Caveat lector: this essay is a little technical and assumes that you have at least a passing familiarity with the basic concepts and issues in quantum mechanics.
Most physicists aren’t all that keen on debating philosophical issues related to quantum foundations, i.e., clarifying what certain concepts in quantum mechanics (QM) actually mean.
(Some physics departments actually go so far as to discourage papers and grant proposals focused on quantum foundations.)
There are also physicists who deny that a problem really exists. But what the “shut up and calculate” crowd, as this group is often referred to, needs to come to terms with is - at the heart of QM, a key pillar of modern physics, are some thorny unresolved theoretical issues.
For example, let’s look at how quantum systems evolve in the textbook formulation of QM (often called the “Copenhagen” interpretation.) There are two incompatible rules as to how the dynamics of a quantum system evolve over time.
Rule one is embodied by the Schrödinger equation. The rule is essentially smooth, deterministic, and reversible. We hand it a quantum state, and it’ll tell us exactly what that state becomes at every future moment. This rule is, as far as we can tell, “exactly true”. We have successfully tested it to great precision.
Rule two is the measurement postulate, or as some refer to it, the so-called “collapse” postulate. When you measure a quantum system, its smoothly evolving superposition (see Rule one) abruptly jumps to a “definite outcome”, with probabilities given by the Born rule. (Explainer: each possible outcome carries a number called its “amplitude.” If we square that number, we get the probability of actually seeing that outcome. That squaring is the Born rule.) This rule is also, as far as we can tell, exactly true. Every experiment confirms it.
The trouble is - these two rules cannot “both” be fundamental. The first says superpositions never collapse. The second says they do, precisely when we “look”. Worse, rule two quietly smuggles in undefined words. What counts as a “measurement”? What counts as an “observer”? A Geiger counter? A cat? A grad student? The math of textbook quantum mechanics genuinely “does not say”. It just tells you to draw a line somewhere between the quantum system and the classical apparatus, perform the collapse there, and not ask too many questions.
This is the “measurement problem”, and it has been sitting at the foundations of QM for roughly a century. If you look closer, it actually breaks down into a few distinct puzzles that people tend to mash together:
The preferred basis problem. A “basis” is just the set of distinct alternatives you choose to describe a state in terms of “alive-or-dead”, say, or even using some bizarre “alive-plus-dead” and “alive-minus-dead” options. The unsettling part is that the math treats all these choices as “equally valid”. That is, a single superposition can be written infinitely many ways depending on “which basis you pick”. So “when we look”, why do we always see the “familiar” alternatives - definite “positions” - rather than the weird combinations (e.g., here-and-there, alive-plus-dead, etc.)? What “singles out” the classical-looking description?
The definite outcomes problem. Why do we observe “one” result rather than a smeared-out superposition of all of them?
Think of this problem as following on from the preferred basis one. Choosing a basis is like choosing a tree from the forest. Choosing a single outcome is like choosing a branch of that tree. Even if you solve the preferred basis problem re tree selection, you still have the definite outcomes selection re branch selection.
Also, please note that I’m speaking figuratively. No one is doing any “actual choosing”. Nothing in the unitary dynamics does any selecting at all. Rather, we might say the branches all coexist and the question is why we find ourselves on one specific one.
The probabilities problem. Where does the Born rule come from? Why does squaring the amplitude to get the probability work, and not some other recipe?
The objectivity problem. This one is (sort of) underrated and often not listed as a problem. Why do “different observers always agree” on the outcome? You and I can both check whether the cat is alive without destroying the result, but we always concur (meaning the cat is either alive or dead for both of us). In quantum mechanics, measurement generally disturbs the thing that is measured, so this “universal agreement” is genuinely strange. But it’s also the very thing that makes the classical world feel “solid” and “out there”.
The orthodox response (the Copenhagen tradition, in the textbook form nearly every working physicist actually operates by) is to keep the Schrödinger equation as the dynamics, and then treat measurement, with its definite classical outcomes, as the place where the theory makes contact with the world. This “works” (amazingly) well as a recipe, as it turns out. Its embarrassment is that “measurement” and “the classical apparatus” are left as undefined primitives. John Bell, in his famous broadside “Against Measurement,” wrote: a fundamental theory of reality shouldn’t have the word “measurement” in its axioms, as though the universe came pre-divided into systems and apparatus. What “physical process”, he demanded, actually constitutes a measurement?
