{"id":"1586782c-97af-457e-9326-c0b1f5633c78","arxiv_id":"2507.05337","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"With a post-selected holographic map for the baby universe, the Engelhardt-Gesteau swap test cannot distinguish the two candidate bulk duals.","lead":"A single boundary state in AdS/CFT can have two possible gravity descriptions, one with a baby universe. This paper proposes a post-selected holographic map that makes a standard swap test unable to pick between the two descriptions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The paper's central claim that S_∂ cannot distinguish the two bulk states depends on the unproven existence of a post-selection state |χ> satisfying the consistency condition (9) for AR's specific states; without such a χ, V_baby does not map ψ(baby) to Ψ(∂) and the swap test remains discriminating.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing condition: the existence of a post-selection state |χ> and a non-isometric map V_baby satisfying the consistency condition (8) for the specific AR states. My analysis confirms that this assumption is not a minor technicality but the linchpin of the entire argument. The equality ⟨S_∂⟩ = ⟨S_baby⟩_ψ(baby) in Eq. (16) is formally correct for any V_baby once S_baby is defined by Eq. (11); however, this equality only carries physical meaning if V_baby actually maps ψ(baby) to the boundary state Ψ(∂). That mapping is equivalent to Eq. (9), which the paper postulates rather than derives. Without an explicit construction of |χ> for AR's states, the proposed 'second holographic map' is not shown to exist, and the swap test retains its original discriminating power. I also noticed a small typo in Eq. (14): the prefactor should be |i|^2 rather than |i| to be consistent with the qubit calculation in Eq. (21). This is easily corrected and does not affect the central logic, so I do not treat it as the primary concern. Since the reader already issued a conditional verdict based on the same assumption, my stress-test does not move the verdict; it reinforces the need for the authors to supply the missing existence proof.","tokens_in":7789,"tokens_out":10231,"duration_ms":113829,"concrete_test":"Take the explicit AR states ψ(nb) and ψ(baby) as defined in [12] and numerically solve the fidelity maximization max_{|χ>∈H_i, ||χ||=1} F(ψ(nb), |i| (1_ab⊗⟨χ|_i) ψ(baby) (1_ab⊗|χ>_i)) in a finite-dimensional truncation of the heavy-operator sector; if the maximum fidelity is strictly less than 1, Eq. (9) has no solution, so condition (8) cannot be satisfied and the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central result (16) is a tautology once V_baby is defined as |i|^{1/2} V_nb ⊗ ⟨χ|_i: for any boundary operator, ⟨O_∂⟩ = tr(ψ(baby) V_baby† O_∂ V_baby) by trace cyclicity. The physical content is that V_baby is a legitimate holographic map for ψ(baby), which requires V_baby ψ(baby) V_baby† = Ψ(∂), i.e., the consistency condition (8). This condition is equivalent to Eq. (9), which asserts that the AR no-baby state ψ(nb) equals (up to normalization) the state obtained by projecting ψ(baby) onto a single post-selection vector |χ> in the baby Hilbert space. The paper simply assumes such a χ exists; it is not derived from AR's state preparation or from the extrapolate dictionary. For generic ψ(baby) and ψ(nb), Eq. (9) is a strong constraint—there may be no unit vector |χ> that reproduces the given ψ(nb) after projection. The toy model in §2.1 evades this by setting V_nb = 1 and choosing |χ*> = |ψ(nb)>, which is circular. If Eq. (9) fails for the actual AR states, then no V_baby satisfying (8) exists, S_∂ has a unique bulk dual via the isometric extrapolate map, and the paper's conclusion that the swap test cannot distinguish the states collapses.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper challenges Engelhardt and Gesteau's recent swap-test evidence against semiclassical baby universes in AdS/CFT. The author proposes that the extrapolate dictionary does not uniquely define a holographic map: in addition to the isometric no-baby map V_nb, one may define a non-isometric baby-universe map V_baby = |i|^{1/2} V_nb ⊗ <χ|_i that post-selects on the baby universe. Under the consistency condition (8), restated as (9), the same boundary state Ψ(∂) is dual to both AR bulk states, and the boundary operator S_∂ acquires two bulk duals, S and S_baby, whose expectation values agree (Eq. (16)). Section 3 argues that when the boundary state is entangled with an external reference, the naive and non-perturbative swap predictions both become exponentially small, so the simplicity of S_∂ does not rule out the semiclassical baby state. Section 4 sketches how recent observer-in-holography rules could be applied to V_baby.","tokens_in":8136,"tokens_out":5802,"duration_ms":68951,"significance":"If the central consistency condition can be established for the actual AR states, the paper would substantially change the interpretation of the EG swap test: the boundary operator would no longer discriminate between the no-baby and baby bulk duals, and semiclassical baby universes would remain viable in at least some regimes. The formal traces leading to Eq. (16) are clean, and the qubit model usefully illustrates