A recent experiment in the lab, billed as a “Wigner’s Friend” experiment, has been interpreted as a test of “local observer independence.” The authors claim the experiment shows that observers must irreconcilably disagree on facts related to quantum measurements. The paper linked below (recently accepted in Foundations of Physics) shows that this is a misinterpretation of the experiment, and that the facts are completely consistent. It is only the attribution of outcomes where no outcomes obtain that causes the inconsistency.
https://arxiv.org/abs/2105.01773
Transactional Interpretation 
Thank you very much for this great analysis!
In my opinion, the authors of the original article themselves actually admit that what is in the original Wigner’s thought experiment does not actually happen in the experiment; they just don’t describe it explicitly. Instead, they keep the point small obscured. This is the part of the article I’m thinking of (page 3):
“Before we describe our experiment in which we test and indeed violate inequality (2), let us first clarify our notion of an observer. Formally, an observation is the act of extracting and storing information about an observed system. Accordingly, we define an observer as any physical system that can extract information from another system by means of some interaction, and store that information in a physical memory. Such an observer can establish “facts”, to which we assign the value recorded in their memory.”
However, it is clear from the article that what is actually happening is that the photon entering the device will be quantum entangled with another photon and the pair of photons in the exit – if there is sufficient coincidence at the photon detectors – can be seen as the “friend photon” measuring the original photon polarization. Except that this “friend-photon” is not a memory, since it does not store a specific measurement result. It is merely a quantum system which is itself in superposition, i.e. the photon pair is entangled in superposition. So nowhere is there anything that would satisfy the “store that information in a physical memory” requirement. Hence the “friend” is not an real observer at all.
Thanks very much, Janos. I think their intent is to define ‘storage of a measurement result’ as the mere copying of one degree of freedom by another due to their correlation–i.e. they treat the copying degree of freedom as a ‘memory’. Of course, that is still wrong, since there was no ‘result’ while these systems remain in a superposition.