Weekly Edition
Return to the Announcements page
| |||||||||||||
Google's dropping H.264 from Chrome a step backward for openness (ars technica)
Ars technica has a lengthy criticism of the removal of H.264 support from the Chrome browser. "Google is now building a community around WebM (similar to that around Theora), but it hasn't taken any steps to submit WebM to ISO, ITU, or SMPTE for formal open standardization. The company is preferring to keep it under its own sole control.
For Google to claim that it is moving to 'open codecs' is quite absurd: H.264 is very much an open codec. WebM is not."
(Log in to post comments)
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 15:43 UTC (Thu) by johndrinkwater (guest, #65840) [Link]
Microsoft wrote a lengthy criticism of a competitor that they would like to see lose Chrome market share.
One or two vaguely correct arguments that pretty much amount to Why didnt Google also say MPEG-LA charges us lots of money.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 15:54 UTC (Thu) by butlerm (subscriber, #13312) [Link]
The idea that WebM isn't "open" in a way that H.264 is positively ridiculous. If the ISO / ITU / SMPTE want to be taken seriously, perhaps they should quit "standardizing" patented formats and protocols. A "patented standard" is an oxymoron. The entire purpose of patents is to grant individuals an organizations a government enforced monopoly. Every patent holder is a "patent troll", by definition.
The argument here can basically be summed up as you didn't go through our official international process to collect royalties from everyone and everybody, and that is bad.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 21:13 UTC (Fri) by rahvin (subscriber, #16953) [Link]
Patent Troll was a term coined by the industry to describe holding firms created to hold patents but no assets such that patent lawsuits could be launched without fear of financial repercussions to a working business. The key being that the company is entirely a law firm LLC with no assets and no business that the patent is protecting. Troll was used because much like the child's tale they sit under a bridge (the market) and extract a toll on industry that uses technology.
To say every firm with patents is by definition a patent Troll does not fit with the history or use of the word combination.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 15, 2011 6:42 UTC (Sat) by butlerm (subscriber, #13312) [Link]
I am aware of the conventional use of the term. I claim that it is a distinction without a difference, on the grounds that the entire purpose of patents - the reason why anyone wants one in the first place - is so that patent holders can engage in trollish behavior.
What good is a patent that can be easily evaded? Nearly worthless, right? No better than a mere copyright. No, what everyone wants is a patent on an technique so fundamental that no rational person thinks it is an invention to begin with, so that they can collect royalties on entire fields of endeavor. An application so fundamental that there are no alternatives except to quit the business entirely.
With perhaps the tiniest of exceptions the patent business is government malfeasance from first to last. The greatest impediment to science and the useful arts ever developed. Rotten to the core, enemy to human health and welfare, entirely contrary to its own stated objective. There doesn't appear to be anything that can be done to fix that, except allow patents for only the rarest of exceptions. Like pharmaceuticals with government mandated years long tests, perhaps.
Usefulness of patents Posted Jan 15, 2011 19:58 UTC (Sat) by jthill (guest, #56558) [Link] Well, I think the RSA patent was a good one, no matter that I was champing at the bit for it to expire along with just about everyone else on the planet who could spell "RSA". That's a fairly weak example -- I think they'd have published even without a patent system -- but nothing else would ensure they could get some of the reward. Or iow I think whether or not you regard someone demanding money for your use of "their" idea as trollish depends on whether you think they have a right to the idea and whether you think they're charging a fair price. Vorbis is technically a much better competitor for MP3 than Theora (but see the 1.2 preview) or VP8 for H.264, so Microsoft^WBright did raise one good point: if it were just about openness, Google would have dropped MP3. I suspect this is more about MPEG-LA's declared intent to abuse the monopoly position they're after.
Usefulness of patents Posted Jan 15, 2011 21:58 UTC (Sat) by butlerm (subscriber, #13312) [Link]
In an ideal world, if you come up with something that no one is liable to independently invent in the next *twenty* years there might be some small merit in the government granting a patent with mandatory licensing at regulated rates.
The procedural problem is that the government has no means of telling whether that is actually the case, so even if such a standard was adopted thousands of illegitimate patents would still issue. The government is completely incompetent at evaluating a much looser standard. The USPTO's conception of "prior art" is so narrow as to be useless. Its conception of "obviousness" is vacuous - as in there is no procedural means to determine that _anything_ is obvious at the time of invention unless it is spelled out letter for letter in prior art, because obviousness is inherently subjective evaluation. Neither the USPTO nor the courts have any reliable means to determine the answer either way.
