Relational Quantum Mechanics and Event Centric View

[sayak83]

It doesn't. Firstly, it isn't even the only relational approach to quantum theory (and it isn't much of a relational interpretation at all, but that's less Rovelli's fault than it is the fact that "relative state" was taken and "relativistic quantum mechanics" is not an interpretation but (more or less) the reformulation of quantum theory with the appropriate invariances but without fields or the full-fledged (so-called) second quantization (the term relativistic quantum mechanics is most often used to refer to the modifications of NRQM we find in e.g., the Klein-Gordon equations or that of Dirac).
The Perspectivalist interpretation of Dieks (see e.g. "Quantum Mechanics and Perspectivalism") and his early structuralist-based work on quantum foundations are more in the general intellectual, philosophical, and scientific traditions associated with relationalist approaches and relationalism.
The same is true of the relational quantum interpretation of Dennis Dieks (see e.g., "Objectivity in Perspective: Relationism in the Interpretation of Quantum Mechanics"), or Teller's relational holism ("Relational Holism and Quantum Mechanics"), or Mermin's Ithaca interpretations with its "correlations without correlata" to name a few.
Then there are a slew of related approaches which do not posit any unobservable worlds or hidden variables and their offshoots, such as the modal interpretation or Healey's pragmatist approach or the informational approaches and corresponding interpretations not to mention a slew of others which claim to resolve the issues that RQM likewise claims via more coherent means that also ultimately (I think) fail, such as quantum logics or quantum probability-as-interpretation (i.e., we need to understand that quantum theory describes objective, real systems but that it does so according to non-classical logic or non-classical probability or both and that if we understand QM in these terms than all the problems evaporate).
None of these approaches, in my view, succeeds. In fact, Mermin is now firmly in the QBism camp. But the point is that Rovelli doesn't offer the only interpretation by any stretch of the imagination which claims to explain quantum mechanics without hidden variables or many-worlds or other such additions to the ontology and/or formalism.
Nor, strictly speaking, is it true that he proposes a no-collapse theory. The problem is that several of his claims about RQM seem to be at least in part in contradiction with one another (see e.g., the reply of the authors to Rovelli's response to their earlier work in "A reply to Rovelli's response to our "Assessing Relational Quantum Mechanics''"). RQM, according to Rovelli, is not a "collapse" interpretation because when he and those who follow RQM speak of the collapse of the wavefunction, they don't take the wavefunction as being physically "real" but rather encoding information or probabilities, or more rather it is understood epistemically. Thus the collapse is simply an updating rule as in QBism. But the problem is that RQM claims to be realist in nature. An there is supposed to be an ontology. Simply claiming that any and all systems count as observers and facts relative to the information obtained in interaction and ONLY via these processes doesn't make for a realist interpretation (at least not readily in any coherent manner) because the entirety of the formalism and the mechanisms and the outcomes of measurements and interactions are understood in terms of epistemic processes, relative facts, etc.
This is much like QBism, which is also a no-collapse approach to QM in the same way that RQM is, does not appeal to any hidden variables or many-worlds, and is entirely consistent. Where RQM and QBism differ is that QBism bites the bullet. In wanting to retain, as Rovelli does, the idea of the completeness of quantum theory, eschew hidden variables and worlds, and provide a consistent interpretation, QBists hold that quantum mechanics is about agents. Like Rovelli, probability is emphasized and the quantum states are understood epistemically. Unlike Rovelli and RQM, QBism doesn't try to claim that somehow realism and an ontology emerges magically from interactions and events (that are ill-defined in general in QM, and worse so when one attempts to go beyond using time as a parameter in either the observables or the states themselves). Instead, no ontology is assumed (but Fuchs at least and most of the Boston-based Qbists hold themselves and their interpretation to be realist, like Rovelli).
QBism does run into some of the same problems as RQM. In particular, it is hard to understand how a betting scheme, no matter how consistent it may be, can tell us much about the external, physical world. RQM has the same problem, but refuses to confront it. Instead, at every turn, fundamentally different descriptions of states and physical properties by observers are all supposed to be equally true because...they ought to be? Because that would be a nice world? It's not clear, but other than declaring that by fiat contradictory state assignments and descriptions of the same physical system are not in conflict no matter how clearly they actually are because agents cannot both observe the same contradictions at the same time is honestly little better of explanation for seeming paradoxes than what Bohr and other founders of the so-called Copenhagen interpretation had to offer.


