correct response to just use worse crypto, or was it to move crypto development outside the US and tell the US gov to go fuck themselves? I feel it was the latter, but I may be wrong.

The reason why people tend to be fans of H.264 is that it has a wide range of hardware that can encode/decode and that hardware has been available for years. This saves greatly on power for embedded devices/phones/etc.

While implementations of VP8 in hardware do exist, I'm not aware of it shipping in any sort of volume. Are there any modern Android phones that can do VP8 in hardware?

Basically none of the H.264 hardware is usable for this purpose. And x264 isnt tuned for this useccase, which makes VP8 the best codec for this purpose in basically every dimension.

But don't expect the political oppostion to be silenced by that, as patents and technical excellence and hardware support and power efficiency were just convenient excuses that they can ignore if the facts change.

edit: just noticed this covers VP9 too. Its probably too early to push for that as mandatory in WebRTC but there's nothing stopping Google using it for Chrome to Android chats, or anyone else picking it up.

Apart from the IPR issues please explain why you have come to this conclusion. x264 has excellent low latency encoding support.

I take it you believe x264 is better than libvpx for these use cases to some noticeable degree? Because even if x264 was tuned for this I thought there were some other considerations in the H.264 format that would make VP8 more competitive in this area than in other, no-holds barred maximum compression of high-bitrate files situations. (And I mean technically, not in the respective chances that people will actually use that specific H.264 encoder)

http://www.nvidia.com/object/tegra-3-processor.html

But your point still stands. Hardware support is still rare, and it's extremely beneficial, nearly required.

amd, arm, broadcom, qualcomm, texas instruments, nvidia

intel is considering hardware based acceleration for their atom based tv chips.

Given that the vp8 hardware acceleration is free of licence costs, and given this patent agreement, I hope that we'll see much increased hardware support for it.

There is no hypocrisy here; I am allowed to love technology while simultaneously hating the practices if its authors. I am allowed to benefit from the good produce of a bad regime. In no way does that compromise my argument for reform.

Those who try to argue that reformers ought to boycott everything they disagree with are making a kind of appeal to hypocrisy; a logical fallacy.

Huh? Whether or not they had VP8 dead to rights, they would agree to a licensing deal. The MPEG-LA's purpose is to make licensing deals for the patents in their pools. They litigate when they cannot get an agreement for a licensing deal.

Litigation is rarely a smart move for a licensing organisation. During the years wasted going back/forth during appeals etc the competitor has a price margin advantage which they can use to increase market share. It's not that easy to quantity that if damages are awarded.

And seriously you either need to put up or STFU. If you have some prior art that invalidates specific patents then speak up. Plenty of companies would be very, very interested. But I'm going to take a guess that you don't.

To quote Seth Meyers -- Really?

So you honestly believe that all the MPEG-LA patents are true innovations? Or just perhaps, having 100+ patents, an army of lawyers, and an FTC antitrust pass is intimidating enough to trump whether the patents are for the most part frankly crap.

Not sure who I'm misquoting here, but in reality you can patent a ham sandwich.

IMHO, MPEG-LA probably just scored some free cash. However, by avoiding a legal battle, and getting a 'clean bill of health' for VP8 from their main detractor, Google's VP8 is now ready for much wider adoption. After all, the principle patent-free niggle issue has now been dealt with cleanly.

I honestly don't know why Google bothered to buy the company in the first place. Seems an odd place to assert a competing position.

I mentioned Google Glass for instance, to which you retorted that Google Glass was 'vapourware'. Funny enough you then spent tons of posts trying to raise concerns regarding Google Glass privacy implications, you sure spend alot of time addressing vapourware.

It's almost as if you had an anti-Google agenda...

Well color me surprised!

Instead I would much prefer to see standards made available under low royalty, FRAND terms. This has worked to great effect for WiFi, USB, Bluetooth etc.

But of course that is looking harder to achieve with FRAND abusers like Google around.

FRAND works fine for hardware because by the time you spend $100 to manufacture and ship the thing you can spend $5 licensing patents.

It fails miserably for software because it's incompatible with free software, which means that the free software people have to create a competing standard that they can implement and we get useless fragmentation that hurts everybody. Produce something compatible with free software to begin with and none of that needs to happen.

I mean WTF, just require the patent holders to license the patents royalty free for free software implementations in order to have their patents included in the standard, and then they can still collect royalties from manufacturers who do hardware implementations. It's not like they were going to get any royalties from free software developers in the first place.

Old habits die hard; old industries die harder.

Abuse of FRAND patents is a different matter, but I will point out that it was in retaliation for asserting some pretty ridiculous patents in the first place. The nice thing about VP8 is that, even if Google turns evil (or, if you're of the opinion they already are, even more evil), the patent promise they granted is irrevocable.

Microsoft ended up giving Barnes & Noble (a bookstore for goodness sake) half-a-billion dollars after going after them for Android/Linux patents, but if you read the press release then they focus on an ongoing royalty payment going in the opposite direction.

Google may have entered negotiation with these 11 patent holding firms and threatened to destroy them with either a giant patent lawsuit or just by bullying them with their massive bank account, team of corporate lawyers, and market control. I would suggest the royalty free licence to these patents for use in VP8 or VP9 would take either that or a very big bag of cash from Google (or some combination of the two).

It's not really to either sides benefit to reveal what happened after they've got the best deal they thought they could though.

Or have they? The press release says that Google can sublicense the patents to other VP8 implementors...

edit: never mind, the rtcweb post (below) nails it down a bit.

(I also see they're playing a bit of the martyr by saying "This agreement is not an acknowledgment that the licensed techniques read on VP8." As if the truth has anything to do with patents.)

http://www.w3.org/2004/02/05-patentsummary.html