{"id":"e7687d33-5f65-496f-a873-c94d90926005","arxiv_id":"2607.27352","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"Transiting brown dwarfs show a systematic, model-grid-independent radius inflation trend: the larger the observed radius, the more evolutionary models under-predict it.","lead":"A comparison of 31 transiting brown dwarfs against four evolutionary models finds that larger observed radii are increasingly under-predicted, with mean radius excesses of 5–14% and an intrinsic scatter of 8–12%. This suggests brown dwarf cooling models are missing something in the atmospheric boundary condition, and the effect would bias masses of directly imaged exoplanets if it extends to planet masses.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 8–12% intrinsic dispersion is likely an artifact of the radius-dependent trend: Equation (1) fits a constant mean, so the trend is absorbed into σ_int. Recomputing σ_int after detrending is required.","rationale":"The paper's headline includes both a shallow R_pred–R_obs slope and an intrinsic dispersion of 8–12%. The slope is a direct regression result; the dispersion is the basis for the claim that object-to-object scatter is physical. Equation (1) is the entire evidence for that dispersion, and it is misspecified: it assumes a constant mean even though the same paper demonstrates a radius-dependent mean. This is not a subtle systematic—it is a structural mismatch between the two statistical analyses. If σ_int is recomputed after detrending and vanishes, the paper loses its 'property of the population' interpretation, even if the slope remains. I therefore regard this as more load-bearing than the reader's age/metallicity concern, which is an external assumption that the paper explicitly acknowledges and that can be mitigated by sensitivity tests. The detrending issue is internal and can be settled immediately with existing data. That said, the presence of this flaw does not by itself refute the slope claim; it requires the authors to revise the dispersion analysis. Hence the verdict remains CONDITIONAL—the paper is not ready as is, but the concern is addressable. The age/metallicity assumption remains a secondary caveat. The proposed concrete test is computational and decisive.","tokens_in":22588,"tokens_out":16766,"duration_ms":171495,"concrete_test":"Re-estimate σ_int using Equation (1) but with r_i replaced by residuals from the best-fit linear trend r_i = a + b (R_obs,i − mean) for each grid, or equivalently fit the R_pred–R_obs relation with an intrinsic scatter term and compare σ_int to zero. If the detrended σ_int drops to ≤3–5% or is consistent with zero at <3σ, the paper's claim of 8–12% intrinsic dispersion is not supported. This requires only the existing Table 2 data and the same linmix/likelihood code.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3 establishes that R_pred vs R_obs has slope 0.3–0.4, so the ratio r_i = R_obs,i/R_pred,i increases systematically with R_obs. Section 3.2 then estimates the intrinsic dispersion using Equation (1), which models r_i as a Gaussian with a single mean μ and an added intrinsic variance σ_int^2, with no trend term. Because μ is constant, the systematic variation of r_i with R_obs is absorbed into σ_int. The reported 8–12% σ_int is therefore not a scatter around the fitted radius trend but largely the amplitude of the trend itself. This also explains why σ_int is similar across grids: the slopes are similar, so the trend-induced scatter is similar. The conclusion that the scatter is 'a property of the population and not the models' (Section 3.2) is unsupported by the analysis as presented. A correct estimate must first remove the trend (e.g., fit r_i = a + b R_obs,i + ε_i and estimate σ_int on ε_i) or fit R_pred vs R_obs with an intrinsic-scatter term. This is an internal statistical issue, independent of external assumptions about BD ages and metallicities.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper assembles a sample of 31 field-age (≥0.5 Gyr), weakly irradiated (Teq < 2000 K) transiting brown dwarfs with published masses, radii, ages, and metallicities. For three field grids (Sonora Diamondback, Sonora Bobcat, ATMO 2020) and, for 23 objects, the M26 irradiated grid, the author computes predicted radii from published masses, ages, and [Fe/H] under the assumption that the brown dwarfs' ages and metallicities equal those of their host stars, and compares with observed radii (17 of which are re-derived from Gaia DR3 SED fits). The headline result is that R_pred versus R_obs has a slope of 0.32–0.44 in all grids, 3.7–8σ below unity; the mean inflation is 5–14%, reaching ~20% at the largest radii. The paper further reports an intrinsic dispersion of 8–12% from a constant-mean likelihood, finds no significant trends with mass, irradiation, separation, tidal heating, or age, and