I am not very familiar with the spec, but is the spec not refering strictly to signing/linking keys to a video? The question still remains who will host the video? The data has to come from someones server. And if you want to offer experiance comparable to YouTube, those servers need to: 1) be fast 2) content will be persistent, like 5 months from now. Futher more if you have 10k people using this the storage and bandwith scales with this number. So still not obvious to me how exactly that can work.