Ok, let me get this straight. HTML 5 is supposed to offer RIA (Rich Internet Application) type abilities to the browser rendered (i.e. no compile, Just In Time-JIT) web. How JIT & no compile, markup based, mixed platform technologies are supposed to offer a comparable performance and UX/UI experience seems sketchy at best. Now let me clause with, I am not writing this idea off, just hesitant to believe that there will be comparable performance to frameworks such as JavaFX, Silverlight, or Flash. With out of browser Silverlight, or Adobe AIR Apps, or heaven forbid WPF based RIAs there really is NOT a performance comparison. I can safely assume that Silverlight & WPF (mainly because I've seen them perform) will smoke HTML 5 for advanced rendering or solid advanced RIA style interfaces. Especially in a Line of Business (LOB) type application. There just is NOT a comparison in this sense.
The video & audio elements are an entirely different rant. Sure, it looks great, the O3D demo looks great, but this is still not going to compare to Silverlight or Flash. These tools are going to smoke HTML 5. The complexity of getting video, audio, and these other advanced elements of HTML 5 supported has the same issue as Silverlight has for penetration. Sure, the browsers will start having these things built in, but that will be completed about the same time as Silverlight has similar market penetration as Flash (i.e. the 98th percentile).
Now some, such as Matthew David, wax somewhat poetically about HTML 5. His article Inside HTML5: The Browser becomes a first class RIA citizen is an interesting read. He obviously has great enthusiasm behind this idea. But I'm not buying it. The technology is not comparable, HTML is band aided already beyond belief (remember, it was supposed to render documents for linking kind of like a library – NOT for all the advanced things people tend to try and use it for these days). HTML 5 is merely another band aid, albeit a bigger band aid, then the last several versions of HTML. The core focus of the markup at this point is basically ignored.
The cool thing, I will admit, is HTML 5 does add a lot of options to the browser based web. I don't think we'll keep going down this rather archaic version of the web forever, but it does provide a stop gap between much more interactive applications. It also provides a stop gap fix for companies that aren't ready, for whatever reason, to jump on the RIA bandwagon. HTML 5 will provide great features, but I just don't see it as a prime mover, but more of a stop-gap between the next best way to connect to the web. Maybe that is Silverlight, Flash, and AIR, or maybe it is some other type of platform or tool.
In the end, something will definitely eclipse HTML 5, but HTML 5 will be great in the meantime. At least until we run out of fingers to poke in the holes of the dam (i.e. HTML 5).