Finally, I don't think there is an basis to claim that an idea of any usefulness is "proper" to anyone. Perhaps if it is so off the wall and unique that no one would independently duplicate in any amount of time (like a novel, or a musical composition). Anything less than that isn't readily attributable to an individual at all. Just an accident of timing and economics, if that.
Usefulness of patents Posted Jan 16, 2011 0:51 UTC (Sun) by jthill (guest, #56558) [Link] I don't think there is any basis to claim that an idea of any usefulness is "proper" to anyone. Which leads directly to regarding anyone wielding a patent as a troll: nobody has any right to, any proprietary interest in, any idea. Ok. You could make the same argument with land. Some cultures did, perhaps there are some left that do. Even here we know ownership is a gross tool. You can see the recognition of that with land use, where "owning" it doesn't allow you to stripmine prime farmland or to salt anything with toxic waste. But if you're going to support the notion that patents are inherently bad, I think you're going to have to maintain that companies would fund and publish basic research even if they couldn't get patents on any resulting gizmos. Look at what IBM's doing with AI. Do you think we'd ever see the results of their work in the absence of patents?
Usefulness of patents Posted Jan 16, 2011 3:42 UTC (Sun) by butlerm (subscriber, #13312) [Link] You could make the same argument with land.You certainly could, but the argument with land would be entirely different, because land is a classic example of natural property, where there is no better example of unnatural property than a proprietary interest in what is rarely little more than an idea. Now while I hardly doubt that we would all be much better off in a world where patents were the rarest of exceptions (if that), how's this for an alternative? Have the government run a contest for the 1000 best basic technological innovations of the year. Grant each one a ten million dollar award. That is ten billion dollars a year, to promote (not pay for) basic research in a world (or at least a country) without patents. About the same as the yearly budget of the National Science Foundation.
Usefulness of patents Posted Jan 16, 2011 21:19 UTC (Sun) by dirtyepic (subscriber, #30178) [Link]
Define "best".
Not only are you just shifting the subjectivity from some idea of obviousness to some idea of importance, you've now created an environment of savage competition and rampant corruption, where everyone is focused on short-term results. Why fund a ten-year project that might not make the cut in the end when you can fund ten one-year projects and increase your odds?
Usefulness of patents Posted Jan 18, 2011 7:32 UTC (Tue) by butlerm (subscriber, #13312) [Link]
"Best" is subjective of course - selection would be by committee. I suggest this as a compromise only. Real innovators don't need government rewards, monopolies, or subsidies.
Usefulness of patents Posted Jan 20, 2011 12:32 UTC (Thu) by Felix.Braun (subscriber, #3032) [Link]
Maybe, but Real Innovators also need to pay for food and housing. They should be able to live off their innovations, otherwise they'd have to work as banking clerks during the daytime. That wouldn't necesserily prevent them from making their inventions, but it would come close to being cruel and unusual punishment, which is unconstitutional ;-)
Usefulness of patents in some fields Posted Jan 20, 2011 16:38 UTC (Thu) by coriordan (guest, #7544) [Link]
In software, the manufacturers (software developers) generally are the innovators.
In some other industries, innovation mostly comes from R&D which is performed by people who are completely removed from the manufacturing process. In those industries, patents might be a good thing, because they allow for a link between the manufacturers and the innovators.
In software, the way to let innovators earn money is by allowing them to write software.
It's really essential to distinguish between different fields when proposing a patent policy.
Usefulness of patents Posted Jan 18, 2011 2:01 UTC (Tue) by rahvin (subscriber, #16953) [Link]
I understand what you arguing, but the problem with the patent system right now isn't that you are creating a government monopoly using patents the problem is the quality of inventions being patented. The abuse of the system is through software patents and business method patents which are on idea's. The system as originally intended and run for most of this countries lifetime was only on actual physical objects. For example the light-bulb where someone had to spend years, tons of money and research finding a material and system that would create light using electricity. Although it may seem trivial today to discover that in a vacuum tungsten would create a white light was quite revolutionary even if several people were working on the problem simultaneously. In addition even once granted the light-bulb patent would have only covered that specific implementation. Replacing the filiment with another material would have avoided the patent.
The problem is when you turn to software and business method patents, the patents themselves are so incredibly vague, they don't cover a specific implementation, they cover the very idea. That's the problem with the system, if software patents were restricted to a very specific implementation like we did in the past such that even a single change such as storing it one dimensional array versus a two dimensional array would evade the patent then the system would work. But the problem is that the patents as granted currently don't cover implementation, they cover the idea. Take multi-touch patents, they don't discuss the specifics such as input, variables, code and processing data, they discuss using more than one touch at a time. It's equivalent to patenting the existence of and use of a single button on a machine, something that in past would have never been granted although in todays patent office very well could.
