Your browser is not supported. Please update it.

9 January 2014

How Apple, Microsoft and Google are hijacking the universal browser

What came first, the browser of the plug-in? Think about it. Some of the most important plug-ins are there to enable codecs and DRM systems and in most cases these were around (in one format or another) long before the browser itself.

Just before Christmas Faultline ran a discussion about how the W3C was sponsoring the future vision where HTML 5.0 was supposedly being brought to a browser near you ‘ whether that was based on Safari, Internet Explorer or Chrome ‘ along with MPEG DASH, in the form of a universal player.

The appeal of a move like this is to eliminate plug-ins, which in turn had already been eliminated in the Apple environment, and the driving force behind the W3C committees, powering, for instance the Encrypted Media Extensions, seemed a step in the right direction ‘ to prevent Apple from blocking apps on its devices. If you only send HTML 5.0 to any device, then even an Apple device would surely have to play it without reference to the rules of its App Store.

Faultline customers and critics alike have been quick to point out some cracks in our analysis, saying that if everything we were told by the W3C had been true, in particular the part about any DRM or CODEC being allowed to work with an HTML 5 browser, in the form of a plug-in, then it would be fine, but many argue that this is simply not the case.

When you add to this the idea that plug-ins are being virtual banned (phased out) by all three of the major browser suppliers and independent companies like Opera, have been forced to drop the webkit browser in favor of the Chromium Blink browser based on code from Google, you see that there is relatively little room for any plug-in vendor to survive.

It is against this backdrop that we hear this week that Google will push VP9, the next generation H.265 class codec from its On2 Technologies group, as a royalty free codec. We have analyzed the changes to the H.265 codec over and above H.264 over a year ago ‘ key elements include 10-bit color, and the replacement of the 16 x 16 pixel macro block with a coding tree block of any size.

Essentially then H.265 achieves many of its gains with more processing at both the server end and the client end, and many of the same ideas based on the same patents as there are in H.264. In order for VP9 to achieve the same type of gains (or close to) H.265, it will invariably use patents from the 30 or so patent holders that are represented in the MPEG LA H.264 patent pool. So how can Google give away technology based on patents that other companies hold?

There is a trick to this and a precedent. Microsoft wrote the VC1 codec as a rival to H.264. Here it used patents it did not acknowledge until it provided details of how the codec worked and asked MPEG LA to create a patent pool. Some 20 companies came forward to examine the product and make patent claims against it. And yet Microsoft had already given away VC1 on millions of operating environments and inside applications and was only punished by having to pay for a full license at the top end of the royalties at what would be today around $5.5 million a year.

Google hopes to avoid this because it went through the same experience with VP8. It tried to build a patent pool and could see that it was going to end up a net payer instead of royalty recipient on the deal, and abandoned the project. It will not put VP9 forward to be patented and instead if any company wants to claim that it uses one of its patents, it will have to prove it in a court of law without access to any of the technical details.

But the codec is only part of the issue, the other area is in the content protection and encryption, which brings us back to the Encrypted Media Extensions (EME).

Here’s what one discomfited rival had to say about the EME, a set of methods being pushed by Apple rivals in order to level the playing field, driven by Netflix, Microsoft and Google.

‘The EME strictly provides Javascript access to the functions of a DRM (and a content decryption engine) wrapped in a browser subsystem generically called a CDM (Content Decryption Module). Unfortunately no specification or guideline exists as to how to write a CDM and have it integrated within a browser.’

Now this may appear quite subtle but unless a major browser supplier wants to let you see enough code to build such a CDM, then you can’t. This has left content protection companies, such as Verimatrix, Irdeto, Cisco’s NDS and Nagravision, with the unenviable task of either working alongside the DRM systems of these major web organizations or finding a browser supplier that will let them in, such as Opera. The Chromium/Blink code (Blink is the rendering engine for Chrome) is supposed to be open source, and it claims that transparency is one of its core values. So you might be able to find a way to implement your own DRM or encryption through knowledge of the source code, but the feeling is that this is actively discouraged and anyway, you would have to include at least APIs to your DRM code in an open source environment.

Now a lot of people ‘ both users and developers on the internet will be delighted that specialist DRM players will be unable to take advantage of the new universal players, but how else are we ever going to see premium content on HTML 5.0 devices?

It’s not a great idea to offer the pirate clues as to how premium content protection works in the environment and anyway the other Content Decryption Modules don’t have to do this. Google offers one for Widevine, Microsoft one for PlayReady, and Apple, as we have said, does whatever it likes in and around Safari.

Without a DRM plug-in mechanism or even any discussion track talking about creating one, and all three of these browser makers declaring a no-plug in world, we are in danger of slipping into a tripartite content universe ‘ the opposite of a Universal player market. Apple devices would need a variant of HTML 5.0 that services for a Microsoft or Google would not fit into; while Internet Explorer and Chrome might cross promote ‘ in other words you may be able to work with a PlayReady service if you are viewing through Chrome, but not an Apple service, and vice versa, since Chrome currently supports both PlayReady and Widevine (although that may not continue).

Ultimately we would live in a worse world than before. Premium content services would have to have a Microsoft version and a Google version, each using different royalty free codecs and DRMs, all native to the browser, while Apple might keep its doors closed, and whereas today it has no problem with Apps using PlayReady or Widevine for protection, and Apple is happy to let IE and Chrome run today on iOS, it doesn’t have to support them natively and perhaps may choose not allow their CDMs to run on iOS.

Today PlayReady is almost the default DRM on most iOS apps, and that may well continue, but does that mean that instead of the entire world moving to a standardized browser and set of cloud services, we have to build an App whenever we want to see a piece of content? At Faultline we have often said that Apps are just web pages, but on slower mobile networks, and that as we go to the cloud, Apps disappear.

The only reason that Microsoft, Google and Netflix control the EME implementations and these particular W3C standards is because they provide 100s of people to the process, and that allows them to control the agenda. The net results could be three separate islands of non-interoperability ‘ one Apple, one Google and one Microsoft, and operators of video services would all have to change suppliers to one of these. In the process we would lose many rich sources of innovation and variety.

The EME is currently defined by Netflix in such a way as to preferentially support PlayReady related protocol architectures between browser and the service provider. One rival supplier noted this as a bug in the spec and supplied a fix, which was rejected by the W3C working group with no explanation forthcoming.

Additionally the EME is strictly architected to support the ISO base media profile MPEG-DASH streaming, so no other format can be used, even though today DASH is a fraction of streaming video delivered today, when HLS is the default streaming protocol. So it is architected more as Anti Apple than to help operators provide services, which is a poor basis for making rules.

If you have to use Chrome and Widevine and a VP9 adapted version of DASH just to view YouTube, and if that is the combination on every Android licensed device ‘ then Google will end up about as innovative as Apple has been since Steve Jobs died or a Microsoft has been since Bill Gates took a step backwards. Not very.

It is likely to remain this way until consumers begin to vote with their feet and begin to stop using consumer software tools, because they only take them to a fraction of the services they want. One way this may appear is for Apps to win out over HTML 5.0, defeating all that work that the W3C has put into it, purely because it won’t stand up to the three major internet bullies.

However if three major organizations carve up a market between them which they did not previously own, then there are anti-trust provisions, certainly in Europe, to ban the practices and the software that made that possible, and the European Commission, banning a standard from the W3C would be a fun item to write up and complex schism to untangle.