{"id":"cd5fd19e-b9f2-466e-b62a-f41b0da21129","arxiv_id":"2505.22219","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":8,"one_line_summary":"In a Buchert-averaged two-domain model, the 21 cm absorption trough is roughly 40 mK deeper for H0=67 than for H0=73, a difference absent in the paper's ΛCDM comparison.","lead":"This paper computes the 21 cm hydrogen signal in a model universe split into dense and empty regions and finds the absorption dip is much deeper for H0=67 km/s/Mpc than for H0=73 km/s/Mpc. It offers 21 cm cosmology as a possible new way to weigh in on the Hubble constant debate, but the calculation assumes the average expansion rate can be inserted into a homogeneous signal formula.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central H0-dependent trough is computed by inserting the Buchert volume-average expansion into the homogeneous 21 cm formula; the sky-averaged signal in a clumpy universe is not given by that proxy, so the 40 mK claim is unestablished.","rationale":"The paper is best read as a forward model in an alternative cosmology: Buchert averaging with scaling laws, calibrated to Union 2.1 SNe and N-body structure formation. The qualitative mechanism—lower H0 increases optical depth—is real and would survive in any model. What is not established is the quantitative claim that the global 21 cm trough differs by ~40 mK between H0=67 and H0=73 in this inhomogeneous universe. Equations (1)-(2) are the standard homogeneous expressions; Eq. (2) contains nHI, Ts, and the line-of-sight velocity-gradient term. The paper replaces only H(z) by the Buchert-average HD(z) and drops δrvr, effectively assuming that the homogeneous formula with volume-averaged expansion gives the sky-averaged signal. That assumption fails in general: the sky-averaged brightness temperature is an average of a strongly nonlinear function of local density and velocity, not the same function evaluated at the volume-averaged expansion. Because the backreaction model is built on order-unity density contrasts, this is not a small correction. The three-domain robustness test in Appendix A uses the same proxy, so it does not validate the observable mapping. Secondary concerns (post hoc choice n1=-1.7, no error bars on T21 curves) would matter for precision but are not the decisive issue. The decisive issue is the missing line-of-sight averaging. A concrete calculation within the authors' two-domain framework—averaging local T21 over M and E regions with their respective volume fractions—would settle whether the predicted H0 difference is a real feature or an artifact of the proxy. Until that is done, the central claim is unverified. This supports the reader's REJECT verdict.","tokens_in":19708,"tokens_out":4762,"duration_ms":53904,"concrete_test":"Within the authors' own two-domain framework, compute the sky-averaged T21 as the volume-weighted average of the local brightness temperatures of the M and E regions: at each redshift, use the local expansion rate HF(z), local density nHI,F ∝ ⟨ρ⟩F, and solve the local spin-temperature equations for each region, then average T21,M and T21,E with weights λM(z) and λE(z). Compare the resulting H0=67 vs H0=73 curves with Fig. 3. If the trough-depth separation changes by more than ~10 mK, or if the ordering reverses, the homogeneous proxy is invalid. A complementary check is to run a standard 21 cm code on an N-body realization with the same expansion history and compare the sky-averaged signal.","verdict_should_be":"REJECT","load_bearing_attack":"The paper's central quantitative claim (Sec. V, Fig. 3) is obtained by taking the Buchert-averaged global Hubble parameter HD(z) from Eq. (31) and substituting it into the homogeneous brightness-temperature formula, Eqs. (1)-(2). This substitution is not justified. In a universe with order-unity density contrasts, the observable global 21 cm signal is a line-of-sight and volume average of the local brightness temperature, which depends on the local neutral-hydrogen density nHI, local spin temperature Ts, local expansion/velocity-gradient term H+(1+z)δrvr, and local Lyα coupling. The Buchert HD(z) is the volume-weighted average expansion of the domain and is dominated by the underdense regions (λE ~0.91), while the 21 cm absorption is weighted by nHI, which is much larger in overdense regions. Consequently ⟨nHI/H⟩ ≠ ⟨nHI⟩/⟨H⟩, and the optical depth can differ substantially. The peculiar-velocity gradient term δrvr in Eq. (2) is dropped on the grounds that it is local, but in a model whose entire premise is large density contrasts, this term is of order H in overdense regions and cannot be neglected. Without a derivation of the averaged 21 cm observable (analogous to the covariant distance relations, Eqs. (36)-(37), used for SNe), the predicted ~40 mK difference between H0=67 and H0=73 is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a two-domain Buchert backreaction model (overdense/underdense regions) with power-law scaling solutions for backreaction and curvature, calibrates the effective matter density parameter