The RSA patents were mentioned, I'd argue they were the prime example of a proper software patent. They were patenting the very specifics of the system because the encryption system is extremely specific in how the mathematical models work. A single change could have avoided the patents but in this case probably would have broke the system and resulted in non-functioning encryption. For a functioning system to work the patent should cover and require disclosure and filing of the code in question with the patent applying to that specific implementation. Admittedly this would gut almost the entire software patent system, but IMO that's exactly what needs to happen. Only the best of idea's should be patentable and worth patenting. Right now there are so many garbage patents filed yearly it's destroyed the credibility of the system. The great hope is that the Supreme court will tackle this with an upcoming case, otherwise we are looking at the continual decline and destruction of the US and other western economies that adopt this crazy policy.
Usefulness of patents Posted Jan 18, 2011 7:22 UTC (Tue) by dark (subscriber, #8483) [Link] If the RSA patents were good example, then what was their benefit? They were a massive obstacle in the deployment of secure internet protocols; I don't see any benefits to weigh up against that.
Usefulness of patents Posted Jan 18, 2011 18:59 UTC (Tue) by rahvin (subscriber, #16953) [Link]
I'll answer your question with a question. If RSA couldn't patent PKI would they have invested the money to create the very creative mathematical models that allow PKI to exist?
From the reports I read RSA invested millions creating the system, proving it worked and backing up the ideas behind it with hard proof. In fact that patent created a ton of research in cryptography, to the point where I'd argue that without the patent RSA probably would have never spent the money and if they had they probably would have refused to sell to anyone outside government. And beyond that had the patent not been granted all the secondary research to find other algorithms wouldn't have taken place and we would have a very limited cryptographic environment available now. How many crypto systems were developed to avoid the RSA patents?
And see that's the whole point of patents and the basis behind them stated in the constitution. It was recognized even back in late 17th century that by allowing a limited monopoly on an invention you would create an environment where people and companies would invest in research and development that benefits everyone in society later. I'll argue that the RSA patent directly caused the creation of an entire encryption ecosystem that today provides security in a digital world. We have more than a dozen different algorithms today that might not have existed without the RSA patent. That's a hell of a benefit for a decade and a half of single ownership.
As an aside I don't think you should be able to patent raw mathematics, but in the case of RSA and PKI IMO the math was only part of the equation, the application and use of the underlying mathematics was what was patentable not the mathematics themselves even if that's what appeared to be patented in the application it was the application and use which was the unique and patentable idea. Maybe I'm full of crap but I believe without the patent system you would have an economy much like China, where very few people invent anything and those that do are quickly copied and put out of business without the means to invent further. There's always someone out there that can copy ideas and either tweak the design or have better advertising and put the real inventor out of business. Maybe that's the way things should work, but why would the ones that come up with the original ideas bother even telling anyone about it if someone can come along and market it better?
Usefulness of patents Posted Jan 18, 2011 7:50 UTC (Tue) by butlerm (subscriber, #13312) [Link] the problem is the quality of inventions being patentedThere is nothing the Supreme Court can do about that. Nor Congress (by all appearances) because there does not appear to be any objective way to determine what innovations are "obvious", or what it is that might make one government granted monopoly social welfare enhancing and others social welfare destroying. What is going on now is the end game of the patent system. The very idea sowed the seeds of its own destruction. There is no legally defensible way to distinguish a "good" patent from a "bad" patent, so the consequence is that the vast majority of patents are bad, which is death to the economy.
Usefulness of patents Posted Jan 18, 2011 15:11 UTC (Tue) by nix (subscriber, #2304) [Link]
That sounds more like the end-game of the *economy* to me.
Usefulness of patents Posted Feb 15, 2011 14:10 UTC (Tue) by Randakar (guest, #27808) [Link]
No. Patents don't work for physical things either. If it doesn't work for the steam engine (http://discardedlies.com/entry/?52916_) which is about as physical as it gets, how on earth is it supposed to work for anything else?
There's a pretty good research paper out there (no link at hand right now, sorry) going into the question wether the early patents - steam engine, light bulb, telegraph, and so on - had a positive effect on innovation. The answer is a resounding: "No".
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 15:58 UTC (Thu) by tterribe (✭ supporter ✭, #66972) [Link]
It's true that ISO, ITU, or SMPTE are not, say, the IETF:
http://www.ietf.org/internet-drafts/draft-bankoski-vp8-bi...