My thesis in this essay is that Wojciech Zurek (and others), building on the theory of quantum decoherence, finally gives that question a physical answer, and that this lets us keep the orthodox framework while replacing most of its hand-wavy ambiguity with calculation. Note that this isn’t a “new theory”. Better to think of it more like a “conservative upgrade” to the existing theory we already use in modern physics.
But before I make this case, let’s look at some alternative approaches to addressing the issue. Over the years, there have been several serious attempts to sort out the measurement problem, pursued by serious physicists (and philosophers of physics).
Objective collapse theories (GRW, CSL, Penrose). Folks who subscribe to some sort of “objective collapse” approach say that wave function collapse is “physically real” and then go on to write down new dynamics that describe how it happens. The Ghirardi-Rimini-Weber model and its continuous cousin (Continuous Spontaneous Localization) add a tiny random “hit” to the Schrödinger equation that “localizes wavefunctions spontaneously”. The hit rate is set so that a single particle almost never collapses, but a macroscopic object (with its ~10^23 particles) collapses essentially instantly. Roger Penrose has a related idea where gravity triggers the collapse. In any case, within this category of methods, the collapse is built into the equations, and no observers are required. The cost? Well, we end up with a “modified” version of QM. Proponents have had to introduce new physical constants and new dynamics, leading to predictions that differ (ever so slightly) from standard QM. Which is great, actually, because that means it’s testable. The catch? Experiments (underground searches for the tiny, predicted heating, interferometry with ever-larger molecules) keep “not seeing” the effect, effectively squeezing the allowed parameter space tighter and tighter. So, this group of approaches is not dead but is definitely on the back foot.
Pilot-wave theory (de Broglie-Bohm). Also known as Bohmian Mechanics. Here you keep the wavefunction exactly as is, (meaning it evolves by the Schrödinger equation), but you add “actual particles” with “actual definite positions” at “all times”, guided by a “pilot wave”. Nothing ever collapses. The particle was always somewhere - we just didn’t know where. This solves the definite-outcomes problem (positions are always definite after all) and reproduces the Born rule given a “quantum equilibrium” assumption about initial conditions. However, this approach comes with a steep price tag. The guiding equation is blatantly, irreducibly “nonlocal” in a way that sits awkwardly with Einsteinian relativity. Moreover, extending pilot-wave theory to quantum field theory (the basis of the current Standard Model of particle physics) is notoriously thorny. There’s also now a whole second layer of “ontology” (a philosophical term for “what the theory says actually exists”) on top of the wavefunction, which now does double duty as a real physical field pushing those particles around.
Many-worlds (Everett). Upfront, this is actually the most austere approach of all. Basically, we take the Schrödinger equation, declare it the “whole story”, and toss the collapse postulate away. There’s no second rule. When we measure a superposition, we don’t collapse it. Rather, we “join” it. The apparatus, then us, then the lab, then the air molecules all get swept into one giant entangled superposition, and what looks like “collapse” is just us finding ourselves in “one branch” of it. “Every” possible outcome happens, each in its own branch.
There are some good aspects to this approach. In fact, the approach I’m going to advocate shares some machinery with the Everettians. Many-worlds is famously minimal in its modifications to the core QM theory. There are no extra particles, no extra guiding equations, no collapse dynamics, no new constants. It just takes the Schrödinger equation we already use and refuses to add anything to it. Note that it’s this “refusal-to-modify instinct” that I’m sympathetic to. Also, I’d like to point out that modern Everett advocates (David Wallace, Sean Carroll, etc.) lean heavily on decoherence to make their approach work. So, this isn’t a “rival” camp to what I’m proposing below, so much as a sort of neighbor.
The bigger problem is the ontological consequences on the backend of the Everrett approach. To get that lean formalism, (full-blown) many-worlds asks you to believe in an unfathomable, ever-branching profusion of (equally real?) parallel universes, all of which are forever inaccessible to us. So, the formalism itself is austere. But the reality it describes is probably the most extravagant in the history of physics. There are also some technical issues here. The Everett approach, on its own, doesn’t (explicitly) tell us “what” a branch is. The Schrödinger equation gives us one enormous evolving wavefunction (the so-called “wavefunction of the universe”). It doesn’t come pre-labeled with saying “world A here, world B there.” Nor does it (explicitly) hand us the Born probabilities. If “all” outcomes happen, in what sense is one outcome “more probable”?
