A lot of research into plasma confinement is completely empirical. No theory at all. The scientists try new settings, see how they work, and draw a line on a graph extrapolating from them to higher power settings. That's why you need all these small experiments, which have no hope of generating net energy: because you're trying to solve plasma confinement by tinkering, without much of a map to guide you.

Confined plasmas have complex behaviors that are really, really hard to model computationally. Experimentation definitely beats computation in this field.