[1] http://www.physorg.com/news/2011-06-pint-sized-particle-nucl...
LFTR proponents figure the simplest approach is to just go ahead and use U233 to seed the reaction. The U.S. has one ton of U233, which it currently plans to dispose of.
The advantage of Thorium is that you can maintain the reaction subcritical. That means it is much easier to shut down by simply removing the neutron source. In a critical reactor, you have to physically inject control rods containing a neutron sink (usually Halfnium or Boron). These elements can jam or be ejected under certain unlikely conditions (this is one of the things that happened at Chernobyl, and has happened at a handful of military test reactors in the US that no one has ever heard of).
In practice, however, modern reactors are designed to take this type of jamming into account. There is no known failure mechanism where it would become an issue.
FWIW, I was an engineer on a nuclear submarine.
Molten salt can be passively cooled (much better thermal conductivity) and can be moved from core to containment by gravity. Also it expands when its hot, which slows the reaction, and it's chemically stable and doesn't crack like uranium fuel pellets.