I'm not sure I'd call a sodium reactor more safe. Heck, liquid sodium explodes in contact with concrete, and the very reactor itself is built out of concrete. They have to clad it in thick steel as a precaution, and after a sodium leak in Japan, the sodium ate over halfway through the steel. Liquid sodium is not nice, friendly stuff.
And I don't think there's anywhere *near* enough data on thorium reactors. All the happy-go-lucky stuff sounds all too much like the sort of sales pitches that accompany each new generation of nuclear reactor.
If I had to pick one that I thought had the most promise, it'd be lead-bismuth. Now, they have their own set of corrosion problems, no question. But at least there's a damned lot of data from the former USSR on how to prevent it. Beyond that, leaks are pretty harmless (apart from economically) - your worst case scenario is that your reactor entombs itself in lead, which most people would consider *desirable* in a worst-case reactor leak. There's no explosion risk from lead-bismuth. It's a breeder approach like sodium, so little waste and highly efficient fuel usage. And the emergency circulation in modern designs is mostly passive.
But honestly, the biggest issue I have with nuclear is cost. The nuclear industry is one of the few industries out there that has demonsingtrated a long-term *negative* learning curve in terms of cost. That is, the longer we run nuclear power plants, the more added risks we learn we have to address (which costs money), the higher the disposal cost estimates versus earlier estimates, and on and on. Scaling factors mean that plants usually have to be very large which means that you don't learn as much from building lots of them with varying approaches. And the generally best way to deal with a problem of escalating costs on a design - start anew with a radically different design - means you start the learning curve over, which takes decades on nuclear due to the slow pace. And the newer approaches are often more complicated in order to solve the previous problems, which introduces new potential avenues of failure.
It's a real problem. All issues of safety and the environment aside, if nuclear can't address the cost issue, it has no future. Cost kept investors out of nuclear more than NIMBY for three decades. They've been trying again with this latest round of nuclear construction (often with citizens picking up the financial risk if not outright the tab), but the results thusfar haven't been very appealing, with lots of cost overruns.