You could spin this further. Me thinks SJ might have been wrong with his analogy of "trucks" and PCs. Perhaps instead, TVs will become the center of processing power and memory/state in the home, with tablets/phones, in combination with an external keyboard, accessing the power like terminals used to in the 70s, except the mainframe is now in your living room instead of n km away. That would also require a unified OS so it's not one or two generations away. Still, this could be something to think about...
No, Microsoft tried to push things this way with their home media server in the early 00s, but it is too complicated for normal people (non hackers) to administrate. So everything is going to the cloud.
Your solution would be great if bandwidth didn't exponentially increase each year.
Where in the US is bandwidth increasing exponentially each year? Can you name even a single residential neighborhood in the US that has had significant increases in available broadband speed in two consecutive years? FTTN/FTTH services are still pretty rare, and basically nobody has the choice between multiple such services. If you've got a choice between DOCSIS 3 and VDSL, you're in a very special market.
(And it should go without saying that cellular ISPs aren't an option for a multi-user household.)
Also found this on Wikipedia (not saying it's exponential here but still supports my point):
>In 1998, all of the United States backbone networks had utilized the slowest data rate of 45 Mbit/s. However the changing technologies allowed for 41 percent of backbones to have data rates of 2,488 Mbit/s or faster by the mid 2000's.
Nielsen's law is basically saying that the highest available bandwidth anywhere increases exponentially. That doesn't mean that any one customer gets an exponential increase, or that the majority of customers get any increase at all. And backbone capacity increases don't help unless they significantly outpace the growth in the number of subscribers.