Hence, my own preference, which is more conservative than any of these approaches listed above. I would prefer an approach that does not add particle locations or guiding equations (Bohm), or add new collapse dynamics (GRW), or commit to a literal multiverse (Everett). I want to keep the textbook QM framework that already works and simply “fill in the gaps” with real, established physics.
Before getting into the core material of this essay, a quick word on a genuine puzzle. If this approach is as promising as I’m claiming, why does it get a fraction of the popular attention lavished on many-worlds and pilot waves? You can buy a shelf of trade books on the multiverse. There’s no equivalent for the Zurek program. Why is that?
I think there are a few reasons - these are based on my observations of the field over the past couple of decades. Note that none of these reasons are “because it’s wrong.”
The Zurek approach isn’t a “flashy” interpretation, so it makes for a worse story. Many-worlds gives us infinite parallel universes. Bohm gives us secret guiding waves. These are “quotable”. Influencers on social media can wax poetic about (or heavily criticize) those approaches. The Zurek approach, on the other hand, gives us... a careful account of why the classical world looks classical. The selling point is mostly “I’ll tell you less about ultimate reality, but I’ll back it all up with calculations.” Kind of a hard pitch for a popular-science story.
A good portion of the Zurek program got "absorbed into mainstream physics so thoroughly it stopped feeling like a position. Decoherence is now a standard part of the QM toolkit. It shows up in quantum computing error analysis, in experimental work, in textbooks, etc. The very success that makes the approach credible also makes it feel like “engineering” rather than “philosophy.” Besides, the QM foundations-and-philosophy media enterprise runs on unresolved drama (or at least it seems that way to me). Many-worlds and Bohm stay newsworthy precisely “because” they remain so contested.
The genuinely bold parts of the Zurek program are either too technical or too new. Envariance (deriving the Born rule from symmetry) is mired in a technical circularity debate that’s very hard to popularize. Quantum Darwinism is recent and was only experimentally confirmed in the lab in 2025. There’s been no breakout popular-science moment or book, and no charismatic popularizer carrying the banner the way Sean Carroll or David Deutsch does for Everett or Tim Maudlin for Bohmian Mechanics.
And advocates of the Zurek program have positioned it as interpretation-neutral infrastructure, which is intellectually honest but… politically weak. A program that says “this is just what quantum mechanics does, no matter your metaphysics” doesn’t recruit any evangelists to the cause. The interpretations that get all the press are the ones with partisans who show up to fight the good fight.
So one could argue that the obscurity is sociological, not evidential. Anyhow, with that out of the way, here’s the actual physics.
Here’s what makes Zurek’s program a “conservative” fix to orthodox QM rather than a radical one.
Bohm adds particles and guiding equations. GRW adds collapse dynamics. Everett adds many worlds. The Zurek approach is - you don’t need to add anything. You just need to take seriously a piece of physics that was sitting in plain sight the whole time, namely “the environment”.
Every real quantum system is relentlessly, continuously entangling with its surroundings. That is, the system is getting its state tangled up with the state of everything it touches. You can no longer fully describe the one without the other. A dust grain in sunlight is struck by something like 10^11 photons a second. A “measuring device” is thus just a big system soaked in an even bigger environment. How does textbook QM deal with this? Well, it sort of performs a sleight of hand in the original formulation - it treats the measured system as “isolated”, then collapses it by (axiomatic) fiat. But, in reality, nothing is usually isolated! The environment is “always in the room”. Zurek’s approach, pursued since the early 1980s and built on the broader theory of decoherence developed since the 1970s (Zeh, Joos, and others), centers around: what happens if you stop ignoring the environment?
The answer, it turns out, is the sort of physical mechanism John Bell was asking for when he wanted a real answer to the measurement problem. The classical apparatus in orthodox QM that Bohr had to “assume” and the definite-outcome world the Copenhagen framework “presupposes” - both of these emerge from ordinary “unitary” QM once you apply it to system-plus-environment (unitary is just the adjective for the smooth, no-collapse Schrödinger evolution of Rule one I mentioned at the beginning of the essay).
Let discuss the core concepts involved.
Let’s start with the preferred basis puzzle. Why "position and not some bizarre superposition basis?
The answer is that the environment is constantly “monitoring” certain observables and not others. The interaction between a system and its surroundings has a particular form. It usually depends on position, because forces depend on where things are. That interaction picks out a special set of states - namely, the ones that “don’t get smeared by being monitored”.