This is simply wrong. I've listed many that do not do so, such as QBism, which takes much more at face value the quantum formalism as does indeed the view of the founders who believed quantum mechanics to be a complete and consistent theory (Bohr and Heisenberg especially, but later e.g., Wigner, Wheeler, etc.). Practically the entire slew of approaches that fall under the so-called informational interpretations and/or statistical interpretations (or ensemble) in the same vein as Ballentine but with the addition of irreducibility and completeness fit your description.
Also, RQM is and has been questioned and found wanting at a basic level of coherence. One can maintain (and people do) that it survives these critiques, but it is hard to see how such built-in, basic contradictions as 1) holding that states in QM are epistemic and evolution and collapse (or equivalents) are to be understood epistemically
and
2) RQM offers a coherent, realist interpretation of QM by claiming that the "relative facts" obtained by observers (which is everything) are objective in some sense and ontological in some sense because truth and facts can be objective as long as we understand that them to be objectively relative to interactions of a formalism that isn't ontological or really even external other than by the kind of fiat Bohr seemed to enjoy with his complementarity.

I always have some difficulty following your arguments. But can you clearly and concisely tell me what is wrong with the following picture below?
Reality consists of a set of interaction events. These interaction events are defined through a set of property variables which have values at these interaction events. These property values can usually (but not necessarily) be expressed in a compact way as a set of objects(or fields whatever) doing the interacting. Objects (or fields etc) are only secondarily real in this context... pragmatic ways to encapsulate the information available during these interaction events.

Next...
These interaction events are often correlated. That is the property value of one set imposes some constraints on the likely property value set of another set of interaction events. Quantum Mechanics describes what these probabilistic constraints are and how it can be used to partially predict the properties and their likely values for a target set given the values of another set is known apriori. Wavefunction is a convenient way to encapsulate these relational probabilities in a mathematical structure.

Now does this contradict anything in QM?

 

[LegionOnomaMoi]

I always have some difficulty following your arguments. But can you clearly and concisely tell me what is wrong with the following picture below?

Well, the first and main problem I have with the idea as you have outlined it here is that this description makes little contact with RQM and perhaps none at all with “standard” quantum mechanics. Of course, this is not a criticism of what you wrote but rather a constraint imposed upon you by the communication medium: this is an online forum composed primarily of non-physicists (let alone physicists whose work concerns or has concerned quantum foundations) and a forum post can only go so far.

But it leaves me in an awkward position. On the surface, there is great appeal in the many presentations of RQM by Rovelli and others. In a concise, summary description of some of the main features like the one you have provided, it seems even more reasonable. But such a description glosses over the nature of the problems that RQM claims to have overcome, which are after all quite nuanced, often technical, and relate the quantum mechanical framework to an operational one alongside a historical and a conceptual development spanning many decades that is where I would have to locate any issues I have with RQM (or any interpretation, for that matter).

I will try to do as you did, and summarize key issues without anything resembling formal sketches or in-depth discussion. I will supplement this with two references that are relatively non-technical I managed to locate which discuss some of the problems I have (albeit with different purposes, emphasizes, etc., in mind) as well as others.

Basically, my problem with RQM concerns whether it can be said to have achieved all that is claimed. I do not think so. But let me quickly summarize some of the key assumptions and claims about RQM in relation to the standard quantum mechanical framework. I recommend looking over these to find areas that may seem to run into conflict and think about how they might be resolved:

1) QM is complete. That is, there are no “hidden variables” (local or not) nor a point at which QM breaks down and must be replaced by a more fundamental theory. Rather, classical physics is at best an approximation of an underlying reality.

2) RQM is realist, and holds that quantum systems are part of an observer-independent reality (external to observers)

3) Observers, observation, and measurement play no special role in RQM. There is no wave-function collapse in the usual sense because there is no ontic interpretation of the wave-function and the “collapse” is interpreted epistemically, i.e., it stems from information gained from local interaction.

4) RQM is local.

5) Events and correlations between systems play the fundamental ontological role in RQM, not states.

6) States and facts are relative to the observer. Properties of physical systems are observer-dependent as are states themselves.

7) In order to keep all of the above and attempt to avoid paradoxes or similar issues, proponents of RQM maintain that systems cannot be correlated with themselves, contradictions normally attributed to differing measurement outcomes in e.g., Wigner’s friend or Schrödinger’s cat type experiments are “resolved” by asserting that different measurement outcomes of the same system cannot occur at the same time, observers cannot consider themselves or their own states quantum-mechanically (i.e., ostensibly because of the no self-correlation rule) and the contradiction between measurements of the same system by different observers can be attributed to different facts obtained via interactions ordered temporally such that one can never actually obtain contradicting information of the same system at the same time due to separation in space, time, or spacetime.


