The third experiment is produced by making several additions to experiment 2.

First, add a fast microprocessor to the output of the result screen that very quickly calculates (from each particles x coordinate on the screen) the probability of each impacting particle being part of a wave or particle pattern using the algorithm generated from the optimally deconflicted (1) and (2) data described in Experiment 2A.

This prediction probability, p, is recorded in R3.

The calculation must have been begun, if not completed, before its associated slit detector signal arrives at the w-w-devices.

According to MBT VR theory, this puts the experimenter in a position to significantly predict the output of a natural (based on an atomic decay process) random number generator.

This is impossible by standard QM theory but easily explained from a VR perspective. See slide below for conceptual set-up.

(The following experiment setup diagrams, are used only to illustrate the concepts, not to define the actual physical setup.)

An interesting variation: if the necessary timing can be achieved (roughly, particle travel time from slits to result screen + algorithm computation time + light speed time from R3 to R1) < delayed detector signal travel time from slits to w-w injector located just before R1), the experiment would be able to turn the w-w device on or off to directly contradict the algorithms prediction immediately after the prediction has been made but before the particle reaches the w-w device.

That would produce wave (diffraction) patterns with recorded w-w data available and particle patterns without any available w-w data recorded — both in direct conflict with particle-wave complementarity.

I believe the system (the VR rendering engine) will treat this variation as two independent experiments to resolve this forced conflict.