A strong no-go theorem on the Wigner's friend paradox. K. Bong, A. Utreras-Alarcón, F. Ghafari, Y. Liang, N. Tischler, E. G. Cavalcanti, G. J. Pryde and H. M. Wiseman in Nature Phys. 16:1199 (2020). What the paper says?
This can be regarded as a founding paper of experimental metaphysics.
The Authors claim that they «rigorously demonstrate that radical revisions [of quantum mechanics] are in fact required», revisions of the type of nonlocality, faster-than-light or retrocausal effects for hidden variables. The most conservative revision is that of "absoluteness of observed events", i.e., not only there is no element of reality before an experiment, there is not either after (in some conditions). Reality is conditional or relative to various observers. In this sense, this seems to be pretty compatible with the many-words interpretation and QBism.
They generalize and make stronger the approach from C. Brukner[1] which tackles "observer-independent facts" (OIF). Here, a stronger so-called local friendliness assumption is made when the three following assumptions are met:
The subtle relationships between OIF and AOE is discussed:
In essence, it extends Peres' "unperformed experiments have no results":
The main result of the paper is then that the universe is not locally friendly:
The Authors use a mixed superposition of states (parameterized with $\mu$) so as to be able to show that Bell inequalities can be violated without violations of local friendliness, i.e., they show something else, and, in fact, something more.
A limitation of their experiment is that what serves as a friend is not an actual observer but one degree of freedom of a simple system (the path of the photon)
Therefore, it leaves open whether an actual observer would similarly be trapped in its Wigner bubble, or fail to violate the LF inequalities. This also pauses the question of, at which point does something becomes an observer (in this case, which path a photon takes not being one). Whether LF can be satisfied by an observer, and in which case, what is an observer, is the most interesting question paused by this work:
If LF are never satisfied, when they should, it means there is no absolute reality but only relatively to the observers. A clear escalation of Quantum Mechanics' insanity.