discusses implications for the substellar luminosity problem, M-dwarf inflation, and direct-imaging mass estimates.","tokens_in":22967,"tokens_out":12556,"duration_ms":126133,"significance":"If the slope result is correct, this is a significant population-level constraint on substellar evolutionary models: four independent grids fail in the same radius-dependent way, and the absence of an irradiation trend argues against hot-Jupiter-like heating as the primary mechanism. The comparison is not circular: R_pred is constructed from mass, age, and metallicity without using R_obs. Strengths include the comparison across four grids, the Gaia-based radius re-derivation for a large subset, explicit treatment of asymmetric age uncertainties, and honest discussion of limitations such as heterogeneous literature values and the assumed age/metallicity transfer. The paper also makes concrete, falsifiable predictions for direct-imaging mass biases and for JWST secondary-eclipse tests. However, the intrinsic-dispersion estimate is statistically conflated with the very trend the paper measures, and the age/metallicity transfer assumption is load-bearing and untested. Both issues are fixable with a reanalysis, so I recommend major revision rather than rejection.","major_comments":[{"comment":"Equation (1) models r_i = R_obs,i/R_pred,i as a Gaussian with a single mean μ and an added intrinsic variance, with no dependence on R_obs,i. But Section 3 and Figure 3 establish that R_pred versus R_obs has slope ~0.3–0.4, which implies r_i increases systematically with R_obs (r = R_obs/(m R_obs + c) with m<1). A constant-mean fit therefore absorbs the trend into the fitted σ_int. The reported 8–12% dispersion is thus not a scatter around the fitted radius trend; it is largely the trend's own amplitude. The cross-grid similarity of σ_int is also expected when the slopes are similar and does not by itself indicate a grid-independent population property. To support the claim that most of the observed scatter is physical, the author should fit a trend term (e.g., r_i = a + b R_obs,i + ε_i) and estimate σ_int on the residuals, or include a trend term in the likelihood. The conclusion in Sec","section":"Section 3.2, Eq. (1)"},{"comment":"The predicted radii are computed assuming the brown dwarfs' ages and bulk metallicities equal those of their host stars. The paper notes the risk (migration-induced tidal-heating age reset; super-stellar metallicities for low-mass brown dwarfs) but does not test it. Because R_pred depends strongly on age (dlnR/dlnt ≈ 0.2, Section 4.3) and metallicity (Figure 1), an age offset of ~1 Gyr at 5 Gyr changes R_pred by several percent, comparable to the mean inflation and to the per-object uncertainty. A systematic correlation between the age or metallicity mismatch and R_obs could alter the measured slope and either create or hide part of the dispersion. I request a sensitivity test, e.g., marginalizing over a plausible brown-dwarf age offset, or recomputing slopes and σ_int with [Fe/H] offsets for the ~16% of the sample below 30 M_J. The central claim would be much stronger if the trend survi","section":"Section 2.2; Table 2"}],"minor_comments":[{"comment":"The phrase \"2 to both focus\" appears to have a missing word or formatting artifact; please rephrase.","section":"Section 2, paragraph 2"},{"comment":"The title and Figure 1 caption use \"T rend\" and \"Sonora-Diamondback\" while the text uses \"Sonora Diamondback\"; unify for clarity.","section":"Title and Figure 1"},{"comment":"TOI-2119b has an M26 predicted radius of 1.321 ± 0.894, far more uncertain than any other entry. State the effect of removing this object on the M26 slope (0.44 ± 0.15) and on the cross-grid consistency claim.","section":"Table 2 and Section 3"},{"comment":"The M26 comparison uses 23 objects while the field-grid slopes use all 31. Since the slopes are said to be consistent across grids, it would be cleaner to also report the field-grid slopes on the common 23-object subset.","section":"Section 3"},{"comment":"The notation ΔR_p in Eq. (2) is introduced as a \"fractional radius difference\" but could be defined explicitly (e.g., ΔR_p = R_true/R_model) to avoid confusion with an absolute radius difference.","section":"Section 4.4, Eq. (2)"},{"comment":"The absence of a trend with irradiation is a central argument, but the Teq < 2000 K cutoff limits the dynamic range. A sentence quantifying the 2σ upper limit on the irradiation slope would make the null result more informative.","section":"Section 3.1"},{"comment":"The Zenodo link should include a versioned DOI and a short description of the machine-readable columns, especially for the T_int values used in Figures 4, 9, 10, and 11.","section":"Data availability"}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a serious referee, but the headline number most likely to get quoted is the one that needs fixing. The paper's real contribution is population-level: four independent grids fail in the same radius-dependent way, with fitted slopes of R_pred versus R_obs of 0.3–0.4 and inflation growing to ~20% for the largest objects. That trend is robust across grids and based on a carefully assembled sample of 31 transiting brown dwarfs, with a sensible Gaia DR3 re-derivation of host-star radii. The linmix fitting with uncertainties on both axes is appropriate, and the significance estimates are internally consistent. The circularity concern is not warranted: R_pred comes from mass, age, and metallicity using external grids, with no radius-dependent term fitted from the residuals.