In quantum mechanics, we are always calculating probabilities. We get results like, “There is a 50% chance this radioactive nucleus will decay in the next hour.” Or, “We can be 30% confident that the detector at position *X* will register a photon.” But the nature and origin of quantum probabilities remains obscure. Could it be that there are some kind of “gears in the nucleus,” and if we knew their alignment, we could predict what would happen with certainty? Fifty years of theorem-proving have made this a hard position to maintain: quantum probabilities are more exotic than that.

But what we *can* do is reconstruct a *part* of quantum theory in terms of “internal gears.” We start with a mundane theory of particles in motion or switches having different positions, and we *impose a restriction on what we can know* about the mundane goings-on. The theory which results, the theory of the knowledge we can have about the thing we’re studying, exhibits many of the same phenomena as quantum physics. It is clearly not the whole deal: For example, quantum physics offers the hope of making faster and more powerful computers, and the “toy theory” we’ve cooked up does not. But the “toy theory” can include many of the features of quantum mechanics deemed “mysterious.” In this way, we can draw a line between “surprising” and “truly enigmatic,” or to say it in a more dignified manner, between *weakly nonclassical* and *strongly nonclassical.*

The ancient Greek for “knowledge” is *episteme* (επιστημη) and so a restriction on our knowledge is an *epistemic restriction,* or *epistriction* for short.

A *trit* system is one where every degree of freedom has three possible values. Looking at Figure 4 of Spekkens’ “Quasi-quantization: classical statistical theories with an epistemic restriction,” we see that the valid states of an epistricted trit follow the same pattern as the SIC-allied Mutually Unbiased Bases of a quantum trit. But that is a story for another day.