Omega_m^D0 against Union 2.1 supernovae via MCMC, and selects the scaling exponent n1 by comparing the model's overdense volume fraction with N-body simulation data. The authors then compute the global 21 cm brightness temperature T21(z) by inserting the volume-averaged Hubble parameter H_D(z) into the standard homogeneous brightness-temperature formula, obtaining a deeper absorption trough for H0 = 67 km/s/Mpc (about -320 mK at z ~ 18) than for H0 = 73 km/s/Mpc (about -280 mK), and claim that this H0 dependence is a clear observational signature of backreaction that is absent in Lambda CDM.","tokens_in":20132,"tokens_out":5355,"duration_ms":60966,"significance":"If the central claim were established, the 21 cm absorption trough would be a novel probe of the Hubble tension and of the backreaction of cosmic inhomogeneities. The paper has some genuine strengths: the MCMC calibration to supernova data is concrete, the comparison with N-body volume fractions gives a physical anchor to the scaling exponents, and the three-domain extension in Appendix A is a useful robustness check. However, the key quantitative result -- the ~40 mK difference between the two H0 values in Fig. 3 -- rests on substituting the averaged Hubble parameter into a local homogeneous observable without deriving the appropriate averaged 21 cm signal, and it is presented without any propagated uncertainty from the poorly constrained Omega_m^D0 and the post hoc choice of n1. These issues are load-bearing for the paper's central claim.","major_comments":[{"comment":"The global 21 cm brightness temperature is computed by substituting the volume-averaged Hubble parameter H_D(z) into the homogeneous formula for T21 and the optical depth tau, but the paper does not derive this as the correct sky-averaged observable in a universe with order-unity density contrasts. The optical depth depends nonlinearly on the local neutral hydrogen density nHI, the local spin temperature Ts, and the local velocity gradient H(z) + (1+z) delta_r v_r. The paper drops the peculiar-velocity term in Eq. (2) as 'local,' yet in overdense regions, which dominate the 21 cm absorption because nHI is much larger there, this term is typically of order H and cannot be neglected. Since <nHI/H> is not equal to <nHI>/<H>, the substitution H -> H_D(z) in Eqs. (1)-(2) is not justified, and the predicted ~40 mK difference between H0 = 67 and H0 = 73 in Fig. 3 is not established.","section":"Sec. II A, Sec. V, Eqs. (1)-(2), Fig. 3"},{"comment":"The central prediction in Fig. 3 is presented as a deterministic curve with no uncertainty band. The best-fit Omega_m^D0 = 0.104 ± 0.056 from Table I has a relative error of roughly 50%, and the scaling exponent n1 = -1.7 is selected post hoc by visual comparison to N-body volume fractions without any quantitative goodness-of-fit criterion. No error propagation from Omega_m^D0 or n1 into T21 is attempted, so the claim that the trough depth and position constitute a 'clear observational signature' is not supported by the stated uncertainties.","section":"Sec. V, Table I and Fig. 3"},{"comment":"The calculation does not specify how the baryon number density nH entering Eqs. (2), (5), and (6) is defined in the inhomogeneous Buchert framework. In the two-domain model, the domain-averaged matter density <rho>_D differs from the local densities in the overdense and underdense regions, and the 21 cm optical depth depends on the local neutral hydrogen density, not the volume average. Without an explicit mapping between nH and the averaged model quantities, the T21(z) curves in Figs. 1, 3, and 4 are not well defined, and the comparison with the standard homogeneous treatment is ambiguous.","section":"Sec. II A - IV, Eqs. (2)-(6)"},{"comment":"The model's 'consistency with structure formation' is achieved by choosing the scaling exponent n1 after inspecting the N-body volume fractions; n1 is thus effectively a calibrated free parameter rather than a prediction. The four values in Table I are used to demonstrate that only n1 = -1.7 'matches consistently' with the simulation data, but no statistical measure of this match is given, and no test of how the final T21 prediction depends on the allowed range of n1 is presented. This weakens the claim that the model is jointly calibrated by supernova and structure-formation data.","section":"Sec. III and Sec. V"}],"minor_comments":[{"comment":"The sentence 'the domination is dictated by the tau part, viz., tau ∝ e^{-H} from (Eq. 1)' is dimensionally and functionally incorrect; the intended scaling appears to be tau ∝ 1/H_D(z), as stated later in Sec. V.","section":"Sec. IV, paragraph after Fig. 1"},{"comment":"The table gives errors on Omega_m^D0 as superscript/subscript asymmetric values, but the text does not state whether these denote 68% or 95% credible intervals; this should be specified.","section":"Sec. V, Table I"},{"comment":"The N-body data points for the overdense volume fraction are described only qualitatively as extracted with a block separation technique; without error bars or a precise definition