HVDC can be as efficient as 97%, compared with the US average of 93%. That 4% "inefficiency tax" would likely pay for long-haul HVDC connectors (as outlined in one of Obama's 2008 energy proposals) within years.
[1] http://en.wikipedia.org/wiki/High-voltage_direct_current#Adv...
http://en.wikipedia.org/wiki/HVDC_Inter-Island
So it is a proven technology to have this as a long distance link (been in operation since 1965) and it is capable of bi-directional power transfer as well.
Mercury-arc rectifiers (the technology replaced with IGBTs and GTOs) were very large and expensive, and also less efficient (even more so before the 1930s or so) so HVDC only made sense for submarine cables (which have huge capacitive losses). With IGBTs and GTO thyristors it start to become feasible to e.g. do an HVDC line across a continent.
As an example the HVDC inter-island was built using mercury-arc valves (it included a submarine leg), but they have since benn replaced with solid-state devices.
HVDC is also a good to tie two separate grids together; since they will be on different time-bases you can't just directly AC couple them.
[1] http://en.wikipedia.org/wiki/High-voltage_direct_current#Dis... (note: some of these are due to AC parts having economies of scale and decades of efficiencies and process that HVDC may not be able to tap into without widescale implementation)
* Difficulty cooling. If you build next to an ocean or river, you can just use some of that water to provide the cold end of the temperature differential that you're using to generate power. Deserts are trickier, and more expensive.
* Transportation. If you build next to navigable waterways, you can ship really big components on barges. In deserts, you can ship some things by rail.
But hey, at least it's politically convenient to stick scary power plants in deserts.
http://en.wikipedia.org/wiki/Thermodynamic_cycle
Water just happens to be a particularly convenient coolant.
Water has some disadvantages. It's a good neutron moderator, but with its low boiling point you have to keep it under a lot of pressure (160 atmospheres for most light-water reactors). That means you need very strong, thick steel, and a huge oversize containment dome, since if a pipe breaks, the steam will flash into 1000 times as much volume. Then some of it will split, and you'll be at risk of a hydrogen explosion, which is what we all saw at Fukushima.
Molten salt, on the other hand, works at atmospheric pressure, and if something leaks it just drips out and cools into rock.
Sodium has a disadvantage in being reactive with oxygen and water, but it also works at atmospheric pressure. The integral fast reactor design uses a big pool of sodium, which provides so much thermal inertia that Argonne was able to switch off the cooling system entirely, and the reactor just quietly shut down.
Either design works at higher temperatures than LWRs, giving better thermodynamic efficiency.
Actually, we have, albeit with varying degrees of success: http://en.wikipedia.org/wiki/Solar_power_plants_in_the_Mojav... http://articles.latimes.com/print/2012/mar/04/local/la-me-so...
If you've ever seen the insulator jacket in an underground HVDC cable it is huge.
How about flywheels?
http://spectrum.ieee.org/energy/the-smarter-grid/superconduc...
http://www.htstriax.com/columbus.html
http://www.ornl.gov/sci/htsc/documents/pdf/fy2003/HTS_Cable_...