the role of post-selection. The paper also makes a concrete falsifiable claim: whenever Eq. (9) admits a solution |χ>, the EG test is non-discriminating; when it does not, the original EG conclusion is untouched. The main weakness is that the existence of |χ> and the Section 3 entanglement structure are assumed rather than derived, so the central claim is conditional on unproved inputs.","major_comments":[{"comment":"The central result (16) depends on the consistency condition V_nb ψ(nb) V_nb† = V_baby ψ(baby) V_baby†, which is imposed rather than demonstrated for the specific AR states. The paper does not prove that a unit vector |χ>_i satisfying Eq. (9) exists; projecting a generic pure baby state ψ(baby) onto a single baby-universe state |χ> yields a particular pure state on H_a⊗H_b, and there is no reason this state must coincide with ψ(nb) up to normalization. If no such |χ> exists, then no V_baby of the form (7) satisfies (8), the equality (16) fails, and the conclusion that S_∂ cannot distinguish the two bulk duals collapses. The qubit model in §2.1 does not fill this gap: with V_nb = 1, Eq. (20) simply sets |ψ(nb)> equal to the projected state |χ*>_ab, which is circular. The author should either prove the existence of |χ> for the actual AR states or state this as an explicit assumption and explain its physical origin.","section":"Section 2, Eqs. (8)-(9)"},{"comment":"The claim that both the naive and non-perturbative swap expectation values become exponentially small in an entangled-reference regime rests on the unproved physical assertion that the heavy operator O^(k) causes the bulk excitation to fall into the baby universe, so that no entanglement exists between ab and R in the baby state. This is a dynamical semiclassical input, not a consequence of the post-selection map; the equality <S>_ψ(baby) = e^{-S_2(ab)} in Eq. (24) is asserted without derivation. If the reference R remains entangled with ab, the naive prediction need not be suppressed in the same way, and the simplicity argument in Section 3 would not apply. The author should derive this entanglement structure from AR's state preparation or identify the approximation under which it holds.","section":"Section 3, Eqs. (22)-(24)"},{"comment":"Once V_baby is defined as in (7) and the consistency condition (8) is assumed, the equality (16) follows immediately from trace cyclicity; it is therefore a consistency condition rather than an independent test. The physical content lies entirely in the existence and correctness of V_baby for ψ(baby), and the paper should provide evidence that V_baby is the map selected by the extrapolate dictionary, rather than an arbitrary auxiliary construction. Without such evidence, the statement that S_∂ 'cannot distinguish' the two bulk states is a restatement of the imposed condition (8), not a new dynamical result.","section":"Section 2, Eqs. (14)-(16)"}],"minor_comments":[{"comment":"The notation |i| = dim H_i is confusing because i also labels the baby universe; please use a distinct symbol such as d_i and define it at first use.","section":"Section 2, Eq. (7)"},{"comment":"The approximate equality symbol ≈ in Eq. (8) is never specified; the author should state whether the equality is exact, or if approximate, in which norm or trace distance and with what tolerance.","section":"Section 2, Eq. (8)"},{"comment":"The qubit model writes |ψ(baby)> = |Φ^+>_{a,i1} |Φ^+>_{b,i2}; please clarify the normalization conventions for the Bell states and the overall normalization of |ψ(baby)>.","section":"Section 2.1, Eq. (17)"},{"comment":"The notation S(ab) = S(R) and S_2(R) is introduced without defining the Renyi entropy convention; please define S_2 and explain why the no-baby entanglement entropy equals the reference entropy.","section":"Section 3, after Eq. (22)"},{"comment":"The sentence 'Either approach can be readily applied to V_baby' is a sketch; since the HUZ and Colorado rules differ, a brief description of how each rule modifies V_baby would make the claim more transparent.","section":"Section 4, Eq. (25)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well written and addresses an important recent controversy. My main reservation is technical: the central claim is conditional on the existence of a post-selection state |χ> satisfying Eq. (9) for AR's actual states, and on the unstated semiclassical assumption in Section 3. These are load-bearing, but they are the kind of gap that could in principle be closed with a derivation or an explicit set of assumptions. I do not see grounds for rejection if the author can strengthen these points, but the current version is not yet sufficient for acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this paper gives a clean, conditional argument that a post-selected holographic map can make the EG swap test inconclusive, but the condition is doing all the work and isn't demonstrated. Section 2's algebra is fine. Define V_baby = |i|^{1/2} V_nb ⊗ ⟨χ|_i, and trace cyclicity gives ⟨O_∂⟩ = ⟨V_baby† O_∂ V_baby⟩. That part is unassailable. The physical content is the consistency condition (8), which is equivalent to demanding that a unit vector |χ⟩ exists such that ψ(nb) = |i| (1⊗⟨χ|)ψ(baby)(1⊗|χ⟩). That's a strong constraint on AR's specific states, and the paper simply assumes it. The toy qubit model evades the problem by setting V_nb = 1 and choosing χ to match the desired ψ(nb), which is circular.