of the grid and counting procedure, the visual agreement claimed for n1 = -1.7 is difficult to assess.","section":"Sec. V, Fig. 2(b)"},{"comment":"The choice n3 = -1.8 for the ambient region is described as 'reasonable' but no motivation from simulations or observations is given; a sensitivity test over the full range -2 ≤ n3 ≤ -1.7 would strengthen the robustness claim in Fig. 4.","section":"Appendix A"}],"recommendation":"reject","confidential_remarks":"The paper addresses an interesting and timely question, but the core quantitative claim -- an H0-dependent 21 cm absorption trough of order 40 mK produced by backreaction -- is derived by inserting the volume-averaged Hubble parameter into a local homogeneous observable without deriving the correct averaged 21 cm signal. This is not a cosmetic or local issue; it is the central result of the paper. The post hoc selection of n1 and the absence of any error propagation further weaken the claim. I therefore recommend rejection. Should the authors substantially revise the manuscript to derive the averaged 21 cm observable or explicitly reframe the computation as a toy-model illustration with clear caveats, a resubmission could be considered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe genuinely new thing here is the application of Buchert backreaction scaling laws to the 21 cm brightness temperature. That combination is not in the cited literature, and the claimed H0 sensitivity of the trough—deeper for H0=67 than for H0=73, with a redshift shift—is a concrete, falsifiable-sounding prediction. The paper is also transparent about its machinery: the MCMC calibration to Union 2.1, the comparison to N-body volume fractions, and the three-domain robustness check are all shown. Credit where due: the authors are not hiding the ball, and the robustness appendix is a good-faith attempt.\n\nBut the central quantitative claim does not hold up as stated. The 21 cm brightness temperature in Eq. (1)-(2) is a homogeneous formula. They substitute the volume-averaged Buchert expansion HD(z) into that formula and call it the global signal. In a universe with order-unity density contrasts, the sky-averaged 21 cm signal is a line-of-sight average over regions with very different nHI, Ts, and velocity-gradient terms. Their own premise—that the underdense region dominates the volume at 91%—makes the proxy especially suspect, since the optical depth is weighted by nHI, which lives mostly in overdense regions. The peculiar-velocity gradient term is dropped because it is \"local,\" but if the model's whole point is that density contrasts are large, that term can be of order H in overdense regions. Without a derivation of the averaged observable analogous to the covariant distance relations they use for SNe, the ~40 mK difference is not established.\n\nThe second problem is the post hoc selection of the scaling exponent n1. They scan n1 from -2 to -1, fit Omega_m from SNe for each, then pick n1=-1.7 because it matches the N-body volume fraction. That is a fair way to calibrate a model, but it means the output is not a parameter-free prediction. And the curves in Fig. 3 have no error bars from the MCMC posteriors, so even internally we cannot see whether the 40 mK difference is significant.\n\nIs the paper worthless? No. The mechanism is physically sane: lower H gives higher optical depth, and the H0-dependence is genuinely absent in standard LCDM. If someone later does the proper inhomogeneous line-of-sight treatment, this paper would be the obvious reference for the idea. But as submitted, the headline result is unsupported.\n\nVerdict: I would accept it for peer review—it is a serious, if flawed, attempt at a new observable—but I would expect the referees to require either a proper averaging derivation or a much more hedged claim. The paper is for readers working on backreaction and 21 cm theory; it will not convince a skeptic without the missing derivation.\n\nRecommendation: send it to referees, but make sure at least one referee is deeply familiar with both Buchert averaging and 21 cm radiative transfer.","headline":"A transparent but premature forward-modeling paper: the 21 cm H0 sensitivity is real in the model, but the observable mapping is unproven and the post hoc scaling choice carries the whole claim.","tokens_in":20647,"tokens_out":755,"would_cite":true,"duration_ms":11345,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that a backreaction-averaged universe makes the 21 cm absorption trough about 40 mK deeper for H0=67 than for H0=73, a difference that does not appear in ΛCDM.","keywords":["21 cm cosmology","Hubble tension","backreaction","cosmological averaging","brightness temperature","cosmic dawn","structure formation","supernova constraints"],"falsifier":"Run the same two-domain backreaction model through a full 21 cm light-cone radiative-transfer code that computes the optical depth region by region from local gas density, spin temperature, and peculiar velocity gradients; if the predicted gap of about 40 mK between $H_0=67$ and $H_0=73$ disappears