Most superpositions, when they entangle with 10^11 incoming photons, instantly leak their “which-state” information “into” those photons (which-state = identity or exact configuration of a quantum system). The delicate “phase relationships” that make quantum superposition a superposition get scattered to the winds. But a few special states are robust. If the system is in one of those special states, the environment scatters off it without disturbing it much, and the state “survives”. Zurek calls these survivors “pointer states” (as in, the states a measuring pointer can actually rest in).
Zurek named the process that selects them einselection (environment-induced superselection). The environment, simply by interacting, acts like a sort of sieve. Fragile superpositions are destroyed almost instantly. Robust pointer states persist. The operational tool for “finding the survivors” is what Zurek terms the “predictability sieve.” You basically run every candidate state forward while coupled to the environment, measure how fast each one degrades (that is, how much entropy it generates), and rank them. The ones that stay predictable (that resist the environmental churn) are your pointer states. For realistic systems, these come out to be the roughly-localized, classical-looking states. So the position basis “wins” not because we put it in by hand, but because that’s what the system-environment interaction selects.
This means that the classical/quantum cut that Bohr drew pragmatically (the dividing line he placed by intuition between apparatus and system) turns out to be a “real physical feature” you can derive. This dissolves the preferred basis problem! And it also answers Bell’s “what counts as classical?” question. Critically, the quantumness itself decays exponentially fast.
(Bit of technical background: physicists track a quantum state as a grid of numbers. Down the diagonal sit the ordinary probabilities: 70% chance of “here,” 30% chance of “there,” the kind of thing a classical coin could produce. The off-diagonal entries are the extra ingredient. They record how the possibilities relate to each other, and they’re what makes a real superposition behave differently from a classical “we just don’t know yet.” When those off-diagonal numbers fade to zero, the quantum state becomes indistinguishable from a shuffled deck, and that fading is what we call decoherence.)
Thus, decoherence is the process of those off-diagonal numbers “shrinking to zero” leaving behind a grid that looks exactly like plain classical odds. And it happens really fast. For a macroscopic object, decoherence times are something like 10^-20 seconds or even faster (see Table 1 on page 11 here). This is vastly quicker than any other timescale in the problem. This is why we never see a superposed cat (dead “and” alive). The superposition isn’t forbidden per se. Rather, it’s just annihilated faster than anything could ever notice it.
Decoherence tells us why certain states survive. But it doesn’t explain why “many independent observers all agree” on the single outcome that emerged.
This is Zurek’s second big idea. Why do we treat the position of the moon as an objective fact? Because the information about it is “redundant”. The moon scatters sunlight in every direction, so its position is imprinted on countless independent packets of photons streaming through space. Any one of us can intercept “some” of those photons and read off the same answer, without anyone’s reading disturbing the moon or interfering with anyone else’s. Thus, the information isn’t held in one fragile place and is broadcast in a zillion copies.
Zurek calls this quantum Darwinism: the environment doesn’t (just) destroy superpositions, it “selects and proliferates” records of the pointer states. The “fittest” states (that is, the einselected ones) are precisely those that can spawn many redundant copies of themselves throughout the environment. Thus, a fact is “objective” exactly when it’s recorded so many times, in so many independently accessible fragments, that everyone who checks gets the same answer.
This essentially reframes the role the environment plays in the framework. In old-school decoherence, the environment is a “sink”, a place where quantum coherence goes to die (as we discussed in the prior section). In quantum Darwinism, it is also a “communication channel.” A witness, so to speak. The environment is the medium through which the classical world “advertises itself” to all of us at once.
Now, what about those probabilities governed by the Born rule? Here Zurek offers his most contested contribution, envariance (environment-assisted invariance). The idea is - instead of postulating the Born rule, let’s derive it (purely) from the “symmetries of entangled states.” Roughly speaking, when a system is maximally entangled with its environment, certain symmetry operations on the system can be “perfectly undone” by operations on the environment alone. (Read that last line twice if necessary). Zurek argues this forces "equal-amplitude branches to carry equal probability”, and then bootstraps the general case for envariance from there.
I will acknowledge that this is the part of the Zurek program where experts genuinely disagree. Critics of the envariance solution to Born probabilities argue that the derivation implicitly “assumes some notion of probability” to get started, rendering it circular. Now, defenders do argue that it doesn’t. I think it’s best to say that this part of the theory should be seen primarily as suggestive. But let’s also note that “even if envariance is not right”, the other pillars (decoherence, einselection, quantum Darwinism) stand on their own.
