I find it hard to accept an argument predicated on "but it would be so inefficient" over a 15% loss when in the next breadth the alternative suggested is nuclear power, whose losses due to inefficiency are somewhere between 65% and 73% depending on the model.
Also, as has been pointed out, such losses would only happen occasionally, rather than constantly.
Yes, and they happen to coincide pretty well with our own cycles of peak power, solar especially. Not to mention we can store energy, in various ways (not just chemically), and we are poised to improve that area of technology enormously in the next few years.
Indeed, which is why I don't feel that centralizing energy generation is a good idea. A distributed system is far more flexible and far more reliable than a single point of failure. Even if it occasionally would require long-distance transfer at a bit of inefficiency.
Heck, even those consume less power once the humans stop opening it to grab stuff from inside.
Grey Wolf