or reverses, the central claim is falsified.","tokens_in":19496,"feed_emoji":"📡","tokens_out":13106,"duration_ms":130511,"temperature":0.7,"pith_summary":"This paper argues that the depth of the global 21 cm absorption trough can act as a probe of the Hubble tension, provided the universe's matter distribution is treated with a backreaction averaging scheme rather than as a homogeneous fluid. The authors model the cosmos as overdense and underdense regions with separate expansion histories, calibrate the effective matter density with supernova distance data, and choose the scaling exponent that reproduces the volume fraction of overdense regions in N-body structure-formation simulations. With these calibrated parameters, a low $H_0=67\\ \\mathrm{km/s/Mpc}$ value produces a trough of roughly $-320\\ \\mathrm{mK}$ at $z\\approx18$, while a high $H_0=73\\ \\mathrm{km/s/Mpc}$ value gives roughly $-280\\ \\mathrm{mK}$. Standard $\\Lambda$CDM changes the signal by only a few mK between the same two values, so the paper presents this gap as an observational signature of inhomogeneous backreaction. If the claim holds, a precise measurement of the cosmic dawn spectrum could weigh in on whether the Hubble tension comes from hidden systematics, new physics, or the way cosmic voids and clusters alter the averaged expansion.","feed_headline":"Lower Hubble constant deepens the 21 cm absorption trough","feed_subtitle":"In a universe averaged over dense and empty regions, H0=67 gives a ~320 mK dip, H0=73 gives ~280 mK; ΛCDM shows no gap.","key_machinery":"The load-bearing object is the backreaction-averaged Hubble parameter $H_D(z)$ of a domain divided into overdense and underdense subdomains, each with its own scale factor and expansion rate. Backreaction here is the averaged effect of differences in local expansion and shear, quantified by the kinematical backreaction term $Q_D$, and curvature is tracked by the averaged Ricci scalar $\\langle R\\rangle_D$. The system is closed by power-law ansatze $Q_F\\propto a_F^{n}$ and $\\langle R\\rangle_F\\propto a_F^{n}$ linked through the integrability condition; the choice $n_1=-1.7$ for the underdense region and $n_2=-2$ for the overdense region is selected because it matches the overdense volume fraction from N-body simulations. This $H_D(z)$ enters the 21 cm optical depth as $\\tau\\propto 1/H_D(z)$, so a suppressed expansion history directly deepens the absorption feature.","core_discovery":"The central claim is that the global 21 cm brightness temperature at $15\\lesssim z\\lesssim30$ becomes a sensitive function of the present Hubble rate once the averaged expansion of an inhomogeneous universe is used in place of the standard homogeneous Hubble parameter. In the calibrated two-domain model, the overdense region is assigned a Friedmann-like scaling $n_2=-2$ with vanishing backreaction, while the underdense region uses $n_1=-1.7$ for the simultaneous scaling of backreaction and curvature; this choice reproduces the simulated growth of the overdense volume fraction. The best-fit effective matter density from the supernova fit is $\\Omega_{D0}^m = 0.104\\pm0.056$. Because the 21 cm optical depth scales as $\\tau\\propto 1/H_D(z)$, the lower expansion history associated with $H_0=67$ deepens the absorption trough to about $-320$ mK at $z\\approx18$, compared with about $-280$ mK for $H_0=73$, and shifts the minimum by $\\Delta z\\sim3$. The paper emphasizes that this $H_0$ dependence is absent in $\\Lambda$CDM, where the same parameter change shifts $T_{21}$ by a few mK at most, and that a three-domain extension with an intermediate-density ambient region modifies the trough's depth and position but preserves the $H_0$ dependence.","pith_inferences":["Editorial inference: the same averaged $H_D(z)$ that deepens the 21 cm trough also changes angular-diameter distances, so a joint fit of this class of models to supernovae, baryon acoustic oscillations, and the 21 cm spectrum could test whether one backreaction history resolves the Hubble tension across all probes simultaneously.","Editorial inference: the quantitative depths are computed by substituting the averaged expansion into homogeneous formulas; a light-cone radiative-transfer calculation using local gas densities and spin temperatures inside voids and clusters would show whether the 40 mK gap survives line-of-sight averaging, and is a natural next step.","Editorial inference: if future experiments detect a deep trough at $z\\approx18$ with the predicted $H_0$ dependence, competing explanations such as baryon–dark matter cooling or an excess radio background would still need to be excluded, since they can also deepen the 21 cm absorption feature.","Editorial inference: the preferred scaling exponent $n_1=-1.7$ for underdense regions could be calibrated further with void statistics and cosmic shear, providing an independent test of the model that does not rely on 21 cm data."],"forward_implications":["A global 21 cm spectrum covering $z\\simeq15$–$30$ can in principle distinguish low from high $H_0$ values in