Instead of socks, lets do photons.

A photon is red shifted or blue shifted depending on the motion of the observer.

Our entangled photon pair are emitted in opposite directions and are observed by Alice and Bob.

Alice has a motion away from the photon and to her the photon is red.

Bob has a motion towards the photon and to him the photon is blue.

When we measure other oscillating properties, like polarization, Alice and Bob similarly get differing results. The photons appear to be independent.

So we filter for successful entanglement, e.g. where the wavelength is the same and thus the motion of Alice and Bob, relative to the emitter, is the same. We only consider that subset of experiments. And magically, other properties also now correlate. We sometimes do filtering explicitly, filtering for similar properties, sometimes implicitly, e.g. mounting detector Alice and Bob to the same rig as the emitter, so their motions correlate for our experiment.

But of course, you say, (realizing the wavelength of light isn't a property of the photon, its a *net*interaction* between some oscillating property of Alice and some oscillating property of Bob and their corresponding photons). You're actually filtering to make that property the same between Alice and Bob and so the net interactions the same.

But then our photons were always mirrored all along, only Alice and Bob's relative oscillating state (compared to the emitter) where undefined, filtering the experiments ensured Alice and Bob were oscillating the same way when the measurements occurred.

Our photons had oscillating properties (which are motions) which added up to the net motion over space. Their macro motion is defined and knowable, so those little oscillationary motions were always defined and knowable, and must add up to the macro scale motion.

And of course such net properties are *combinations* of properties between photon and observer, so there are more combinations than unique properties. You only need to filter for a few combinations, for other combinations to also correlate.

Now imagined you modelled wavelength as if it was a property of the photon. You believe the photon has a particular wavelength. But you could never know what that will be, because you do not yet know the observer's oscillating pattern until you select which observer. To Alice it will be red, the Bob, blue.

Then you have a probabilistic model trying to predict the future, such a model will be complex and imprecise and unknowable, almost magical. i.e. the Quantum model.

And it would be prone to logic flaws, e.g. calculating the threshold by which correlation is no longer random, claiming that model has no prior knowledge of the system (which is untrue, you filtered for that subset of experimental results, that model has only the filtered data) and thus is proof of the magical nature of said system. i.e. the Bells test.

So, here we are. A fork in the road. Down one path the universe is locally unreal, down the other path Schroedingers model is locally unreal until you select the observer with its local oscillating pattern.