{"id":"81cfdd74-76c1-453c-a104-80004895e8fc","arxiv_id":"2608.02635","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"Under a Copenhagen variant with observer-independent collapse in isolated labs, the extended Wigner's Friend 'ok, ok' probability becomes 1/4 instead of 1/12, which the authors claim resolves the paradox.","lead":"A measurement is assumed to always produce one real result, even inside an isolated laboratory. Applying this to the extended Wigner's Friend paradox changes a key probability from 1/12 to 1/4 and is claimed to dissolve the contradiction.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Collapse to |ok>_L leaves friend's memory in a superposition, so Table 2's definite-knowledge claim fails and the paradox resolution collapses.","rationale":"The paper's central claim is that adding collapse as a real process inside isolated labs and distinguishing state from knowledge resolves extended Wigner's Friend. The probability calculation (13) is a straightforward consequence of applying the collapse axiom to the coin and spin measurements: the collapsed branches (12a-c) replace the coherent branch (14), yielding p_ok,ok=1/4 rather than 1/12. That part is internally consistent. The problem is at the final W measurement. If collapse is a projection onto the eigenspace of the observed observable, then W's complementary-basis measurement of the lab projects the lab onto |ok>_L or |fail>_L, which are superpositions of the two pointer/memory states. The internal observer's memory is therefore not a definite eigenstate after the measurement. Table 2 step 3 nevertheless asserts that F has perfect knowledge of the lab state ('probability 1 for the state in the corresponding lab'); this contradicts the very collapse rule used to obtain it. The resolution depends on being able to tell a story in which each agent's knowledge is well-defined at all times, but here the friend's knowledge is undefined because her brain is in a superposition. This is not a quibble about consciousness; it follows from the paper's own assumption 2, which treats the friend's memory as a quantum system (5a,b). The weakening of criterion (Q) to (Q') is also a red flag, as it makes the no-go theorem vacuous, but the complementarity issue is the decisive technical flaw. Thus the central claim is unsupported; the verdict should remain REJECT.","tokens_in":13020,"tokens_out":7081,"duration_ms":72399,"concrete_test":"Analytical check: start from state (12b), |t>_R ⊗ |−1/2>_L, apply W's measurement on L with basis (6b), and condition on outcome 'ok'. Using the paper's collapse rule, the post-measurement state is |t>_R ⊗ (|−1/2>_L − |+1/2>_L)/√2. Now decompose |±1/2>_L via (5a,b); the reduced state of F's memory is (|'ψ_S=↓'>_F − |'ψ_S=↑'>_F)/√2, an off-diagonal superposition, not a definite eigenstate. If this derivation is correct, Table 2 step 3's K_F=1 is contradicted. (Alternatively, a 3-qubit simulation confirming this suffices.)","verdict_should_be":"UNCHANGED","load_bearing_attack":"Assumption 2 (Section 2) makes collapse a real physical process inside isolated labs. Applied to the final step, W's measurement of lab L in the |ok>/|fail> basis (eq. 6b) projects the lab onto, e.g., |ok>_L = (|−1/2>_L − |+1/2>_L)/√2. This is a coherent superposition of the two pointer/memory states (5a,b). Consequently the friend's record and conscious memory are left in the superposition (|'ψ_S=↓'>_F − |'ψ_S=↑'>_F)/√2, not in a definite eigenstate. Table 2 step 3 nonetheless assigns F probability 1 for the state in her lab ('Knowledge is perfect about the state in the corresponding lab'). This is inconsistent: the collapse postulate does not provide a definite internal register when the measured observable is complementary to the pointer basis. The paper's resolution requires that at all times every agent either knows or can assign probabilities to a definite true state; but after W's measurement the friend's cognitive state itself is indefinite. Thus the central claim 'we can resolve any inconsistencies' fails on the paper's own axioms.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a modified Copenhagen interpretation of quantum mechanics based on three assumptions: (i) a universal, observer-independent quantum state; (ii) wave-function collapse as a real physical process that is part of quantum time evolution, including inside isolated systems; and (iii) a sharp distinction between the objective quantum state and the epistemic state of conscious observers. It applies this framework to Wigner's Friend and to the extended Wigner's Friend setup of Frauchiger and Renner. The central computation is Eq. (13), which gives p_ok,ok = 1/4 instead of Frauchiger-Renner's 1/12, and the paper claims that with these assumptions all inconsistencies are resolved. The paper explicitly acknowledges that the measurement problem is not solved and that the collapse postulate is an ill-defined placeholder.","tokens_in":13361,"tokens_out":8122,"duration_ms":91859,"significance":"If the proposed interpretation were internally consistent, the paper would be a useful contribution: it offers