\n\nSection 3 is softer still. The claim that a heavy operator's bulk excitation falls into the baby universe, so that the reference R purifies i alone, is a semiclassical assertion with no derivation. It might be true, but it isn't established.\n\nWhat's genuinely new: the identification of S_baby as the dual of S_∂ under the post-selected map, and the observation that the same boundary operator can have two bulk duals with matching expectation values. That's a useful counterpoint to EG, and the connection to the observer program is timely. The citation pattern is unobjectionable, and the paper properly notes simultaneous related work [22].\n\nThis is for readers following the AR/EG debate in quantum gravity and holography. It deserves a serious referee because the question is important and the conditional logic is worth having on record. But it should come back with either a derivation of χ for the AR states or an explicit reframing as a consistency argument conditional on that existence. Section 3 needs more support. I wouldn't cite it as a settled result, but I'd send it out for review.","headline":"A clean but conditional construction: the swap-test conclusion only follows if a post-selection state χ exists for the AR states, and that existence is assumed, not derived.","tokens_in":8649,"tokens_out":2417,"would_cite":false,"duration_ms":29415,"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 boundary swap test cannot distinguish a baby universe from a no-baby bulk state, so semiclassical baby universes survive in AdS/CFT.","keywords":["baby universes","AdS/CFT","holographic dictionary","swap test","post-selection","non-isometric maps","bulk reconstruction","observers in holography"],"falsifier":"A direct first-principles boundary computation of the swap expectation for the heavy-operator CFT state in a solvable model (for instance a two-dimensional dilaton-gravity theory or a large-N tensor model) would settle the claim: if it matches the naive baby-state value rather than the no-baby value, the equality fails. Alternatively, attempting to derive the post-selection state from the bulk path integral and finding no solution would collapse the construction.","tokens_in":7529,"feed_emoji":"🌌","tokens_out":19232,"duration_ms":166122,"temperature":0.7,"pith_summary":"A state in a conformal field theory (CFT) was previously shown to admit two equally valid bulk duals, one containing a baby universe and one not. A recent swap test concluded that the boundary dual of a bulk swap operator favors the no-baby state, under the assumption that the extrapolate dictionary defines a unique isometric holographic map. This paper argues that the extrapolate dictionary, the standard boundary-to-bulk reconstruction rule, does not fix a unique map for a closed universe because a baby universe has no asymptotic boundary. Post-selecting on the baby universe defines a second, non-isometric holographic map under which the same boundary operator has a different bulk dual whose expectation value exactly matches the no-baby prediction. The swap test therefore cannot distinguish the two candidate states, leaving semiclassical baby universes viable and opening the way to including observers in the baby universe.","feed_headline":"Baby universes survive the boundary swap test","feed_subtitle":"Post-selection gives the swap operator a second bulk dual, so the test can't choose between the states.","key_machinery":"The load-bearing object is the second holographic map $V_{\\text{baby}}$, defined by $V_{\\text{baby}} \\equiv |i|^{1/2}\\, V_{\\text{nb}} \\otimes \\langle\\chi|_i$: it post-selects the entire baby-universe Hilbert space onto a fixed bra $\\langle\\chi|$, making the map non-isometric while keeping the extrapolate-dictionary action on the two asymptotic AdS components identical to $V_{\\text{nb}}$. The induced operator map $V^*(O_{\\partial}) = V^\\dagger O_{\\partial} V$ converts the boundary swap operator into $S_{\\text{baby}} = |i|\\, S \\otimes |\\chi\\rangle\\langle\\chi|_i \\otimes |\\chi\\rangle\\langle\\chi|_{i'}$, a swap on the AdS factors dressed by projectors on two copies of the baby universe. Trace cyclicity then forces the equality of expectation values, which is the entire argument.","core_discovery":"The central claim is that the boundary operator $S_{\\partial}$ has two consistent bulk duals. With the isometric no-baby map $V_{\\text{nb}}$ it is the bulk swap operator $S$; with the new post-selecting map $V_{\\text{baby}} = |i|^{1/2}\\, V_{\\text{nb}} \\otimes \\langle\\chi|_i$ it is $S_{\\text{baby}} = |i|\\, S \\otimes |\\chi\\rangle\\langle\\chi|_i \\otimes |\\chi\\rangle\\langle\\chi|_{i'}$. The two maps are chosen so that $V_{\\text{nb}}\\,\\psi^{(\\text{nb})}\\,V_{\\text{nb}}^\\dagger = V_{\\text{baby}}\\,\\psi^{(\\text{baby})}\\,V_{\\text{baby}}^\\dagger$, and trace cyclicity then gives $\\langle S_{\\partial}\\rangle = \\langle S\\rangle_{\\psi^{(\\text{nb})}} = \\langle S_{\\text{baby}}\\rangle_{\\psi^{(\\text{baby})}}$. Measuring $S_{\\partial}$ therefore does not amount to measuring