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 16:40 UTC (Thu) by jch (guest, #51929) [Link]
> draft-bankoski-vp8-bitstream-00
That's an individual submission, not an IETF draft. IETF drafts have names that start with "draft-ietf-". It also says
> Intended status: Informational
which says that it is not being submitted for publication as a standards track RFC.
There is no simply no denying that WebV has never been submitted for standardisation. Whether that is a problem or not is a matter of opinion.
--jch
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 21:10 UTC (Thu) by sgros (subscriber, #36440) [Link]
Many RFCs published are only informational, and actually, every vendor has possibility of publishing something via RFC so that everyone can implement and use what's published.
And, also, many RFCs that are on standard track never reached that status, and thus formally are not "standards". You can take HTTP as an example. Last RFC was published in 1999 as a draft standard and never moved from that status.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 16:17 UTC (Thu) by me@jasonclinton.com (subscriber, #52701) [Link]
What is stunningly missing from the article is any discussion on how the adoption of H.264 would effectively kill Firefox. I love Chrome as much as the next guy but Firefox 4.0 Beta 7 and higher are really putting Firefox back in the performance game and it's still the defacto browser for web development.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 16:23 UTC (Thu) by magnusr (guest, #72358) [Link] If I understand it correctly, Google is not dropping H264 support from Chrome. They will still bundle the Flash player, which includes it. What they are doing is in effect just disallowing H264 outside a flash container.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 16:36 UTC (Thu) by Zizzle (guest, #67739) [Link]
This guy is the Ars Microsoft editor.
Which means he always cheer leads for everything in Microsoft's favor.
I'm not going to even bother reading the article I can already imagine what it says.
But it's amazing to see how many people are screaming about this decision.
From the quote it is an amusing jab to take. Didn't MS submit OOXML for standardization? Which they proceeded to fast track and get? And then completely not implement in office making the whole standard moot and useless?
Software standard bodies are obsolete. Open source code and free patent grants are the new standard.
don't knock all standards bodies Posted Jan 13, 2011 18:58 UTC (Thu) by stevenj (guest, #421) [Link] Software standard bodies are obsolete. Open source code and free patent grants are the new standard. You could argue that RAND standards are obsolete in favor of royalty-free standards, and that software standards excluding FLOSS software are obsolete, and that some of the traditional standards bodies such as ISO provide little value for software/data (or even do harm). But saying that standards bodies for software are obsolete in general goes too far. W3C and OASIS are two exemplars of standards bodies that do a lot of good in building consensus among multiple contributors (including FLOSS), providing quality control and rigorous review, and making sure that published standards are widely available, implementable (including by FLOSS), and royalty-free.
re: don't knock all standards bodies Posted Jan 13, 2011 21:25 UTC (Thu) by rlhamil (guest, #6472) [Link]
In my view, the first importance of a standard is that it is a useful specification _separate_ from an implementation, that (with or without royalties) could be used as the basis for multiple, interoperable implementations.
Something meeting that definition is a valuable engineering discipline that should often improve the quality (if not necessarily speed) of the resulting code. Such a standard, even if after-the-fact to an existing implementation, can serve as the basis for future development. Consider SVID (System V Interface Definition), which evolved into the more participatory standards POSIX, XPG, and Single Unix Specification.
Nevertheless, one could argue that for widest implementation and support, a standard should be
The whole open-source-as-an-ideology line of thinking leaves me totally cold. I like open source because it lets me do as much of my own troubleshooting (or occasionally modifications or extensions) as I want to. But I want code based on standards (or at least some co-development there) rather than code in _place_ of a standard. So I'm more interested in the ability to implement something royalty-free (yes, I agree that software patents generally stink) than in what the license on the source code is, as long as I can look at (but perhaps not even modify) the source code without spending an arm and a leg.
Still, the unfortunate reality, especially for codecs and data formats, is that people will probably want to be able to
That means that for some time at least, until not so much the ideology as the practical value sinks in (i.e. until those that create and consume content take control away from those that merely provide the tools to them), like it or not, there will be a demand for interoperability with non-royalty-free codecs and formats; and quite understandably so: most don't care so much about the broad currents of history as they do about just getting something done today.
That's one example of where I think a pragmatic rather than ideological approach could end up at the common goal of opening up communications and interoperability more quickly than going straight for the jugular by eliminating encumbered codecs and formats right away. Just draw a line, and say that after that line, new standards must be unencumbered, and leave it at that. That simple approach also takes away the foolishness of trying to jockey for market position by dominating, not the implementations, but by trying to sponsor or pick the dominant specification and make sure one's competitors pay for using it in ways other than just coming up with their own implementation. Compete on total results and value please, not on "tricks".
re: don't knock all standards bodies Posted Jan 13, 2011 23:19 UTC (Thu) by drag (subscriber, #31333) [Link]
To me standards are something that should be implemented AFTER a implementation is made.