a concrete, falsifiable probability difference (Eq. 13, p_ok,ok = 1/4 vs 1/12), engages seriously with collapse models and the Frauchiger-Renner theorem, and is transparent about its limitations. However, the central consistency claim fails at the final step of the extended Wigner's Friend setup. The paper's own collapse rule, applied to the external measurement of the lab in the |ok>/|fail> basis, leaves the friend's memory in a superposition. Table 2 nevertheless ascribes definite internal knowledge to the friend. This is not a minor presentation issue; it is the load-bearing step of the claimed resolution. The paper also weakens Frauchiger-Renner's criterion (Q) to (Q'), which reduces the force of the 'resolution' unless the modified axioms are independently well motivated.","major_comments":[{"comment":"The collapse of lab L onto |ok>_L = (|-1/2>_L - |+1/2>_L)/√2 is a projection onto a coherent superposition of the two pointer/memory states (5a) and (5b). Under Assumption 2, this is a real physical collapse: the friend's measurement device and the friend's memory are left in the superposition (|'-1/2'>_F - |'+1/2'>_F)/√2. Table 2 nonetheless reports the internal registered result as 'ok' and states that knowledge is perfect about the state in the corresponding lab. That is inconsistent with the collapse postulate: F does not occupy a definite belief state after W's measurement, so the epistemic probabilities in Table 2 are undefined. This invalidates the paper's claim that the modified assumptions resolve all inconsistencies at all times.","section":"§4, Eq. (6b) and Table 2, step 3"},{"comment":"The paper replaces Frauchiger and Renner's criterion (Q) with a weaker criterion (Q'), requiring consistency of 'a suggested version of quantum mechanics' rather than a pre-defined version. The paper itself concedes that 'It is trivial that modified assumptions (and modified criteria) can lead to different conclusions.' The original Frauchiger-Renner theorem is about the consistency of a fixed, given quantum theory. Demonstrating that a different theory with a modified criterion avoids the contradiction does not resolve the original paradox unless the modified axioms are independently justified at the required level of precision. The collapse postulate is explicitly acknowledged to be not well defined, so the justification is currently incomplete.","section":"§5, criterion (Q')"},{"comment":"Assumption 2 makes collapse a real process 'also within isolated systems,' but it does not specify when a measurement process occurs, what counts as a macroscopic registration, or what selects the measurement basis and the timing of collapse. The paper acknowledges this ('As such a measurement process is not well-defined, this is not satisfying'), but the central probability shift in Eq. (13) depends on collapse occurring at the specific internal steps and not at the coherent-branch level of Frauchiger-Renner. Without a precise criterion, the universal consistency claim cannot be uniquely evaluated. This is a foundational gap, though the more immediate technical inconsistency in Table 2 step 3 is the decisive problem.","section":"§2, Assumption 2"}],"minor_comments":[{"comment":"Typographical errors: 'underlay' should be 'undergo'; 'ocurred' should be 'occurred'.","section":"§2"},{"comment":"Typographical errors: 'There is a an intensive discussion' should be 'There is an intensive discussion'; 'intergral' should be 'integral'.","section":"§5"},{"comment":"The column headers for 'registered result' are confusing because 'ok'/'fail' in step 3 are results of the external measurements, not of the internal devices D and D. The footnote clarifies, but the table's layout makes it easy to misread the content of the cells.","section":"Table 2"},{"comment":"The notation uses the same subscript L for the two labs (⟨ok|_L ⊗ ⟨ok|_L), although earlier in the paper the two labs are distinguished as L and L (or L and ar L). This makes Eq. (13) unnecessarily ambiguous.","section":"Eq. (13)"}],"recommendation":"reject","confidential_remarks":"The paper contains a transparent probability calculation and a clear statement of its assumptions, but the central consistency claim fails because the external collapse to |ok>_L leaves the friend's memory in a superposition, contradicting Table 2. This is not a matter of presentation: it is the load-bearing step of the resolution. The weakening of (Q) to (Q') further limits the interest of the result. I do not see a straightforward revision within the current framework that would preserve the paper's central claim; a substantial reconceptualization of the role of observers after external measurements would be needed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe paper is worth reading for the calculation: they get p(ok,ok)=1/4 under their collapse-in-isolated-systems Copenhagen, whereas Frauchiger–Renner get 1/12. The derivation is clear and the difference is exactly traced to replacing the coherent branch (eq. 14) with collapsed branches (12b,c). That is a genuine, testable-in-principle consequence of their axioms. The authors are also honest about the unresolved measurement problem and they don't hide the fact that they have weakened FR's criterion (Q) to (Q').