this model, because the predicted trough is about 40 mK deeper and shifted by $\\Delta z\\sim3$ for the lower value.","The absence of this gap in standard $\\Lambda$CDM means a measured $H_0$-dependent trough would provide the claimed observational signature that averaged inhomogeneities, not only dark energy, shape the background expansion at cosmic dawn.","The calibrated model prefers an effective matter density near $\\Omega_{D0}^m\\simeq0.104$, noticeably below the standard value; if the model is right, low-redshift matter-density probes should converge to similarly low values.","Because the three-domain extension preserves the qualitative result, the signature is robust to at least one natural generalization of the domain partitioning.","The predicted trough position lies in the band targeted by global 21 cm experiments, so forthcoming observations can confront the model without needing high angular resolution."],"supporting_citations":[{"why":"Supplies the averaged Einstein equations and the definition of kinematical backreaction that the model builds on.","marker":"[48]"},{"why":"Defines the two-domain scaling solution, the volume-fraction method, and the N-body comparison procedure used to select the scaling exponent.","marker":"[62]"},{"why":"Provides the two-region overdense/underdense gravitational model whose partitioning the paper adopts.","marker":"[80]"},{"why":"Supplies the type Ia supernova dataset used for the MCMC calibration of the effective matter density.","marker":"[90]"},{"why":"Provides the N-body structure-formation data used to test the evolution of the overdense volume fraction.","marker":"[91]"},{"why":"Supplies the covariant distance-redshift relation used to compute distance moduli for the supernova fit.","marker":"[50]"},{"why":"Supplies the 21 cm brightness temperature and optical depth expressions used to compute $T_{21}$.","marker":"[93]"},{"why":"Supplies the Ly$\\alpha$ heating and cooling terms and the injection fraction $r=0.1$ used in the thermal evolution of the baryon temperature.","marker":"[87]"},{"why":"Provides the CMB-based $H_0\\approx67$ km/s/Mpc value used as the low Hubble constant baseline.","marker":"[7]"},{"why":"Provides the distance-ladder $H_0\\approx73$ km/s/Mpc value used as the high Hubble constant baseline.","marker":"[4]"}],"fun_headline_variants":["Cosmic voids deepen 21 cm absorption at low H0","Inhomogeneity makes 21 cm dip a Hubble ruler","H0=67 digs a deeper 21 cm trough than H0=73","21 cm signal could break Hubble tension deadlock","Clumpy universe sharpens 21 cm Hubble probe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The calculation assumes that substituting the averaged expansion rate of the backreaction model into the standard homogeneous 21 cm brightness-temperature formula gives the true global signal, even though the real signal is built from local gas densities and radiative transfer through dense and empty regions.","fun_headline_variants_meta":{"raw":{"variants":["Cosmic voids deepen 21 cm absorption at low H0","Inhomogeneity makes 21 cm dip a Hubble ruler","H0=67 digs a deeper 21 cm trough than H0=73","21 cm signal could break Hubble tension deadlock","Clumpy universe sharpens 21 cm Hubble probe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000581,"raw_usage":{"total_tokens":2781,"prompt_tokens":1035,"completion_tokens":1746,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":1662}},"tokens_in":651,"tokens_out":1746,"duration_ms":12701,"temperature":1.0,"reasoning_tokens":1662,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:13:10.920977+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same two-domain backreaction model through a full 21 cm light-cone radiative-transfer code that computes the optical depth region by region from local gas density, spin temperature, and peculiar velocity gradients; if the predicted gap of about 40 mK between $H_0=67$ and $H_0=73$ disappears or reverses, the central claim is falsified.","supporting_citations":[{"cited_title":"Buchert, M","cited_arxiv_id":null,"evidence_quote":"Defines the two-domain scaling solution, the volume-fraction method, and the N-body comparison procedure used to select the scaling exponent."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the two-region overdense/underdense gravitational model whose partitioning the paper adopts."},{"cited_title":"Haario, M","cited_arxiv_id":null,"evidence_quote":"Supplies the type Ia supernova dataset used for the MCMC calibration of the effective matter density."},{"cited_title":"Haario, E","cited_arxiv_id":null,"evidence_quote":"Provides the N-body structure-formation data used to test the evolution of the overdense volume fraction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the 21 cm brightness temperature and optical depth expressions used to compute $T_{21}$."},{"cited_title":"Chuzhoy and P","cited_arxiv_id":null,"evidence_quote":"Supplies the Ly$\\alpha$ heating and cooling terms and the injection fraction $r=0.1$ used in the thermal evolution of the baryon temperature."}],"review_version":1}