Actually ideally you want to have a good standard you should have no less then 3 separate competing implementations of the draft well in advance of it getting ratified. That way your more likely to arrive a superior solution with competing answers to unknown questions.
If you make the standard first then you have no idea of whether it works right or not.
re: don't knock all standards bodies Posted Jan 14, 2011 9:44 UTC (Fri) by salimma (subscriber, #34460) [Link]
Indeed; so far we have two implementations of WebM (Google's and FFMPEG's); judging from the opinion of one of the lead developer of FFMPEG, it's probably a bit too early to get WebM standardized for now.
Not saying that Google should not start documenting the codec better -- but this showcases what tends to happen when a codec is developed behind closed doors (when it was VP8); too much of the documentation is implicit, only existing in the heads of the developers working from the same office.
re: don't knock all standards bodies Posted Jan 15, 2011 7:55 UTC (Sat) by rlhamil (guest, #6472) [Link]
> To me standards are something that should be implemented AFTER a implementation is made.
I don't have a problem with that; the SVID came after SVR2, IIRC.
The objective would be that the standard would be sufficient to guide the creation of an independent implementation capable of the specified interoperability.
The end result should be that the standard is the reference, and the initial implementation(s) are simply _sample_ implementation(s). I understand that for any given implementation, the source is obviously authoritative as to what it does. But the point of the separation between standard and source is that the standard is authoritative for what it _should_ do (but not how).
Most likely, I imagine as the ideal a prototype that becomes the first sample implementation, co-developed with the standard (but with some consideration to having the resulting standard able to work well on other platforms, etc), followed by one or more independent implementations based on the standard alone, and interoperability testing.
re: don't knock all standards bodies Posted Jan 14, 2011 10:46 UTC (Fri) by nhippi (subscriber, #34640) [Link]
> In my view, the first importance of a standard is that it is a useful specification _separate_ from an implementation, that (with or without royalties) could be used as the basis for multiple, interoperable implementations
However, "multiple implementations" is truly only necessary in the proprietary software world.
re: don't knock all standards bodies Posted Jan 14, 2011 11:17 UTC (Fri) by sorpigal (subscriber, #36106) [Link] There is value in multiple implementations, even open source ones. If there's only one consumer it's hard to know whether the spec is good because all you can say is that one consumer managed an interpretation which functions, not that the consumer was able to write an interoperable program from the spec. A spec which is so unclear that you can't expect two implementations to interoperate is a bad spec and you really need two or three implementations to find out whether this is the case.
re: don't knock all standards bodies Posted Jan 14, 2011 14:43 UTC (Fri) by farnz (guest, #17727) [Link] It's necessary anywhere; the point of multiple independent implementations is to catch things that are unclear in the spec, and clarify them before they become an issue. For a hypothetical example, imagine a spec that says "audio samples are stored as 24-bit values". The reference implementation may handle this as an unaligned 32-bit read from an array of bytes - in C-like pseudocode:
Now, you have a problem. Is this meant to be little endian or big endian? Is the picking up of 8 bits too many deliberate or accidental? Was the unaligned read deliberate, in which case I have to write code to handle it, or did the designer just not realise that unaligned reads are expensive on some architectures? These are all questions that get resolved as more implementations come along. Lots of the answers are going to be obvious, but some aren't - and the extra thought the questions trigger may result in a better format at the end of it all.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 20:17 UTC (Thu) by drag (subscriber, #31333) [Link]
> But it's amazing to see how many people are screaming about this decision.
They attract a lot of Apple users with their articles.
Apple says that Vp8 is slow and inefficient and H.264 is a standard, so therefore that is true and everybody who disagrees is a moron or a nerd and does not understand how the real world works.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 20:37 UTC (Thu) by Trelane (subscriber, #56877) [Link]
"nerd"s are actually kinda popular. I think you mean "neckbeard."
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 20:57 UTC (Thu) by elanthis (guest, #6227) [Link]
Nonsense! Nerds are ridiculed and picked on. Geeks are where it's at.
Nonsense! Geeks are awkward and unattractive. Dorks are hip, though.
Nonsense! Dorks are lame and only good for dodgeball targets. Nerds are the future.
Nonsense! Nerds are ridiculed and...
