The real threat will be an increase in the amount of video streamed and watched over the internet, NOT an increase in the video size.
Twitter commentary and Facebook will not be a huge increase in network bandwidth. Small bits of text. It's NOTHING. Text entered by humans. You can be the fastest typist EVER and you won't even touch what an audio stream is. You might say that everyone wants to provide their own audio tracks -- but that'll still pale to the existing skype traffic, which we're handling right now.
Sound might get more resolution, but we're already at the point where sound is a tiny portion of a movie. And there' s a limit to what the human ear can hear, and we're pretty close to that already.
You are also ignoring any increase in compression. We're much better at compressing sound and video than we were 10 years ago, I don't think it's unreasonable to assume that we'll get better still in the next 10 years. This will offset some of the increase in resolution.
Really, the only driver in traffic (looking at file sizes) is video.
The highest resolution movies considered right now are 4k (4096×3072) or about 6x what 1080p is. We can pretty much guarantee that with the adoption rates of tvs that it'll take at least a decade to get there. No one is even making screens at that resolution that cost less than $30k yet. 3D might change that, but at present, it seems 3D will fizzle out. And it's only (at worst) a 2x increase in the video size, but a 3D compression algorithm would likely change that to a small % over a 2d movie (the 3D frames are largely the same).
Google is able to deliver in small doses, the network they are talking about, it'll only get easier to implement in the years to come. That network would serve up a 4k movie in 1.5 minutes assuming 15 seconds for a 1080p movie (maybe they aren't using 1080p for the 15 second quote though).
In other words, you are harping on a 10x increase in traffic due to file sizes, in the face of a 100x increase in bandwidth (10Mb to 1Gb).
As I said, the real threat will be the increased # of hours of video that people watch over the internet. That can scale higher than 10x, as people shift from watching video through cable/broadcast to over the internet.