"If we discover the mechanism, we could theoretically manufacture a room-temperature superconductor (above 0 C). This would really be a leap forward :)"
Superconductive power transmission lines are huge. You now can move electricity anywhere with zero loss. I've seen loss numbers for the US electrical grid as high as 20%. Imagine if we got 20% more power to use from the same generators?
MRI machines are a second use case. Currently, most MRI's use a superconductive metal to run gigantic currents through a coil to create gigantic magnetic fields which they measure. In order to do this, the MRI needs to be cooled enough to make the coil superconductive which requires at least liquid nitrogen and sometimes liquid oxygen. This is what makes MRI's expensive. Without the cooling requirements, they would likely be as expensive and thus as common as X-ray machines.
Believe it or not, getting the extra 20% out of the grid is peanuts compared to the ability to move power wherever.
In West Texas there is a huge amount of wind power installed but it's difficult to get that power from West Texas to anywhere that people use it like Houston or Dallas.
Right now people pay different amounts for electricity depending on where they are in the globe because electric power is actually a locally produced, locally consumed good. It's a commodity to be sure, but it's actually harder to transport than oil is. So oil prices are fairly flat worldwide but electricity prices can vary by a factor of 10 or more.
High temperature superconductors would make it really feasible to interconnect the world's energy grids and allow anyone with the ability to generate utility-scale power to sell into the wholesale market.
That means you could cover the Sahara, the outback in Australia and the deserts in the Americas with solar panels and run a fairly flat and smooth solar-only electric grid.
I'm thinking this falls out as: we only see losses up to 20%, because for transmissions where it would be higher, it's not feasible, so not done, so we don't see it.
I guess your scenario of outback solar powering the world would work with undersea (why not?) superconducting cables. But orbital power stations need a space elevator (perhaps more feasible, if it's just a cable, not people/goods transport).
I worry about room-temperature superconductors... what if they carry massive power and get a tiny bit hotter? I suppose there's the same solution of circuit breakers as for regular cables, but seems more dangerous with little head room and no environmental cooling. Don't touch that! You'll warm i ka-BOOOM
There will be engineering challenges no doubt, but they're technically quite easy to overcome. Expensive and a bit impractical, but very doable.
A few thoughts:
1. Make the total capacity up of multiple superconducting strands. That prevents a single failure from totally blowing everything up.
2. Create variable loads at the power generation equipment as well as distributed throughout the grid. They could be giant resistors or single-use loads that are designed to burn up, or probably some combination of both. If/when something goes wrong they can be brought online to absorb the energy currently in the grid and prevent really nasty failures.
3. Any kind of really long distance transmission is likely going to be HVDC and that means capacitance isn't a huge problem. Which means you can bury the cables underground (perhaps 500 or more feet) where the ground temperature is relatively constant in the 50-90f range (depends on latitude). http://en.wikipedia.org/wiki/Geothermal_heat_pump
4. There probably won't be too many of them and that means you can go to some pretty serious lengths to ensure that they're highly reliable.
I would envision a single cable from Australia to China to Russia to Europe. A dogleg down through Africa and head onwards to cross North America horizontally with another dogleg down to South America. If we've got crazy money, maybe even connect back up over the Pacific and have the South American and African doglegs connect over the Atlantic.
The whole thing would easily cost hundreds of billions but the benefits would be really big. New power generation could be done where it's easiest rather than near/in population centers.
You'd definitely encase them in a pipeline. Something durable. That would be much cheaper than multiple pulls. 1/2" wall 8" diameter steel pipe is real strong and pretty cheap compared with doing a second pull for a redundant cable. But you make a good point about redundancy.
Not to be a cynic, but even if we got a room-temperature superconductor it might not be useful for power lines. The material may end up being too brittle or too expensive (most of these superconductors are complex ceramics, which are very brittle and hard to form into wires). Alos, superconductors carry electricity with no resistance, but only up to a point. If you try to stuff too much current into a superconducting wire, its superconductivity will turn off.
From what I have read I'm sure MRI machines use liquid helium because the superconducting magnet needs to be very cold and liquid nitrogen isn't cold enough. LN2 was used as a sort of buffer/insulator but has been replaced by cryocooler cooled dewar.
MRI machines are a second use case. Currently, most MRI's use a superconductive metal to run gigantic currents through a coil to create gigantic magnetic fields which they measure. In order to do this, the MRI needs to be cooled enough to make the coil superconductive which requires at least liquid nitrogen and sometimes liquid oxygen. This is what makes MRI's expensive. Without the cooling requirements, they would likely be as expensive and thus as common as X-ray machines.