1) His general point is that no single energy source will be sufficient. We're going to need a combination of many different sources of power, combined with as much improvement in efficiency as we can muster. He concludes that it is very difficult, but not impossible, and that tradeoffs will be necessary.
2) Covering the electrical needs of the world is an easier task than covering the total energy needs, and even covering the current energy needs will not account for increasing demand by the developing world. Still, this is an approach worth exploring. It's covered in considerable detail in Chapter 25: http://www.inference.phy.cam.ac.uk/withouthotair/c25/page_17...
There are experiments being done with transferring large amounts of electricity across countries in Europe based on the assumption that it will always be windy or sunny somewhere and that if power can be transferred long enough the uncertainty of supply will be cut drastically.
I don't envisage huge vats of molten salt meeting our energy needs throughout the night but I suspect that energy storage will continue to have a part to play in meeting our energy needs in the future. It seems to be relatively expensive but then so is long distance transfer.
But then, you read a quote like this from another source and the solution sounds almost easy:
"Less than 1% of the world's deserts, if covered with concentrating solar power plants, could produce as much electricity as the world now uses."
http://www.trec-uk.org.uk/images/CSP_map_squares2.jpg
What am I missing?