Theora More of a Patent Threat than H264? Wait, What? | Hacker News Reader

Theora More of a Patent Threat than H264? Wait, What?

osnews.com

33 pointsnnutter26 comments

This article is an attempt to win support for Theora over H.264 not by refuting the basic facts of the case, but by making us feel angry about those facts.

Gruber in fact does not appear to be wrong: there are indeed two classes of patents that Theora is exposed to (general but poorly known patents covering modern codecs, and the patents MPEG LA owns). There clearly are codec patents that MPEG LA doesn't own. There clearly are codec patents that they do own. MPEG LA clearly isn't going to sue, in the immediacy, over their own patents.

Meanwhile, it is indeed entirely in MPEG LA's own hands whether or not they will sue over use of Theora. They sure are a bunch of evil jerks. But that changes the facts of the case not one whit. Being upset about MPEG LA's IP position doesn't change what that position is.

The article refutes the assertion that Theora is infringing on patents held in the MPEG LA patent pool. The MPEG LA has been shaking a stick over Theora's head, but has yet to point out which specific patent(s) Theora is indeed infringing.

In fact, noone has come forward with a specific patent that has been shown to infringe after scrutiny and noone has yet sued Firefox for shipping this so-called infringing codec.

The tone of the article is basically a "put up or shut up" directed at the MPEG LA members. When someone is trying to spread FUD about your product, there's not much else you can do except ask that they stop talking and start showing evidence. Since they've been unwilling to do so for some time, it's natural to conclude that this is nothing more than sabre-rattling to try and drum up some extra MPEG LA licensees.

Do people familiar with the state of the art in codecs seriously think that MPEG LA doesn't have patents they could reasonably claim are infringed by Theora? That isn't the impression I had.

Noone, either familiar with the state of the art in codec or not, has come up with a reasonable set of patents that Theora's ancestor codec from 2001, VP3, could have possibly been infringing. Individuals are free to believe whatever they like, but there hasn't been any evidence of infringement beyond the MPEG LA's CEO making vague assertions.

The MPEG LA has a financial interest in keeping the patent situation around Theora murky, which they've clearly done well.

Just so it's clear (and some of the quotes by the MPEG LA's CEO make more sense from this page http://www.streamingmedia.com/article.asp?id=11746):

* Theora is a patented codec (ie: there are active patents today on the concepts used within VP3)

* One of the MPEG LA's patent holders, On2/Google does hold those patents

* On2 has, however, licensed those patents globally, indefinitely and royalty-free, making the above two points true but entirely a non-issue

Google/On2 are not members of the MPEG-LA patent pool. Google has an H.264 licence which is separate.

The MPEG LA has a financial interest in keeping the patent situation around Theora murky

Web streaming is free until 2015. Theora will not be found in Set-top-boxes and mobile phones which is where the real money is made in licensing.

How did you get this impression? One of the ffmpeg guys had been hinting about this for a while and finally posted in a LWN thread to point out 3 Nokia patents he thought Theora infringed.

http://lwn.net/Articles/372416/

Which were shot down pretty quickly: http://www.0xdeadbeef.com/weblog/2010/02/some-additional-inf...

You could argue that the ffmpeg guys should be the ones to know best about MPEG patents that Theora infringes since they implement those codecs and as this example shows they are happy to jump to conclusions even when two of the patents clearly couldn't apply for the very basic reason of being filed after VP3 was made public. So the fact that they've not named names could be seen as a positive sign, but as with all things patents, proves nothing.

Indeed. This is the same as Microsoft saying that Linux infringes on its patents. It's probably true, but nobody cares.

One of the main points is that code behind Theora has been made public for years, and in fact Xiph has been actively asking people to tell them if there is a patent violation. The significance of this comes into play in a potential lawsuit, because patent enforcement is simultaneously a burden on the patent-holder: if they are aware of a violation, it is their legal duty to enforce the patent, and if they fail in that duty then eventually they lose the right to claim damages from infringement (big oversimplification here).

The point being, even if MPEG-LA does have infringing patents, the patent holders can reasonably be expected to have known about it for years, and so if they decided to sue they would have to explain why they didn't enforce their patent in the years that Xiph has been asking them to enforce the patent.

Conclusion: all this talk about Theora potentially being in patent-jeopardy is FUD and BS.

It's trademark owners that have an affirmative duty to enforce their marks. The situation for patentholders is far murkier. It is far from true that a patentholder that knew about Theora for years would be simply unable to enforce their patent against Theora users.

It is far from true that a patentholder that knew about Theora for years would be simply unable to enforce their patent against Theora users.

It's not necessarily false either. The doctrine of laches (http://en.wikipedia.org/wiki/Laches_%28equity%29) says that you can't deliberately delay legal action in order to run up the alleged damages or prevent them from taking alternate paths. By publicly making potential patent holders aware of what they're doing, Xiph should be in a good position to claim that defense.

(In case it's not blatantly obvious, IANAL).

A laches defense in a patent suit generally only affects damages, limiting the plaintiff to damages based on infringement AFTER the suit was filed. If all the plaintiff is interested in is telling the defendant to stop infringing, laches isn't much of a problem.

It's far better that they exist than that they don't. If they didn't, there is a chance that any of the patents in their pool could have been acquired by a patent troll and exploited to even more devastating effect.

