The ISP model is hard because it just requires fucking shit tons of money to solve, and there's no getting around the tragedy of the commons problem. With telecom infrastructure, you can either use wired (expensive and with a heavily fragmented regulatory environment) or wireless (expensive with extremely limited spectrum - see the $12+ billion spectrum auction happening now).
Disruption is well and good, but it requires new technology that delivers similar services for a fraction of the cost. Until someone figures out a way to bring the economics of delivering content to end users down by a factor of several hundred, we're stuck with the current ISP situation.
I know someone who has this service in San Francisco - they only have it because it's literally their only option besides ADSL.
The meme that dealing with problems of physics and mathematics is somehow harder or more noble than dealing with problems of people should probably be considered harmful.
Just take a look at Aaron Swartz, Chelsea Manning or Edward Snowden.
But I agree with your premise, that hard problems with large rewards require a different sort of VC pitch and may take longer to see a payoff. And are completely worth pursuing!