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 16:41 UTC (Thu) by RogerOdle (subscriber, #60791) [Link]
This is about the battle for the phone market. Google is not supporting Apples and Microsoft H.264 while at the same time, Apple is not supporting flash. Apple strongly controls the applications that can be run on its platforms and it does not support GPL applications. So who is and who is not supporting open standards?
The article goes into detail about why WebM is not "open" but falls short of fully explaining the status of Theora. The amount of content available on the web in the Theora format is growing. The article seems to want to crow about how H.264 is here to stay but so is Theora.
The W3C consortium's position to support only those standards that are not encumbered by royalties is the position that we should all get behind. Not just for the web but for all our entertainment electronics as well. I look forward to the day when bluray discs are encoded with truly free codecs. There are no technical reasons it can not be done now, only economic and political ones.
What will make this possible? Open hardware? Our entertainment electronics are full of reprogrammable logic circuits that allow us to upgrade them over the Internet. This has gotten to the point where bluray players are sold in a state where they are not fully operational until you connect them to the Internet to get the latest bug fixes. Is H.264 really so entrenched that it can not be replaced? How hard is it to insert Theora decoders into these devices? Hardware isn't as hard as it used to be.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 17:17 UTC (Fri) by jzbiciak (✭ supporter ✭, #5246) [Link]
There are many SoCs out there with fixed or very-nearly-fixed function hardware accelerator blocks that support a small number of standards. These can't be upgraded to support arbitrary codecs merely by flashing a new image onto the device.
Some SoCs have CPUs that are powerful enough to pick up the oddball codecs, but CPUs consume orders of magnitude more power to achieve the same end, and thus supporting arbitrary codecs represents a large battery life hit in a portable device.
Some newer accelerators try to be more flexible, but there are limits to how much flexibility you can build in before you lose your energy efficiency advantage.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 15, 2011 16:48 UTC (Sat) by RogerOdle (subscriber, #60791) [Link]
That's today. Reprogrammable logic extends into all areas more and more everyday and it is adapting to the needs of battery powered devices. This is pushed by the need for field upgradable designs and reducing time to market. We are getting hardware in the field in advance of the standards they are supposed to implement and we are getting comfortable with that because we are getting used to connecting them to the Internet for little fixes. This is especially true for the phone market.
Goggle is big enough to play the hardware game now. It can design chips for hardware decoding of non-H.264 video formats itself and offer them to the market. As more Internet content gets provided on alternative formats, the pressure for flexibility will increase.
The problem with H.264 is that it got pushed on us whether we liked it or not. It is covered by vague and over-reaching patents that try to lock down the very concept of video encoding. It seems to me that patents were originally intended to cover whole things that were completely implemented. Now they are used to cover pieces of things which represent incomplete ideas and I do not think that this was intended when the idea of patents was created. Patents were created to promote innovation by encouraging people to share ideas and combine them in new ways. How does this modern interpretation of the patent system accomplish this?
Remember that IP rights are not natural ones. They are not at the same level as the right to life, liberty, and the happiness which are natural rights. They are created by the state for a purpose and if that purpose is not fulfilled then they have no justification for existence. Is IP law being implemented and enforced in a manner that is in accord with its purpose? I do not think so. I see it as concentrating market share in the hands of the few who have the money to influence law makers. The law is being stood on its head where the IP system is used to stifle innovation instead of encouraging it.
An unusually bad article Posted Jan 13, 2011 16:44 UTC (Thu) by renox (subscriber, #23785) [Link]
Articles at Ars technica are usually quite good, so I was very surprised/dismayed to see such a poor article: H.264 is a codec made in a process which share the revenues depending on the number of software patents owned by the contributors: so supporting H.264 is in a way supporting software patents.
Which makes very difficult to see the removal of H.264 as "a step backward for openness"(*)..
I hope that this is only a one time mistake and not a signal of a reduction of the quality of articles at Ars..
*:Yes, WebM could be "more open" but that's a different issue..
An unusually bad article Posted Jan 13, 2011 16:51 UTC (Thu) by magnusr (guest, #72358) [Link]
There was also the anti-google take on the IE vulnerabilities that they published the other day:
http://arstechnica.com/microsoft/news/2011/01/internet-ex...
Ars used to be the best. I wonder what's going on there now.
An unusually bad article Posted Jan 13, 2011 18:46 UTC (Thu) by drag (subscriber, #31333) [Link]
> [Editor's Note: The original version of this story was published before receiving proper vetting, and many of you rightly chastised us for it. We apologize and present the following coverage, which more completely examines the issue.]
> Ars used to be the best. I wonder what's going on there now.