Now let’s talk about some of the most recent developments related to this approach.
Decoherence tells us the classical world emerges as a set of stable, redundantly-recorded, non-interfering components - the so-called “branches.” But there’s a deep question that physicist C. Jess Riedel asks, most recently in a 2025 piece pointedly titled “Wavefunction branches demand a definition!”
(Technical note: a “branch” here is a term in a decomposition of the global state vector relative to a chosen factorization of Hilbert space into system, apparatus, and environment.)
Riedel’s challenge goes like this: the decoherence story usually starts by “assuming” a split of the world into “system” and “environment.” It also assumes we know which variables are the macroscopic, classical-looking ones. But imagine you’re handed the raw wavefunction of every atom, with no labels. Meaning, nobody tells you which atoms are “the coffee cup” and which are “the air.” Could you write a formal procedure, an actual algorithm, that takes the bare wavefunction and spits out the branches as an output? That is, we are not going to rely on just human intuition about what’s “macroscopic.” We won’t use some sort of pre-chosen system-environment split. We want a procedurally calculated “state in, branch structure out”.
Per Riedel’s challenge, if we can actually do that, we will have something remarkable. We replace the vague, observer-dependent notion of “measurement” with a precise, universal, derivable definition of what the distinct outcomes of any quantum process actually are. And this would work for for the photodiodes in our labs, the whole universe, the aftermath of cosmic inflation, etc. Riedel’s point is that the somewhat abstract problem then becomes a concrete, well-posed math problem.
Now here’s the fun part - there are people making real attempts to actually do this. Recent proposals try to define a branch by how “hard” it is to undo. (Riedel reviews the work of Taylor & McCulloch and Weingarten). The intuition is simple (and elegant). Two parts of the wavefunction count as separate branches when they’re easy to “tell apart” but very hard to “stitch back together” into a superposition. That asymmetry - easy to distinguish, near-impossible to recombine - is then the fingerprint of an irreversible, classical “fact”. (This kind of works the same way it’s easy to scramble an egg and effectively impossible to unscramble it.) One proposal even ties the stability of branches to the general tendency of this kind of “un-mixing difficulty” to keep growing over time, linking the appearance of collapse directly to physical irreversibility.
These attempts aren’t finished, of course. Riedel is candid about the gaps. For example, one proposal offers unique branches but may force them too sharply (Weingarten). Another is better motivated but doesn’t yet provide a unique answer (Taylor & McCulloch). Neither has a clean relativistic version. In any case, progress continues on this particular thread of the program.
So, where does this leave us? Well, the orthodox QM framework pretty much left “measurement” as a black box. Decoherence pried open that box and found an underlying explanatory mechanism. Now, the frontier work is making that mechanism more precise, such that it could be stated as an algorithm.
At the end of the day, we need evidence to back up theories. And this is where the Zurek approach distinguishes itself from most of the other interpretations. Large chunks of it have been experimentally confirmed. Meaning - these chunks are now established physics, not just a stance.
Decoherence has been extensively verified. Since the 1990s, experiments have observed decoherence happening “in slow motion” by engineering systems in which it’s slow enough to time. Serge Haroche’s group (who won the Nobel Prize in 2012) famously watched superpositions of light decohere, photon by photon, in a cavity, measuring the decay of the off-diagonal terms and confirming that larger superpositions decohere faster. Exactly as the theory says! Matter-wave interferometry with large molecules (fullerenes etc.) shows interference fringes vanishing as we allow the molecules to interact more with their environment (emit thermal photons, collide with gas). Decoherence has become textbook, lab-confirmed, engineering-relevant physics. The entire field of quantum error correction exists because decoherence is real and quantitatively well understood. The 2025 Nobel Prize in Physics went to Clarke, Devoret, and Martinis for showing that a superconducting circuit large enough to see can tunnel and can occupy discrete energy levels, behaving as a single quantum object. The same platform became the superconducting qubit, where coherence and its loss are now measured as a matter of routine engineering.
Quantum Darwinism has now been directly observed. This is another recent development. The redundancy-and-proliferation story got its first experimental support around 2019, in photonic systems and nitrogen-vacancy centers in diamond. These showed the characteristic signature: information about a system saturating once an observer captures a small fragment of its environment, with extra fragments just repeating what you already know. That plateau is the smoking gun of redundant encoding. (Something to note: several of these experiments are best described as “quantum simulators” of the effect, i.e. engineered systems that realize the Darwinism dynamics rather than a naturally decohering object caught in the act, a distinction worth noting.)