\n\nThe problem is that the resolution fails at the final step. When W measures the whole lab in the |ok>/|fail> basis, their own collapse postulate projects the lab onto, say, |ok>_L = (|−1/2>_L − |+1/2>_L)/√2. That is a coherent superposition of two pointer/memory states. So the friend's memory and recorded outcome are not definite. Table 2, step 3, still assigns F perfect knowledge of 'the state in the corresponding lab.' That is inconsistent with the collapse rule they rely on. The collapse to an eigenstate of W's observable is not a collapse to an eigenstate of the friend's pointer observable, and their postulate doesn't magically make the friend's memory definite. This is the same issue that plagues naive collapse pictures when external measurements are in a complementary basis. The stress-test note gets this exactly right.\n\nThe weaker criterion (Q') is a second soft spot: they change the no-go theorem rather than satisfying it. That's not fatal if the new axioms are independently motivated, but it's a significant concession. The mechanistic definition of collapse (threshold, basis, timing) is also admittedly missing, and they explicitly punt on it.\n\nSo: the calculation is new and clean, the writing is transparent, and the limitations are stated. But the central claim—'we can resolve any inconsistencies'—is not supported. The paper deserves a serious referee because the calculation and the interpretation move are interesting and the mistake is instructive, but it should not be published as a resolution. A good referee might suggest testing the p=1/4 prediction in a small quantum simulation, but that won't fix the conceptual inconsistency.\n\nFor whom is this? People working on Wigner's friend and collapse models may want to cite it as an example of a collapse-based attempt that fails at the complementary-measurement step. I wouldn't cite it in my own work. I would send it to peer review rather than desk reject it, though the likely outcome is a clear rejection or a major rewrite that fixes the final step.\n\nRecommendation: engage with it enough to point out the step-3 error; don't let it through as a resolution.","headline":"A clean probability calculation and a clear limitation statement, but the resolution fails at step 3: W's complementary-basis collapse leaves the friend's memory in a superposition, so Table 2's definite-knowledge claim doesn't follow.","tokens_in":13748,"tokens_out":3187,"would_cite":false,"duration_ms":32134,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A modified Copenhagen interpretation—with wave-function collapse as real time evolution inside isolated labs—eliminates Wigner's Friend inconsistencies and predicts a joint 'ok' probability of 1/4, replacing the previous 1/12.","keywords":["Wigner's Friend paradox","extended Wigner's Friend","Copenhagen interpretation","wave-function collapse","observer-independent quantum state","measurement problem","Born rule"],"falsifier":"Run the extended two-lab procedure many times, with internal measurements performed by automated devices, and tally how often both outside observers get 'ok'. The modified collapse picture predicts 25%; the coherent-branch picture predicts about 8.3%. A statistically clear rate at one of these values would settle between the two analyses.","tokens_in":12934,"feed_emoji":"⚛️","tokens_out":5275,"duration_ms":55402,"temperature":0.7,"pith_summary":"This paper proposes a modified Copenhagen interpretation that removes Wigner's Friend and extended Wigner's Friend inconsistencies. The key move is to treat wave-function collapse during measurement as genuine quantum time evolution that also occurs inside isolated labs, and to separate that objective state from the partial knowledge conscious observers may have about it. Under these assumptions the earlier extended-setup calculation changes: the probability that both outside observers see the outcome 'ok' becomes 1/4, not 1/12. Because the altered number comes from collapse rather than from a coherent superposition of branches, the difference is in principle observable. The authors acknowledge that a full mechanistic account of collapse remains an open problem.","feed_headline":"Wigner's friend paradox becomes a 1-in-4 vs 1-in-12 bet","feed_subtitle":"Counting collapse inside isolated labs predicts joint 'ok' 25% of the time—a difference an experiment could settle.","key_machinery":"The machinery is the axiom that measurement consists of a projection onto the eigenspace of the observed eigenvalue, with the outcome realized according to Born probabilities, and that this projection is part of the system's time evolution even when the system is isolated from any external observer. Applied inside Wigner's friend labs, this axiom turns an entangled superposition of 'heads' and 'tails' branches into a set of definite alternative states, each with its own probability. The calculation in equation (13) is the point where the axiom changes the result: each definite branch