Gruber's right about there being two distinct sets of patents Theora has to worry about (basically big companies that are part of MPEG and random other companies/trolls) but he's wrong about whether they are "likely" to infringe either set and wrong about basically everything else he says about patents in general e.g. misusing the term "submarine patents" even though he links to an accurate explanation on Wikipedia, and in their specific relation to Theora and H.264.

For example, why should Theora worry about H.264 patents? Theora is from the previous generation of codecs, which means it is both simpler and older, two very good things when it comes to patents. That's two very good reasons why Theora is less likely to be exposed to random non-MPEG patents than H.264.

If Theora infringes any MPEG patents then it's more likely patents from MPEG-1, 2 and 4 part 2 that they need to worry about. But these are well known and Theora has worked around them.

Thom (of OS News)'s arguments are, as you say, mostly rabble rousing, and probably wrong wherever he gets into specifics, but that doesn't magically make Gruber right.

But this does remind me of a million other tech battles where you have the choice of supporting the plucky underdog or rewarding the source of the actual problem (e.g. people building sites that work on IE6 only) and I think people getting angry, at least as a first step, is a potentially useful strategy.

There seems to be some confusion around the term "submarine patent".

One use of the term, which came around during the 1995 GATT/WTO debate and which is described in the Wikipedia article, refers to patent holders manipulating the PTO in order to keep them from issuing a patent until later when they are ready to sue. This practice was struck down by the courts and made impracticable by the 1995 law.

The other use of the term is to refer to unknown patent infringement. This comes about because software patents are such a joke that nobody knows who exactly owns the patent to what. Even the inventor might not know that something infringes on their patent until lawyers start rummaging through their patent war chest looking for something to sue with.

This second use of the term seems to be what Apple and MPEG LA are referring to.

"No one in the market should be under the misimpression that other codecs such as Theora are patent-free," Horn claims, "Virtually all codecs are based on patented technology, and many of the essential patents may be the same as those that are essential to AVC/H.264.

Translation: There's enough vague patents floating around out there that read "Its like TV but on the inter-computer-web-thingee" that no coding technology can ever be considered demonstrably free from infringement.

So just pick a codec and know that if you make money, the trolls are going to be at your door, no matter what the pretext. If its not your video codec, it'll be your "one click shopping cart" or something else equally inane.

Big games are shipping with both theora and vorbis. Nobody sued them. Also Youtube used VP6 and VP7 for a while which were just slight modifications of VP3, just like Theora. And nobody sued Youtube. So why would they wait for theora to become used if they could have just sued google? Doesn't make sense.

Because they don't want to sue google first. (There are much easier things to sue google/youtube over right now.) They want to sue a smallish company who will "license" rather than fight because they can't afford to. After a few of those build up some precedence, then they go after the really big fish.

For the submarine to work on this type of thing, you have to wait until a great many players of all sizes adopt a technology and commit to it.

I'm not saying we shouldn't use theora. We should keep our options open and use everything. I'm just saying that whatever becomes the dominant standard, be prepared for "I'm a pc and windows 7 was my idea", only with lawyers and stuff.

Youtube have never used any On2 VPx codec to my knowledge, but Adobe Flash and basically every video site on the web used them up until about 2 years ago when they started to transition to H.264 (though many still use it for certain bitrates) so your basic point remains that lots of companies, big and small, used or are using it right now without any patent problems.

(I assume Flash on mobile phones means they'll now all be exposed to the exact same patent risk too)

Whilst there are many big games which use Vorbis (Halo comes to mind) as far as I know there are none that use Theora. Most use Bink because of its simplicity.

http://wiki.xiph.org/Games_that_use_Theora

That's not a complete list since it is missing e.g. Starcraft 2

None of those I would consider "big games"

http://en.wikipedia.org/wiki/No_true_Scotsman ?

It would be if "big games" was as clear a category as "Scotsman".

You can't write an open letter and expect patent owners to tell you if you're infringing or not. The accepted thing is to conduct patent searches (money and lawyers), and make sure that your product does not run afoul of something that's already in there.

I'm not saying that this system is ideal, (It is not! Let's make something better to replace the patent system - something based on private contract alone) but Theora is not necessarily legally OK (under the existing patent regime) just because some guy wrote a few open letters and posted them to a website.

What would I do? I would go with whatever is the better technology. (And from what I've read, Theora is a terrible format.)

Making sure that some piece of software doesn't run afoul of an existing patent is basically impossible no matter how many billable lawyer hours you throw at the problem. You'll either come to the conclusion that everything has already been patented and thus you can never build anything ever again, if you read patents broadly, or you won't find anything at all, if you read them more narrowly and throw out the ones that obviously have prior art and should never have been granted.

The only 100% reliable method of patent lawsuit control is software development abstinence . . .

Xiph and Mozilla have both done patent searches. The lawyers that they spent the money on told them that publicizing the results would reduce their options if they were ever sued so the only evidence they haven't found problems is that they haven't altered Theora to work around problems and that they ship it in their product, respectively.

I would assume Google (and Opera and a bunch of other companies with slightly less impressive amounts of cash and lawyers) have done the same, though I don't think they've mentioned it if they have.

Reply on news.ycombinator.com