Now for my core concerns:
Mostly, all of the issues I see involve or result from the problem of interactions. This is at the core of RQM, and equally it is vital that the time of interactions be well-defined in order to make basic claims about the consistency of RQM. Yet it is not at all clear that such times, or indeed interactions at all, are well-defined. RQM, as an interpretation of QM, clearly makes use of the formalisms that do after all involve interactions we call measurements that are made at specific times and places. But one thing that enables these interactions to be well-defined is a division imposed between the systems, the measuring apparatus, the environment, and the observer (as well as perhaps other crucial components of both realized experiments and experimental schemes such as ancillary systems and so forth). By its radical egalitarian interpretation of observers, as well as the demand that properties and states make sense only in terms of interaction events, it is hard to imagine what it means for an experimentalist to even observe a “pointer” position in an experiment or better still how to make sense out of such simple properties as that of spin in systems like single electrons when the values these take (given a well-defined interaction time) are nonetheless basis dependent (this is related to the so-called “preferred basis” problem or just basis problem).

In any case, it is easy to talk about an interaction in QM and apply the formalism as has been so successfully done for decades and decades now. But this was and has been due to making certain assumptions about the nature of quantum systems, properties, states, and measurements that are problematic and that in fact RQM is claimed to have resolved. Namely, we never obtain the states predicted by the theory for almost any sort of interaction we can conceive of or realize experimentally. In particular, treating the measuring apparatus quantum-mechanically yields states never observed, making it necessary to supplement the theory with the projection postulate or collapse or something like this to ascribe definite values to measurements. RQM denies that these are objective, ontological properties or states but are rather relative to the systems involved in the interactions that are supposed to produce measurement outcomes. In particular, the interaction time at which we would have spoken of a measurement is instead defined to be a probability distribution that can be derived in terms of correlations between systems. But this in turn means that would-be measurement by the would-be observer is defined externally to the observer by another “observer” for whom the values obtained are ill-defined or denied completely and indeed for whom the “measurement” interaction cannot be said to have taken place.

In short, RQM gives a radically “democratic” approach to observers and observations that renders all systems equivalent while maintaining a complete interpretation of QM without additional supplements beyond the Born rule (or its equivalents or generalizations of these). In relativity, this is fine because the “observables” like velocity or time that are relative to reference frame are not probabilistic but governed by particular symmetries and ensure all properties and states are always and everywhere well-defined. In QM, the symmetry is given by unitary evolution, but this leads to the superpositions and entangled states we never observe without introducing an asymmetric interaction we call a “measurement” or observation. Rovelli keeps both the unitary evolution AND rejects the asymmetry, and so RQM contains blatant contradictions. These are supposed to be resolved by insisting that measurement outcomes be defined locally in terms of the interacting systems involved. But these “quantum events” require additional, external aspects in order for the required correlations to yield the distributions at a particular time. Moreover, as even Rovelli has had to admit, RQM cannot be defined locally in the manner it was initially claimed but must yield to a kind of non-classical causality that is not explained at all because the only thing we no about such explanations would involve admitting non-locality or giving up one or more additional assumptions of RQM.

As for realism, I will say simply that one can get an idea for how difficult or impossible it is for there to be any kind of coherent, ontological approach to a world with objective systems by considering that according to RQM facts are always relative to the observer, no observer is privileged, and different interactions will yield different assignment of “facts” even in the same interaction depending upon whether we consider it from the “observer” perspective making a measurement or that of the system under observation. All facts about all systems (including that these systems exist in any sense in this or that state) are relative to all other systems, so the states and properties of all systems are relative, and there is never any quantum event that we can say took place at any particular time in any particular way with any particular outcome. So what, exactly, is “real” about anything? What is the good of maintaining an interpretation like this as opposed to one which gives up at least one claim made by RQM, such the incoherent claim that RQM is a realist theory describing an objectively-existing external (quantum-mechanical) world?