The most comprehensive demonstration to date came in 2025, when Zhu, Salice, Touil and collaborators used a superconducting quantum processor to probe the effect. Where the earlier experiments had only detected information-theoretic signatures in narrow settings, this team was able to map the branching structure of the global state directly. They were able to watch the mutual information saturate as predicted and confirm the geometric picture that underpins the framework. The system is still an engineered one, of course, rather than an everyday object decohering in the wild, but it is the closest anyone has come to seeing the whole mechanism at once.
The implications are fascinating - as long as observers eavesdrop on a suitably large fragment, they always agree on their conclusion if they’re in the same branch, illustrating how classical reality emerges from a structured quantum universe!
Now, this does not prove that Zurek’s view is the “uniquely” correct one. In fact, I’m not sure that experiment could really “settle” a question of interpretation. What it does prove is that the “mechanisms” the Zurek program is built from (einselection, redundant proliferation, branch structure, etc.) are “real, physical, and measurable”. This makes the program stand out when we look at the alternatives. Bohmian particles? Never observed (and by construction unobservable?). GRW collapses? Searched for, never seen. Everett’s other worlds? Inaccessible in principle. But decoherence and quantum Darwinism? We have empirical data that confirms important aspects of the framework. Since the goal here was a “conservative fix”, the Zurek framework does that by (primarily) leaning on mainstream confirmed physics rather than introducing new (speculative) metaphysics.
Let’s also look at the some of the main criticisms of the Zurek program. I’m going to group these into two buckets - what the Everettians would say (Sean Carroll, David Deutsch, David Wallace, etc.) and what realists like Tim Maudlin might say.
The Everettian objection: “You’re just using the Everett framework and refusing to admit it.” (I’m paraphrasing here, but I think that’s a fair summary of one of their main points.) A committed Everettian would say this whole “conservative upgrade” framing is a dodge. Look at what we’ve to actually commit to in the Zurek framework. We have “the Schrödinger equation is complete, it always evolves unitarily, nothing collapses”. Well, those are the defining characteristics of the Everett program! If the global wavefunction never collapses and the other branches genuinely contain observers seeing definite outcomes (which Zurek and decoherence says they do), then those branches are '“as real” as the many-worlds ones. Declining to call them “real”, from the Everettians’ perspective, shouldn’t be seen as “conservative”, but rather a refusal to fully outline the (Zurek framework) ontology while helping oneself to all of Everett’s machinery.
My response: Ouch. Well, the Everettians are right that the Zurek program maintains a good chunk of the Everett program’s upfront framing (no collapse, the wavefunction as the complete story, etc.). I think they are also right that this puts the Zurek program in the “same broad family” as Everett. But I think the difference comes down to what the Zurek program is “obligated” to assert. The Everett folks think taking the formalism seriously “forces” us to grant “full reality to every branch”. I think that’s an extra metaphysical step, not a logical entailment. The physics (decoherence, einselection, redundancy) is identical whether or not you make that step, because no experiment can reach the other branches to confirm or deny their “reality.” So we take the conservative path: assert what can be empirically verified, stay quiet about what can’t. But yes, I’ll concede the main framing point: this Zurek view is indeed sort of “Everett-adjacent”.
The realist’s (e.g. Maudlin) objection: “Decoherence does not produce definite outcomes, and you’ve admitted as much.” Maudlin has actually said something exactly that like that. And we’d have to concede that he’s right. But, let’s take a look at the difference between two situations that look identical on paper. First - let’s look at a genuine “mixture” scenario. That is - the cat really is either alive or dead. Well, one of them for sure, we just don’t yet know which (kind of like a coin already flipped but still covered by your hand). Second, let’s take a look at “superposition” scenario. That is - the cat is in the strange both-at-once state that has no classical counterpart. What decoherence does is make a superposition “look”, on paper, exactly like an ordinary “we just don’t know which” mixture, by driving those off-diagonal numbers to zero. Now the rebuttal to this would likely be that “looking identical to” a real mixture is not the same as “being one”. The realists would say - the cat is still “formally” in the both-at-once state. Decoherence has merely guaranteed that the two possibilities can no longer interfere with each other, not that one of them has actually become the case. Maudlin would likely invoke his famous argument here (which I like quite a bit) that the measurement problem just “is” the inconsistency of three claims: 1) the wavefunction is the complete story, 2) it always evolves smoothly with no collapse (unitarily), and 3) measurements have single determinate outcomes. Maudlin says you always have to deny one. So any sort of “conservative upgrade” that touches none of those still has the contradiction.