They are among the best on some things, but they still make mistakes. :)
An unusually bad article Posted Jan 13, 2011 18:49 UTC (Thu) by drag (subscriber, #31333) [Link]
Oh. And you have to pay attention to the authors. They certainly do not all agree with one another and certainly do not have the same viewpoints. But the point is to have interesting articles full of information that will generate ad revenue so that means people are free to put their own opinion up as long as it is well reasoned.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:11 UTC (Thu) by ms (subscriber, #41272) [Link]
Also, the article mentions CPU/GPU/dedicated-hardware support for H.264. But isn't that mostly DCT and such generic blocks that can be used by most codecs? Is there really lots of hardware being made that is absolutely specific to H.264?
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:16 UTC (Thu) by magnusr (guest, #72358) [Link]
Perhaps theory, but someone has to implement it. The DSP code tends to be the most closed thing there is in any proprietary system, so it's not likely that we'll see hardware-accelerated webm until (the firmware divisions of) hardware manufacturers get enough incentive to do it.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:30 UTC (Thu) by ms (subscriber, #41272) [Link]
So they don't just appear as generic extra ISA extensions then at a suitably fine-grained level - it's all too high a level and thus the individual functional blocks can't be repurposed?
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:35 UTC (Thu) by theICEBear (subscriber, #23193) [Link]
I think Google will leverage Android to get WebM into a lot of hardware. The issue would more likely be PC Hardware, but with Nvidia onboard with the standard I doubt that will take too long (and if in their game of one-up-manship continues AMD will soon follow).
I also saw a blog from the webmproject mentioning that there is already initial discrete hardware. Firmware/FPGA code can't be far behind.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 10:12 UTC (Fri) by bawjaws (guest, #56952) [Link]
The Tegra2 has accelerated Theora, Vorbis and VP8 decode, and VP8 *encode* and "videoconferencing":
http://www.nvidia.com/object/tegra-2.html
There was an avalanche of Tegra2-based tablets (and a smattering of phones) at CES. I believe the first one is due out this month (the Dell Streak 7"):
http://blogs.nvidia.com/2011/01/tegra-2-tablets-tear-up-ces/
They guys that make the chips for all the cheap, interesting/rubbish tablets are also shipping support for 1080p WebM this quarter, so devices in users hands maybe first half of this year?:
http://www.prnewswire.com/news-releases/rockchip-and-webm...
There was some valid arguments above that Google could have put WebM through a proper standardization process, or at least let the Xiph and FFMPEG guys a bit more time before locking everything down. The excuse at the time was that they needed to get the hardware guys working because of their longer lead times. Since they've delivered on that in spades, I'd say it was a good call in hindsight. The perfect is the enemy of the good etc.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:40 UTC (Thu) by hp (subscriber, #5220) [Link]
The way the hardware works is that you just feed in the H.264 and it magically appears on the screen. It isn't providing generic building blocks.
Some of the newer hardware is programmable and can support multiple codecs. Also you can just use the GPU to accelerate video, which isn't the same as full hardware decode but can be good enough if your CPU is somewhat decent (say a netbook, rather than a phone).
Lots of existing devices could never decode WebM though, is the bottom line.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:49 UTC (Thu) by jonabbey (subscriber, #2736) [Link] The way the hardware works is that you just feed in the H.264 and it magically appears on the screen. It isn't providing generic building blocks. Really? I would have guessed that you just feed the H.264 to the hardware's driver, and it uses the hardware for portions of the decoding and the rest is done on the system's main CPU.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 18:05 UTC (Thu) by hp (subscriber, #5220) [Link]
I don't know how every decoder on the market works, but the ones I've messed with aren't adaptable to a different codec. Even if they were they don't usually have open source drivers.
If you had to pull data back from the GPU (or decoder) to the CPU, then push it back, that would be a nasty performance hit most likely. You don't want to keep going back and forth. So any decoder/CPU split would probably want to be a single preprocessing step. (Some decoders may well have this, a bit hard to tell when the driver isn't open source.)
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 15, 2011 17:36 UTC (Sat) by kleptog (subscriber, #1183) [Link]
The only decoder I have any experience with was the one in the Series 1 Tivo. It was an all-in-one chip essentially. You configured a circular buffer and it would DMA de MPEG-1 data straight from memory and decode it into TV-compatible signal. It had some stuff to do overlays with text. It could send interrupts to the CPU which would then read blocks into the buffers as needed. That's why the system only needed an 8MHz CPU to run. It also couldn't do anything useful other than decoding/encoding MPEG data.