Here are the two papers which go into more depth than I could here about these and other issues. I would say more than either paper and I also do not agree on the whole with either paper completely, but this is already a long post and I have only touched upon a sketch of some of the aspects of a few issues.
A quintet of quandaries: five no-go theorems for Relational Quantum Mechanics
Assessing Relational Quantum Mechanics

 

[LegionOnomaMoi]

It might be useful at this point to briefly cover what issues or problems QM is often thought to have or to present us with in order to clarify how RQM does or does not resolve these issues. After all, popular/sensationalist presentations ascribe all manner of “sins” to QM in all manner of ways. Some of these are relatively innocuous simplifications, but many are deliberately phrased in order to elicit the “wow”-factor rather accurate understanding, are intended to align the treatment of quantum theory with some particular worldview advocated for (often spiritual or even paranormal), and/or are generally designed to make QM seem as bizarre as possible in ways that are not at all necessary and do far more to mislead than anything else.

Most of the problems that physicists and others (e.g., philosopher) have or have had with QM concern the measurement problem (and most of the other issues they have or have had with QM are result from or are related to the measurement problem). The term “measurement problem” is used in different ways even in the technical literature and more broadly still in popular presentations. One can find it used in relation to basically every paradox or conceptual problem or seeming inconsistency and so forth in QM, and even as a term to mean any and all of these. In quantum foundations, the term is used more narrowly, and as most of the problems/issues/paradoxes/etc. that sometimes serve to define “the measurement problem” are instance of, equivalent to, or result form this narrower usage, it is the narrower usage I will briefly describe.

First, a touch of textbook QM. Quantum mechanics, like many (or most!) theories in physics rests upon certain postulates. One of them usually given as something like “to every physical system there corresponds a ray in some Hilbert space we take as describing the state of that system.” At this point, one means by “state” that one is actually dealing with a “pure” state, which formally means that the phase is irrelevant (i.e., doesn’t change predictions upon measurement). As Hilbert spaces are equipped with a norm (by definition), we often implicitly or explicitly describe the ray as a normed vector in the same Hilbert space.

Next there are measurements. I won’t present this part axiomatically but descriptively in a manner similar to the textbook presentation. Basically, to every would-be measured property one associates an operator called an “observable”, and then one describes measurement as being modeled by this observable acting on the state of the system. More formally, the “observable” operator “acts” on the system’s state (read “ray” or “vector”) by projecting it onto a subspace of the Hilbert space originally associated with the system.

Put more simply: Upon measurement, the state of the system is said to “collapse” to some definite, determined “value” and this is modeled formally by projecting the mathematical representation of the state onto a subspace.

In this presentation, which corresponds rather closely to what Bohr had in mind, there is no measurement problem and most of the other problems of QM are likewise absent (or swept under the rug) because the presentation carefully avoids dealing with the fact that the measuring apparatus, which we describe using an “observable” operator, is not treated from within the formalism as a physical system. Rather, it encodes information about different outcomes that may be obtained by the measuring apparatus, which is treated “classically” in the sense that it is represented not by a quantum state but something that “acts” on quantum states and tells us what we will find in the read-out of our approximately classical apparatus.

But quantum theory is the theory that we take to describe atoms and subatomic particles, and our measurement apparatus is composed of such things. So it should, at least in principle, admit a quantum description. And according to the postulates of QM (in particular, the one I’ve given above) there exists some ray in some Hilbert space corresponding to the state of the apparatus.

This means that we can (at least in theory) use more advanced methods from QM to describe the measurement process by representing it in terms of an interaction. Using this approach, measurement becomes an interaction between two quantum systems. In the formalism of quantum theory, we can say that these two systems become entangled.

And that’s where we run into a wall. Once the systems become entangled, the state of the new composite system “Quantum System + Apparatus” no longer corresponds to anything we ever observe, and indeed in general we wouldn’t even know what it would mean to see something like this (other than via famous(ly cruel) examples involving cats that are both dead and alive and similar spectacles).

That, in basically a single page or less, is the measurement problem and therefore at the heart of the “paradoxes” and so forth of QM. The first approach, in which we superimpose a classical world over a quantum world by relegating quantum mechanical states and dynamics to systems we prepare and then measure classically using instruments we treat classically, was more or less the original Copenhagen interpretation (more akin to the view Bohr advocated, as Heisenberg insisted on a classical treatment of measurement for other reasons). It is consistent, coherent, and forms the basis for the tremendous success of this now nearly 100 year old theory as well as the foundations upon which the modern operational approach is based (as well as the foundations for many closely aligned interpretations, including RQM, although RQM is less aligned with Bohr’s view and based more upon that of Heisenberg).