My response: Got to concede the central point to Maudlin here. Decoherence does “not”, by itself, deliver a single determinate outcome. But let me state the actual claim precisely - decoherence and quantum Darwinism don’t dissolve the measurement problem, but they “maximally narrow” it. Of the four puzzles I opened with, three (preferred basis, objectivity, and, more tentatively, probabilities) get genuine physical answers from the Zurek program. What’s left is the bare determinacy question, why “this” particular outcome is experienced rather than the superposition, and we have to concede that this remnant is basically Maudlin’s “deny one of the three.” But I will also assert that the Zurek program shrinks the trilemma down to a single, sharply isolated, “empirically inert” question, while explaining everything around it with confirmed physics. Maudlin will then say probably “inert is not solved,” and he’s right.
Riedel concedes something along those lines too: even a full branch definition “would hardly eliminate all the mystery of the measurement problem.” But the point stands that the Zurek program has made more progress than any other approach.
(A note about claims being “empirically inert”: Imagine a claim that a tiny, invisible gremlin is hiding in your room. If this gremlin makes no noise, leaves no footprints, cannot be felt, and interacts with absolutely nothing, it is “empirically inert”. Because you can never run a test to prove the gremlin is not there, the idea is untestable.)
Maudlin’s likely follow-up: “Objectivity is answering a question nobody’s measurement problem was asking.” He’d also likely say: quantum Darwinism is a nice result about “information structure”, but it’s orthogonal (irrelevant) to the real problem. Observers agreeing with each other presupposes there are outcomes for them to agree on, and that’s the very thing in dispute.
My response: I think the criticism undersells what’s been accomplished. Sure, objectivity presupposes definiteness rather than producing it. But the objectivity problem is a “legitimate explanatory target” in its own right. I’d also like to point out that Bohmian Mechanics and GRW would also have to explain why records across the world agree, and the answer isn’t clear at all! Quantum Darwinism is a genuine, now experimentally confirmed contribution to “that” question. (So this was more like a successful side-quest.)
The realists’ parting shot: “The fact that you still can’t define a ‘branch’ after sixty years is a symptom, not a frontier.” The realists would read the Riedel review material very differently. If the approach can’t even non-arbitrarily say “what the outcomes” are without an after-the-fact algorithm to dig them out of the universal wavefunction, that, they would say, is evidence the framework tries to carve structure that isn’t fundamentally there. Whereas some alternatives (a precise microphysics, Bohmian corpuscles, GRW mass-density, etc.) would tell you what exists up front, no extraction procedure required.
My response: If you already believe the fundamental furniture of the world should be specified up front, then yes, needing an algorithm to “locate” the branches looks like an admission of failure. But if you think macroscopic structure is “emergent” (like temperature, fluid flow, etc.), then needing a principled procedure to extract that structure from the microphysics is completely normal. We do that for every other higher-level science. Nobody calls thermodynamics ill-founded for doing that. So whether the branch-definition problem reads as “symptom” or “promising frontier” depends on a prior commitment about emergence that the physics alone won’t settle. I’m going to go ahead and say that this is a promising frontier. Because, the tools now being used to tackle the problem (quantum circuit complexity, approximate error correction, etc.) genuinely did not exist when the measurement problem was originally posed. So “we have a precisely stated problem and powerful new tools aimed at it” is a very different situation from “we have an incoherent muddle.”
Okay. Let’s look at some genuinely open questions that the framework is still addressing. I’ll split this up into what I think is merely philosophical vs. actually physics.
After einselection picks the basis, decoherence kills the off-diagonals, quantum Darwinism makes the outcome objective, and envariance hands you the probability weights, one thing remains: the global state is still, formally, a superposition of all the branches. The residue, as we discussed in the prior section, is the bare question of why one outcome is 'experienced as actual.
Now you might ask - why I am being dismissive and filing this under under philosophy rather than physics? Well, the physics is “identical” no matter how we answer it. Every branch contains the same einselected, redundantly-recorded, Born-weighted structure whether or not we decide the other branches are “real.” No experiment can distinguish “the other branches exist” from “they don’t,” because any experiment we run happens “within” a branch and sees the same thing either way. Embrace the branches (The Everett path) and the question dissolves into “why am I me?”. Reject the branches and we (have to) posit a brute fact about which branch is actual. Either way, more decoherence calculations will not settle it. Gonna call one this philosophy and leave it to taste.