the naive swap operator in the baby state, and the swap test does not favor the no-baby state. When the boundary state is entangled with an external reference, both the naive and the post-selected swap expectation values in the baby state are exponentially small, so the simplicity of $S_{\\partial}$ cannot rule the baby state out either.","pith_inferences":["The consistency condition relating the no-baby and baby states is imposed rather than derived; a first-principles derivation of the post-selection state from the bulk path integral would convert this consistency argument into a predictive dictionary.","The same post-selection logic should apply to other boundary operators, not just the swap operator; if it does, the holographic dictionary becomes state-dependent in a way that may conflict with boundary linearity in settings beyond this one.","Section 3's exponential-suppression mechanism suggests a concrete scaling test: entangle the boundary state with references of different sizes and locate the transition where the swap test loses its discriminating power.","If observers inside the baby universe have no equivalent description in the no-baby state, the observers themselves supply the extra data that resolves the uniqueness puzzle; the paper leaves this as an open question."],"forward_implications":["The boundary swap operator is not evidence against a semiclassical baby universe, because the same measurement is equally well described by the post-selected bulk operator in the baby state and by the original swap in the no-baby state.","The extrapolate dictionary is compatible with non-isometric, post-selecting holographic maps, so a closed universe does not need a uniquely defined boundary-to-bulk map.","When the CFT state is entangled with an external reference, both the naive and the post-selected swap expectation values in the baby state are exponentially small, making the two predictions indistinguishable to any simple experiment.","Applying observer-inclusion rules to $V_{\\text{baby}}$ enlarges the fundamental Hilbert space to $\\mathcal{H}_A \\otimes \\mathcal{H}_B \\otimes \\mathcal{H}_I$, giving the baby universe non-trivial holographic content and new bulk operators.","The correct reading of the earlier swap-test result is the non-uniqueness of bulk reconstruction rather than the invalidity of the semiclassical baby universe."],"supporting_citations":[{"why":"Supplies the two candidate bulk states and the requirement that both map to the same CFT state.","marker":"[12]"},{"why":"Defines the bulk swap operator, its boundary dual, and the conclusion that this paper reinterprets.","marker":"[19]"},{"why":"Provides a rule for including observers in holographic maps, applied to the new map in Section 4.","marker":"[8]"},{"why":"Provides an alternative observer rule and the non-isometric post-selection viewpoint.","marker":"[10]"},{"why":"Gives earlier evidence that holographic maps on closed universes act by post-selection.","marker":"[20]"},{"why":"Motivates the simplicity and non-isometric-code reasoning used in Section 3.","marker":"[21]"}],"fun_headline_variants":["Swap test cannot choose between baby and no-baby universes","Post-selection makes baby universe duals indistinguishable","New holographic map blurs swap test result","Swap test indecisive on baby universes","Post-selection saves baby universes from swap test"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction assumes, rather than derives, that there is a post-selection state in the baby-universe Hilbert space making the two holographic maps agree on the two candidate bulk states in question, with this state exhibited only in a toy qubit model with a trivial extrapolate dictionary.","fun_headline_variants_meta":{"raw":{"variants":["Swap test cannot choose between baby and no-baby universes","Post-selection makes baby universe duals indistinguishable","New holographic map blurs swap test result","Swap test indecisive on baby universes","Post-selection saves baby universes from swap test"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001207,"raw_usage":{"total_tokens":5000,"prompt_tokens":1002,"completion_tokens":3998,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":3923}},"tokens_in":618,"tokens_out":3998,"duration_ms":27762,"temperature":1.0,"reasoning_tokens":3923,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:28:58.018200+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct first-principles boundary computation of the swap expectation for the heavy-operator CFT state in a solvable model (for instance a two-dimensional dilaton-gravity theory or a large-N tensor model) would settle the claim: if it matches the naive baby-state value rather than the no-baby value, the equality fails. Alternatively, attempting to derive the post-selection state from the bulk path integral and finding no solution would collapse the construction.","supporting_citations":[{"cited_title":"Akers, G","cited_arxiv_id":null,"evidence_quote":"Provides an alternative observer rule and the non-isometric post-selection viewpoint."},{"cited_title":"Akers, N","cited_arxiv_id":null,"evidence_quote":"Motivates the simplicity and non-isometric-code reasoning used in Section 3."}],"review_version":1}