It is also deeply unsatisfactory to large numbers of physicists because it resolves the measurement problem by forbidding it. The classical world and quantum world have two very different (but complementary) associated theoretical frameworks that must be understood as necessarily distinct. But this means that the things like measuring instruments, labs, cats, and so forth are made up of quantum systems they cannot be considered in terms of. Quantum theory seems (to just about everybody except Bohr and a few willing to follow his line of thought to this radical conclusion) to be not so much a theory of the physical world as a theory of what happens in laboratories.

Today, however, even accepting this view is problematic as (apart from anything else) we need quantum theory to describe very large systems for very sensitive experiments such as the detection of gravitational waves. It is also necessary for small, unimportant events such as the origins of the cosmos, the processes that allow for stars, the nature and values of the fundamental parameters of the universe, etc. And that’s without addressing the fact that a physical theory which is supposed to describe the systems out of which everything else is composed of ought to be able to explain how these things can exist at all (the quantum-to-classical transition, the emergence of classicality, etc.).

It is for this latter reason that Schrödinger originally proposed his infamous thought-experiment involving a cat both dead and alive due to a quantum-mechanical trigger that released or didn’t release a poison. He and Einstein saw clearly that the formalism of quantum theory might seem adequate for the description of the atomic and subatomic realm, but this was at best an illusion. His point was that one could in principle couple a sufficiently isolates system “Cat + container with poison gas + atomic trigger “ such that we must treat the atomic decay in terms of the superposition principle (i.e., both decayed and not decayed until observation) and therefore the entire system, cat and all, as existing in a superposition state.
So the many-worlds interpretation, QBism, Bohmian mechanics, RQM, etc., all involve various ways to try to explain why we get different answers when we treat measurement "classically" rather than quantum mechanically or (and very similarly) why it is that quantum systems seem to evolve according to one set of dynamics when nobody looks and a wholly different one when observed.

 

[sayak83]

Well, the first and main problem I have with the idea as you have outlined it here is that this description makes little contact with RQM and perhaps none at all with “standard” quantum mechanics. Of course, this is not a criticism of what you wrote but rather a constraint imposed upon you by the communication medium: this is an online forum composed primarily of non-physicists (let alone physicists whose work concerns or has concerned quantum foundations) and a forum post can only go so far.

But it leaves me in an awkward position. On the surface, there is great appeal in the many presentations of RQM by Rovelli and others. In a concise, summary description of some of the main features like the one you have provided, it seems even more reasonable. But such a description glosses over the nature of the problems that RQM claims to have overcome, which are after all quite nuanced, often technical, and relate the quantum mechanical framework to an operational one alongside a historical and a conceptual development spanning many decades that is where I would have to locate any issues I have with RQM (or any interpretation, for that matter).

I will try to do as you did, and summarize key issues without anything resembling formal sketches or in-depth discussion. I will supplement this with two references that are relatively non-technical I managed to locate which discuss some of the problems I have (albeit with different purposes, emphasizes, etc., in mind) as well as others.

Basically, my problem with RQM concerns whether it can be said to have achieved all that is claimed. I do not think so. But let me quickly summarize some of the key assumptions and claims about RQM in relation to the standard quantum mechanical framework. I recommend looking over these to find areas that may seem to run into conflict and think about how they might be resolved:

1) QM is complete. That is, there are no “hidden variables” (local or not) nor a point at which QM breaks down and must be replaced by a more fundamental theory. Rather, classical physics is at best an approximation of an underlying reality.

2) RQM is realist, and holds that quantum systems are part of an observer-independent reality (external to observers)

3) Observers, observation, and measurement play no special role in RQM. There is no wave-function collapse in the usual sense because there is no ontic interpretation of the wave-function and the “collapse” is interpreted epistemically, i.e., it stems from information gained from local interaction.

4) RQM is local.

5) Events and correlations between systems play the fundamental ontological role in RQM, not states.

6) States and facts are relative to the observer. Properties of physical systems are observer-dependent as are states themselves.

7) In order to keep all of the above and attempt to avoid paradoxes or similar issues, proponents of RQM maintain that systems cannot be correlated with themselves, contradictions normally attributed to differing measurement outcomes in e.g., Wigner’s friend or Schrödinger’s cat type experiments are “resolved” by asserting that different measurement outcomes of the same system cannot occur at the same time, observers cannot consider themselves or their own states quantum-mechanically (i.e., ostensibly because of the no self-correlation rule) and the contradiction between measurements of the same system by different observers can be attributed to different facts obtained via interactions ordered temporally such that one can never actually obtain contradicting information of the same system at the same time due to separation in space, time, or spacetime.


