However, if we are willing to call one simulation real, then we get into the slippery slope problem which I have no idea how to avoid whereby any physical phenomena implementing any program from the perspective of any universal Turing machine must really exist.

Can you expand on this a bit?