January 21, 2010

Miscalculation and video standards

It's been one week since I've been trying to write an entry about how data collected about users on social networks could lead to nefarious results, but after reading it I found it was not funny at all to read. It was even pretty depressing. So I'll do a summary and switch to another thing after. Here goes : If people are not careful, they may misjudge others without even meeting them in the first place because of unrepresentative data. Instead of ending this small summary in a vision of a post-apocalyptic world (for that you've got your non-Star-Trek science fiction movies), I'll just say to be careful. Not only with what you post online, but how also you collect your information. You may have a lot, but don't let bits judge for yourself.

Now that this is done, that was just in in my news reader : Introducing YouTube HTML5 Supported Videos. As you may know, the supported video codec is H.264. In all honesty, I tend to think they should also support Theora. And I won't go into the juicy technical and juridical details. A very nice article has been made on Ars Technica about the debate not so long ago - go read it, it's very interesting.

My opinion on all this? It's a big mess. I mean, look at some statistics here. Pay attention to the quantity of technologies. In alphabetical order : Flash, Java, Quicktime, Real, Silverlight, Windows Media Player... Lots of choices here and while having choice is good, having functional ones are always better. And when it comes to the Web, they must preferably be open. When developing Web pages, it is the developer's job to ensure that he chooses a technology that will satisfy most of the users and of course make sure that the whole thing stays accessible. None of the above options does the job when it comes to video streaming. None of these work seamlessly on all major browsers and platforms. They all require to install or configure a new component into the browser in order to work.

You look me in the eyes and say it's simple. Without laughing. You are not? Well I am!

Multimedia content is in strong need of standardization and browser built-in functionality. It's not like it hasn't been done before in other contexts. Image handling (on the non-vectorial side anyway) has been pretty much standardized and implemented. JPEG for photos, PNGs (or GIFs) for transparency and animations. None of them require you to install a plug-in to view them and they work on pretty much everything that can handle graphical elements. Even text-based browser can have text equivalents (if it's done right).

When it comes to video, nobody wins. Go with Flash? Install supplementary software. Java? Install supplementary software depending on the platform. Quicktime? Install the codecs if you're not on Mac OS. Real? Install the codecs. Silverlight? Install supplementary software and the supplementary codecs if you manage to make Moonlight work on Linux. Windows Media Player? Install the codecs if you're not on Windows. And in some cases, some of these don't even work on all browsers or platforms. A site even managed to require Flash and Silverlight at the same time (do a little scavenger hunt for that one). If you mix all this in a blender, what do you get? Picture yourself people fighting in mud. That's the actual mess that multimedia content is at the moment over the Web.

But now comes the HTML5 recommendation. With audio and video specifications. There has been an intense debate over the choice of technologies and the whole thing ended up in segmentation. Again. You have the Theora camp, who pushes forward accessibility and openness and the H.264 camp who pushes forward performance. And you have the Microsoft camp : Nothing to see here for the moment, move on. Me? I'm a bit in favor of Theora since it's not encumbered by patents and since H.264 is, it cannot be freely distributed. Theoretically, Theora is the only one that can be implemented everywhere without too much restrictions. And for those who come and say "but it's not efficient, it's not this and it's not that", encourage the effort instead of burying it. We don't need kick-butt performances, we need something that works everywhere. Like a lot of technologies, Theora should improve in time. And since it is open source, it's not like nobody can't put their two cents in it. But in all honesty, I'd say : Please, Get along. I'm sure we can work it out.

*Whew*. Got that out of the chest. Next stop, SVG support in Internet Explorer. Naaah, just kidding!

Click Here!