Good point. On the other hand, DRM will come to HTML5 anyway, whatever we try. So, I think it's better to give it a spec, and have one, unified and well-known way for the DRM to operate, instead of 20 different implementations.
Specs are supposed to make things simple. I'd rather have DRM vendors struggle with the problem individually at every step (and that includes browser integration)
For one, because they'll fail in the most hilarious ways (and each one separately), so there will be backdoors, or at least a good laugh or two. But also because it drives up the cost of DRM, hopefully making it less attractive.
Should DRM manage to get hold even despite such problems and Amazon notices that they can reduce the price of their media package by 1$/month if they drop the DRM, the resulting massacre will be great to watch.
