If a bridge worked 99% of the time, or a damn worked 99.999% of the time it would be considered outrageous, but here we are in the era of Web 2.0 and still the fundamental engineering of the internet remains flawed. Of course, the flaws are almost by design, but there are plenty of technologies and protocols such as IPv6 and multicasting which could improve the performance of video over the internet.
The problem currently is that the money that was poured into infrastructure in the late nineties still hasn't paid its way, so investment in technologies already with us will have to wait.
The next problem is that there is an obsession with retro-fitting and supporting standards that are well past their sell by date.
Nothing shows up these issues more than the current state of the browser market where there are, once more, a handful of browsers, all hugely incompatible, although purporting to use the same standards.
That's not the problem - the trouble comes when your drive your car over a certain bridge and it breaks the bridge.
Just as dial up and broadband created a two tired system for the internet, expect this to further break into QoS (quality of service) and non-QoS markets in the future and not just for your broadband; similar services are likely to exist for your mobile devices also.
After all, wouldn't you pay a bit more if you were sure the damn wouldn't break...