Now the part where there’s unfinished physics.
1. The branch-definition problem. As discussed, we don’t yet have a rigorous, universal, observer-free definition of a branch derivable from the bare wavefunction. We have promising complexity-based proposals with real gaps (the Riedel summary I discussed previously outlines these). None are yet simultaneously unique, well-motivated, and relativistic. How it might get resolved: we should pay attention to the quantum-complexity approaches. If someone shows the “easy to distinguish, hard to interfere” criterion gives a unique decomposition that (a) forms a proper tree in time, (b) recovers the known decoherence of ordinary hydrodynamic variables, and (c) survives a relativistic limit, that would essentially '“complete” the program, converting “measurement” from a primitive into a theorem about wavefunction structure.
(Re hydrodynamic variables, what I mean is - a natural test for the program’s ambitions is whether it can account for the classicality of hydrodynamic variables. Gell-Mann and Hartle argue that the classical variables of a many-body system are the local densities of conserved quantities, energy, momentum, particle number, integrated over small volumes. Zurek’s framework locates classicality in coupling to an environment instead. Both approaches work. Both are plausible. Whether they are two faces of one mechanism or two distinct routes remains open, and settling it would tell us how general einselection really is.)
2. Pointer states in the hard regimes. Einselection works beautifully when the system-environment coupling dominates. But there are regimes where “which states get selected” becomes subtle and timescale-dependent. Examples of these scenarios include when the system’s own internal dynamics compete with environmental monitoring, or when it starts in a mixed state. Recent work, notably by Sebastian Deffner’s group, has been turning the loose “pointer states usually exist” into rigorous, checkable conditions, by providing precise algebraic criteria for exactly when objectivity emerges for a broad class of two-body interactions. How it might get resolved: this looks like a tractable classification program, that might be plausibly wrapped up (in the sense of “we know the necessary and sufficient conditions”) within a few years.
3. Making objectivity fully rigorous. Zurek’s original quantum Darwinism used a quantity called quantum mutual information as its measure of “how much the environment knows” about the system. (That’s just a QM analog of Shannon mutual information.) J. K. Korbicz and collaborators pointed out, constructively, that mutual information is actually “too weak” to guarantee genuine non-disturbing objectivity on its own. They developed a sharper notion called “Spectrum Broadcast Structure” (SBS) that demands the global state literally factor into perfectly distinguishable records across observers. Effectively, when a state has SBS form, objectivity becomes a “theorem”. This would be a strong self-correction within the research program. Status: nearly resolved on the theory side; the remaining work is showing SBS emerges generically rather than in hand-picked models.
So where does this leave us?
What Zurek offers (augmented by Riedel, Korbicz, Deffner, and a growing community) is, I think, the most attractive resolution on the table to address some of the issues within QM we laid out at the start. It’s attractive precisely because of “how little it asks us to add” to conventional QM. If we want to keep the textbook QM framework that every physicist already uses, whose primary embarrassment was leaving “measurement” undefined, and patch that hole with physics we’ve confirmed in the lab - then the Zurek approach is the best way forward today.
Decoherence is real and measured. Einselection is calculable. Quantum Darwinism was directly observed in 2025, with multiple observers provably driven to agreement.
Final scorecard: the preferred-basis problem is “solved” (einselection). The objectivity problem is “solved”, and now experimentally demonstrated (quantum Darwinism, sharpened by SBS). The probability problem is “plausibly addressed but contested” (envariance). The definite-outcomes problem splits cleanly into a derivable structural part (branches, with real open physics) and an empirically inert metaphysical residue (which-branch-is-mine, where people are free to choose their poison).
Now, the Zurek framework is not a complete theory (yet). But this is just the incompleteness of competing, irreconcilable philosophical approaches. And more importantly, it’s just the ordinary incompleteness of an active research program with well-posed problems and scientists making measurable progress on them.
When the dust finally settles on the measurement problem, I’d bet that the resolution looks “less” like a brand-new collapse mechanism or a hidden layer of particles, and “more” like the orthodox QM framework augmented by the Zurek program.
So, Eureka is too strong. But Zureka!? I think we’re allowed that one.
Thanks for reading! Subscribe for free to receive new posts and support my work.