Now for my core concerns:
Mostly, all of the issues I see involve or result from the problem of interactions. This is at the core of RQM, and equally it is vital that the time of interactions be well-defined in order to make basic claims about the consistency of RQM. Yet it is not at all clear that such times, or indeed interactions at all, are well-defined. RQM, as an interpretation of QM, clearly makes use of the formalisms that do after all involve interactions we call measurements that are made at specific times and places. But one thing that enables these interactions to be well-defined is a division imposed between the systems, the measuring apparatus, the environment, and the observer (as well as perhaps other crucial components of both realized experiments and experimental schemes such as ancillary systems and so forth). By its radical egalitarian interpretation of observers, as well as the demand that properties and states make sense only in terms of interaction events, it is hard to imagine what it means for an experimentalist to even observe a “pointer” position in an experiment or better still how to make sense out of such simple properties as that of spin in systems like single electrons when the values these take (given a well-defined interaction time) are nonetheless basis dependent (this is related to the so-called “preferred basis” problem or just basis problem).

In any case, it is easy to talk about an interaction in QM and apply the formalism as has been so successfully done for decades and decades now. But this was and has been due to making certain assumptions about the nature of quantum systems, properties, states, and measurements that are problematic and that in fact RQM is claimed to have resolved. Namely, we never obtain the states predicted by the theory for almost any sort of interaction we can conceive of or realize experimentally. In particular, treating the measuring apparatus quantum-mechanically yields states never observed, making it necessary to supplement the theory with the projection postulate or collapse or something like this to ascribe definite values to measurements. RQM denies that these are objective, ontological properties or states but are rather relative to the systems involved in the interactions that are supposed to produce measurement outcomes. In particular, the interaction time at which we would have spoken of a measurement is instead defined to be a probability distribution that can be derived in terms of correlations between systems. But this in turn means that would-be measurement by the would-be observer is defined externally to the observer by another “observer” for whom the values obtained are ill-defined or denied completely and indeed for whom the “measurement” interaction cannot be said to have taken place.

In short, RQM gives a radically “democratic” approach to observers and observations that renders all systems equivalent while maintaining a complete interpretation of QM without additional supplements beyond the Born rule (or its equivalents or generalizations of these). In relativity, this is fine because the “observables” like velocity or time that are relative to reference frame are not probabilistic but governed by particular symmetries and ensure all properties and states are always and everywhere well-defined. In QM, the symmetry is given by unitary evolution, but this leads to the superpositions and entangled states we never observe without introducing an asymmetric interaction we call a “measurement” or observation. Rovelli keeps both the unitary evolution AND rejects the asymmetry, and so RQM contains blatant contradictions. These are supposed to be resolved by insisting that measurement outcomes be defined locally in terms of the interacting systems involved. But these “quantum events” require additional, external aspects in order for the required correlations to yield the distributions at a particular time. Moreover, as even Rovelli has had to admit, RQM cannot be defined locally in the manner it was initially claimed but must yield to a kind of non-classical causality that is not explained at all because the only thing we no about such explanations would involve admitting non-locality or giving up one or more additional assumptions of RQM.

As for realism, I will say simply that one can get an idea for how difficult or impossible it is for there to be any kind of coherent, ontological approach to a world with objective systems by considering that according to RQM facts are always relative to the observer, no observer is privileged, and different interactions will yield different assignment of “facts” even in the same interaction depending upon whether we consider it from the “observer” perspective making a measurement or that of the system under observation. All facts about all systems (including that these systems exist in any sense in this or that state) are relative to all other systems, so the states and properties of all systems are relative, and there is never any quantum event that we can say took place at any particular time in any particular way with any particular outcome. So what, exactly, is “real” about anything? What is the good of maintaining an interpretation like this as opposed to one which gives up at least one claim made by RQM, such the incoherent claim that RQM is a realist theory describing an objectively-existing external (quantum-mechanical) world?

Here are the two papers which go into more depth than I could here about these and other issues. I would say more than either paper and I also do not agree on the whole with either paper completely, but this is already a long post and I have only touched upon a sketch of some of the aspects of a few issues.
A quintet of quandaries: five no-go theorems for Relational Quantum Mechanics
Assessing Relational Quantum Mechanics

Sorry for the very late reply, but assessing these things take time, which is in short supply.
This is not a response but more about providing a link to a research article which tries to explain what type of realism is being described by an RQM approach. Let me know what you think.
The Bundle Theory Approach to Relational Quantum Mechanics - Foundations of Physics