Another way is with the readyState of the media element. In particular you want to look for numeric value 4, HAVE_ENOUGH_DATA. From the WHATWG:
All the conditions described for the HAVE_FUTURE_DATA state are met, and, in addition, the user agent estimates that data is being fetched at a rate where the current playback position, if it were to advance at the rate given by the defaultPlaybackRate attribute, would not overtake the available data before playback reaches the end of the media resource.
I could understand this in mp4, which has a global index giving locations of all frames in the file, but how would this be possible in ogg, which has no global index? Without a global index, you can't know anything about time regions that you haven't yet downloaded.
I personally think it's a huge mistake for Firefox to support only ogg. It's a bad time for fragmentation in the push for open video on the web.
It should still be possible to get a rough estimate for when to start playback. Good enough for buffered autoplay.
The best solution would be to simply use system decoders, like Safari does. On Windows, this would be Directshow, on Linux this would be Gstreamer, and on OS X this would be Quicktime. The browser should have ABSOLUTELY NOTHING to do with video and audio decoding.
They do not. For Debian and Ubuntu, it is in Multiverse (non-free) repositories. For Fedora, it in RPM Fusion repository. In other words, there is no distribution that distributes H.264 or AAC, so system decoders would not help at all.
I know! While we're at it, who had the brilliant idea to put image decoders in browsers? I say let QuickTime decode GIFs and PNGs.
Although Chrome supports H.264 video, I believe Chromium does not.
Mozilla already has compile flags that generate binaries other people aren't allowed to distribute -- the logo and trademarks are decidedly non-free, hence IceWeasel.
My understanding is that HTML5 browsers will attempt to buffer as much as possible; is there any reason you'd want less buffering?
If this were abstracted to the browser level (discounting the fact that these long-format providers will likely never move away from their current flash/silverlight implementations), my choice of online tv watching would no longer be dependant on the competence of the content-providers buffering algorithm: currently, despite the fact that I enjoy several shows provided by CBS.com, I will not watch them online due to their shoddy buffering algos. I will not watch them via antenna, either (I watch on my own schedule, thanks).
Of course, as stated, these providers are hopelessly locked to their delivery platform due to DRM and contractual obligations. Ergo, most of this point is moot when it comes to long-format TV-oriented content distribution.
[1] in my experience, CBS does not buffer enough before playing, and does not buffer during the pause state. It also does not indicate how 'full' the buffer is during pause state, like hulu does. Netflix is, in my opinion, the best on this as it indicates on the seekbar the buffer level, and adjusts bitrate based on perceived bandwidth. When your bandwidth drops below the perceived bandwidth, it gives you an accurate status indication as to buffering activity. CBS, contrarily, simply buffers to the original perceived value and resumes play. This is problematic if you begin with high bandwidth but midway through stream something happens to negatively effect your bandwidth. If this situation occurs, you're likely to encounter a slideshow: video resumes for 5< seconds, whereupon it buffers once more. This results in >2s stutters every 3-10 seconds. Because CBS does not buffer during pause states, this renders the video unwatchable except during time periods when my bandwidth is constant: living on a highly variable cable node makes this time period nigh impossible to predict. Therefore I do not watch CBS shows - antenna or streaming - at all.
There's a lot wrong with Theora, but it's irrelevant to the discussion: what matters is what people actually use. If today it came up that there was a big patent on JPEG, would you propose that the entire web stop using JPEG tomorrow?
No matter how good or bad Theora is, the internet will not simply drop the existing standard formats and switch. It doesn't work that way.
This is also why browsers should use system decoders: it would make it easier to use a different video or audio format, since one could use anything that people had decoders installed for. This would stop the internet from being locked into a single format.
"Use the system's video" sounds attractive, but just pushes the problem down the stack. Your mileage may differ, but I've seen people try it before....
Until HTML5 has basic feature parity with Flash's video capabilities, it will probably not be used very widely. This is of course ignoring the fact that Firefox doesn't support a video format that anyone actually uses, and the fact that IE (still unfortunately the most popular browser) doesn't support anything at all.
The sad reality is that a single system often has the advantage over a group of conflicting, incompatible systems; even if the single system is bad, one at least only has to compensate for its weaknesses, and not for everything else's.