Too much focus seems to be spent on some binary yes/no vote for a codec of choice, when the subject is the HTML Language. For HTML 5 to evolve with forward momentum, I think it needs to increase support for the objects it models...

LSL, the Linden Scripting Language used in Second Life, greatly impressed me by how it structures the language very closely to the models it controls. As an object branches into a hierarchy of model elements, the methods controlling those elements follows much the same hierarchy.

HTML currently treats video, and thus codeces, with utter agnosticism... Something of a certain size fits the page, and its function is ignored. If the language, however, were crafted to fit the model of video frames and their playback functions over time, there could be far more interactivity programmed in HTML. More importantly, "HTML 6" or later progeny could heap on more methods for human input to interact with video output... or even other input/output combinations. (If the CANVAS tag in HTML can read 3D gestures like a Wii controller, why not webcam gestures?)

In order for HTML documents to properly support the use of the data, I think it ought to provide some support for its standard formats. The result should allow greater interactivity, and I find that a high priority.