18-20 TW is somewhat more than the current 16 TW, of which renewables (mostly large hydro) account for about 9%.

A recent estimate found potential for 72,000 GW of wind power alone, worldwide.

That is 2100+ quadrillion BTU of electricity (not fuel) per year.  Humanity currently uses about 400 quadrillion BTU of energy from all sources.  That should tell you how much headroom we have.

(I'm going to regret entering this massive thread so late.  I think I would have regretted it, period.)

Sure, the potential is great. But I guess we still need to address the variability issue and get portable fuel. We could solve it by electrolysis and ammonia production, but what would it cost?

From 55 kWh, you get a kg of hydrogen, and then I guess you need some more to go to ammonia, say 70 kWh in total to get 140 MJ ammonia. Gasoline is 32 MJ/litre, so you'd need 16 kWh for one litre of gasoline equivalent, or 60 kWh for one gallon equivalent. I guess this would be somewhat acceptable, at least to Europeans. However, nuclear and HTE would make this cheaper and more efficient.

But I guess we still need to address the variability issue

Redundancy. Overbuild. The wind is always blowing somewhere. With climate going crazy, it'd be stupid not to have a massively distributed and robust energy system.

But, then, we don't appear to be any smarter than yeast.

Cheers

we still need to address the variability issue

Pumped storage.

As I've mentioned to before, it's a mature, reliable, efficient, simple technology that's already widely deployed to smooth out mismatches in supply and demand of electricity. Storage needs are surprisingly low - a few days - so existing reservoirs could be retrofitted with more generator turbines and lower catchment areas to quickly provide enormous amounts of pumped storage capacity.

Moreover, modest overbuilding of capacity allows a small amount of storage - about 5 days - to reliably provide a region's electricity supply (at least for the case of Ontario, given hourly supply and demand data for 2007). The technical aspects of this aren't a particularly hard problem.

Wind combined with pumped storage is more expensive than nuclear.

Nuclear is already combined with pumped storage and gas power plants.