It doesn't seem logical to decode MPEG and then transport that data to another device. Decoded video is quite high bandwidth; often if you know the output format you can be more efficient about the decoding process.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 17:31 UTC (Fri) by jzbiciak (✭ supporter ✭, #5246) [Link] DCTs are cheap these days. Problem is, that people keep changing them. You now have some integer DCTs that are specified to be bit-exact, for example, as opposed to relying on looser standards such as IEEE-1180. (Even IEEE-1180 sucks. I know, having implemented an IDCT that tested fine against it, very well in fact, but ended up still failing to be useful in some low bit-rate codecs due to limitations in IEEE-1180's testing model.) H.264's CAVLC and CABAC are a wonderous nest of complexity that are truly horrifying to behold. There are so many little twists and tweaks in there that it could drive an engineer mad trying to understand it all. (I watched my neighbor at work descend into that pit of madness, actually.) All the gates spent on hardware acceleration for those guys are likely to be wasted if you aren't running H.264 or something close enough to H.264 that it runs afoul of its patents. There's all sorts of other fun, such as loop deblocking filters and so on that also take part in the pipeline. Now, there's been work (and if you google for it, you'll find papers for them) on reconfigurable accelerators that hope to support standards other than H.264 and retain much of the efficiency. That's great, and I'd love to see such things shipping in volume. They aren't yet, though. In the meantime, there's oodles of smartphones and other gizmos out there today that won't benefit from that new work, but will take a hit when asked to display a codec their accelerators can't handle. I'm not arguing in favor of supporting H.264 everywhere, unless that also means donating its patents to the public domain and making it free for everyone. I'm just saying recognize that arguments mentioning hardware acceleration unfortunately have some merit for the time being.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 17:15 UTC (Thu) by iabervon (subscriber, #722) [Link]
WebM has a reference implementation and a published document that agrees with it, which makes it a more real standard than, say, HTML has ever been.
This is why I never follow Ars links anymore Posted Jan 13, 2011 19:07 UTC (Thu) by zaitcev (guest, #761) [Link]
The low-quality journalism plus transparent political bias.
This is why I never follow Ars links anymore Posted Jan 13, 2011 19:42 UTC (Thu) by Trelane (subscriber, #56877) [Link]
The thing that really got me turned off was the sanctioned MSFT astroturfing.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 19:41 UTC (Thu) by jthill (guest, #56558) [Link] I gotta ask, newbie or not: did this really need to be posted here? Anyone who cares what ars reporters post (which I do) already found it, and I think the criticisms above are spot on. It's not worth reading.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 19:56 UTC (Fri) by marcH (subscriber, #57642) [Link]
I find it was worth it just for the discussion happening here.
VP8 specification Posted Jan 13, 2011 19:47 UTC (Thu) by DonDiego (subscriber, #24141) [Link]
The VP8 specification still has a very long way to go. It is neither complete nor does it agree with libvpx on all accounts. The FFmpeg implementors of VP8 had to use libvpx as reference in many places while doing ffvp8.
VP8 specification Posted Jan 14, 2011 9:31 UTC (Fri) by tzafrir (subscriber, #11501) [Link]
Good. Ffvp8 is a second independent implementation of the decoder. Is there a second independent implementation of the encoder?
VP8 specification Posted Jan 14, 2011 12:16 UTC (Fri) by bawjaws (guest, #56952) [Link]
I've heard that Google has sponsored Ronald S. Bultje to spend a year on xvp8, an encoder counterpart to ffmpeg's ffvp8 decoder (which he also worked on). Apparently some work has begun but he's just had a baby boy and so is taking a break to focus on being a dad. I don't know if the name is intended to indicate that it builds on the x264 encoder's codebase.
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 13, 2011 20:28 UTC (Thu) by b7j0c (subscriber, #27559) [Link]
the rancor over this discussion is hilarious, as if all the h264 advocates were assured world domination prior to this decision
the day mozilla decided h264 would never be part of the base ff build, h264-uber-alles was doomed, the google decision just put one more nail in the coffin as it was being lowered into the dirt. you can say what you want about mozilla, but with 350 million users, they most certainly can kill a codec if they want to
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 0:02 UTC (Fri) by AndreE (subscriber, #60148) [Link]
It's funny (or sad) that people willingly conflate "standard" with "open"
Google's dropping H.264 from Chrome a step backward for openness (ars technica) Posted Jan 14, 2011 11:27 UTC (Fri) by sorpigal (subscriber, #36106) [Link] Especially "De-facto standard." Remember GIF vs PNG, the early years? Or it's like being told "Jabber is not open, use the standard OSCAR instead." Time will tell.
|
|||||||||||||