Peak vs off-peak prices for electricity aren't that far apart—up to a factor of 3 in my area for residential service. And that's for a fixed schedule of peak/shoulder/off-peak hours. More dynamic demand-based pricing of electricity doesn't work all that well for residential service; it basically requires automated load-shifting that's far more practical for industrial customers than residential.

The cost curve for internet service during peak hours is a lot steeper. I think it would take much more than a 3x price multiplier during peak hours to get any noticeable demand reduction beyond what streaming applications already do by dropping down to lower resolutions automatically. (Assuming that the base cost for off-peak usage is remotely realistic, ie. orders of magnitude lower than the metered prices we pay for cellular data in the US.)