I think what Breakable meant by “path of least resistance” is a biasing network, similar to biasing grids in a tetrode or pentode vacuum tube. Back then we wanted to minimize grid current. Now we need to maximize it in order to reach and exceed unity.
The most effective collecting layers will tend to be more negative, even before adding components to the biasing network. If we’re dealing with a fluctuating ‘signal’ (current flow) we may be able to use fast switching and/or capacitors to minimize power loss.
Frankly, I’m having problems seeing foils carrying huge currents. Anybody want to take a stab at the power, voltage, and/or current that Baby will output from the X-ray converter?