overlaps with the external 'ok' state with amplitude 1/2, giving 3 × (1/3) × (1/2)^2 = 1/4, whereas the coher","core_discovery":"The central claim is that all inconsistencies in Wigner's Friend and extended Wigner's Friend setups disappear if measurement collapse is part of quantum mechanical time evolution even or especially in isolated labs, the quantum state is universal and observer-independent, and observer knowledge is tracked separately via intersubjective probabilities. The decisive calculation concerns the extended setup's final double measurement: after the friend in the second lab measures the spin along z, the lab state collapses to one of the three definite branches (12a)-(12c) instead of remaining a coherent superposition. Computed from these branches, the probability that both external observers registe","pith_inferences":["The 1/4-versus-1/12 gap suggests a concrete experimental target: realize the two internal measurements with controllable 'friend' systems and count the joint 'ok' rate; the two branches of quantum mechanics predict different rates.","The paper's distinction between objective state and intersubjective knowledge could be exported to other quantum puzzles, such as delayed-choice and counterfactual measurement scenarios, where partial information is usually conflated with state update.","The framework implies that there is a definite time (when the macroscopic device registers the result) at which the branch structure changes; experiments that tune the size or sensitivity of the 'measurement device' might map the crossover from coherence to collapse-like behavior."],"forward_implications":["The extended Wigner's Friend setup becomes a consistency check rather than a contradiction: observers can disagree only in knowledge, not in state.","The joint 'ok'-and-'ok' outcome occurs with probability 1/4; each of the four outcome pairs is equally likely.","The same reasoning carries over to spontaneous collapse models and to unitary-plus-statistical-mechanics accounts of measurement, since both realize collapse without an observer.","If a future mechanistic theory replaces the collapse axiom, the argument is meant to remain valid because it only relies on the existence of some projection-like transition inside isolated labs."],"supporting_citations":[{"why":"Supplies the original Wigner's Friend thought experiment that the paper's modified Copenhagen interpretation is designed to resolve.","marker":"Wigner (1961)"},{"why":"Defines the extended Wigner's Friend setup and the previous probability calculation (p=1/12) that the paper replaces with equation (13).","marker":"Frauchiger and Renner (2018)"},{"why":"Shows an analogous observable probability difference between quantum interpretations, which the paper invokes to argue its collapse-based difference is testable.","marker":"Deutsch (1985)"},{"why":"Provides the condensed statement of the extended setup (two friend labs, two external Wigner measurements) used in Section 4.","marker":"Lazarovici and Hubert (2019)"},{"why":"Supports the assumption that collapse can occur as internal time evolution via stochastic modifications of the Schrödinger equation.","marker":"Bassi et al. (2013)"},{"why":"Cited as evidence for a universal, observer-independent quantum state, one of the three modified assumptions.","marker":"Pusey et al. (2012)"}],"fun_headline_variants":["Collapse inside labs ends Wigner's friend paradox","Modified Copenhagen: collapse runs in isolated labs","Wigner's friend 'ok' odds: 25% vs 8% — testable","Observer-independent state resolves friend paradox","Friend paradox falls if collapse obeys time evolution"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The whole resolution hangs on the premise that measurement produces a genuine wave-function collapse inside an isolated system as part of time evolution; if collapse happens only when an outside observer looks, the branches stay coherent and the paradox returns.","fun_headline_variants_meta":{"raw":{"variants":["Collapse inside labs ends Wigner's friend paradox","Modified Copenhagen: collapse runs in isolated labs","Wigner's friend 'ok' odds: 25% vs 8% — testable","Observer-independent state resolves friend paradox","Friend paradox falls if collapse obeys time evolution"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000221,"raw_usage":{"total_tokens":1272,"prompt_tokens":712,"completion_tokens":560,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":456,"completion_tokens_details":{"reasoning_tokens":482}},"tokens_in":456,"tokens_out":560,"duration_ms":6809,"temperature":1.0,"reasoning_tokens":482,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T04:21:53.052736+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the extended two-lab procedure many times, with internal measurements performed by automated devices, and tally how often both outside observers get 'ok'. The modified collapse picture predicts 25%; the coherent-branch picture predicts about 8.3%. A statistically clear rate at one of these values would settle between the two analyses.","supporting_citations":[],"review_version":1}