[tehwalrus]

I think the point is that a "lab model" would be useless, since the problems change non-linearly at scale and speed... and computer simulations of fluid dynamics and physical systems are pretty good these days, since computers got so much faster (physicists add more and more complexity/realism to their modelling such that a simulation always takes ~3 days to run, however good their computer. Our computers are awesome now.)

Also, some parts of it are tested, e.g. the roller coaster magnetic thrusters/brakes.

I do understand the difference, but the point stands for all revolutionary, as opposed to incremental, technologies. We will need some, how do we fund them?