\n\nThe first soft spot is internal to the statistics. The intrinsic dispersion in Section 3.2 is estimated from Equation (1), which models r_i = R_obs/R_pred as a Gaussian with a single mean. But the paper just showed that r_i grows systematically with R_obs, so the constant-mean model is misspecified and the trend is absorbed into sigma_int. The reported 8–12% intrinsic dispersion is probably largely the amplitude of the trend itself, not a separate population property. The claim that it is grid-independent and physical is unsupported until the trend is removed, e.g., by fitting r_i = a + b R_obs + epsilon_i and estimating sigma_int on epsilon_i. This is fixable and should be fixed.\n\nSecond, the age and metallicity assumption. The paper explicitly assumes BD ages and metallicities equal host-star values, and acknowledges the risks of tidal-heating resets and super-stellar metallicities in low-mass BDs. Given the strong age sensitivity of R_pred at field ages (dlnR/dlnt ~ 0.2), a subset of younger-than-host BDs could mimic both the mean inflation and the scatter. The paper does not marginalize over this; a sensitivity test with free BD age/metallicity would make the trend much more convincing. Third, heterogeneous literature parameters are a limitation, but the paper is upfront about it and calls for a uniform reanalysis.\n\nThe no-irradiation conclusion is weaker than advertised because the sample is cut at Teq < 2000 K, but within that range the lack of a trend is a fair result.\n\nThis paper deserves a serious referee. The trend is likely real and important for substellar models and directly imaged planets, but the intrinsic-dispersion interpretation needs to be redone before acceptance.","headline":"The radius-dependent trend versus all four grids is a genuine new result and worth refereeing, but the 8–12% 'intrinsic dispersion' is contaminated by the same trend and needs a detrended reanalysis before it can be interpreted.","tokens_in":23390,"tokens_out":3961,"duration_ms":39666},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Transiting brown dwarfs are systematically larger than substellar evolutionary models predict, with the deficit growing with radius.","keywords":["transiting brown dwarfs","radius inflation","substellar evolutionary models","atmospheric boundary conditions","intrinsic dispersion","M dwarf radius inflation","directly imaged planets","irradiation"],"falsifier":"Repeat the comparison allowing each brown dwarf's age and metallicity to be free parameters fitted to its measured radius, or independently constrain ages for a subsample; if the R_pred-versus-R_obs slope and the 8–12% intrinsic dispersion survive, the trend is physical, and if they collapse toward zero, it is an artifact of the host-star assumption.","tokens_in":22498,"feed_emoji":"🪐","tokens_out":5048,"duration_ms":42943,"temperature":0.7,"pith_summary":"This paper tries to establish that the observed radii of field-age, weakly irradiated transiting brown dwarfs deviate from all four major evolutionary model grids in a coherent, radius-dependent way: the larger the brown dwarf, the more its measured radius exceeds the predicted value, by 5–14% on average and up to ~20% for the largest objects. It also claims that the object-to-object scatter about this trend is mostly intrinsic—8–12%—rather than measurement noise, and that this scatter is consistent across grids even though the mean offset varies. Because the deviation does not correlate with irradiation, orbital separation, mass, or age, the paper argues against hot-Jupiter-like stellar heating and instead proposes that current models under-represent the range of atmospheric boundary conditions that control how fast brown dwarfs lose interior heat. A sympathetic reader would care because this is the first population-level test of substellar cooling models using directly measured masses and radii, and it connects to known inflation problems in M dwarfs and to possible mass biases for directly imaged planets.","feed_headline":"Brown dwarfs run 5-14% larger than evolutionary models predict","feed_subtitle":"The mismatch grows with radius and defies irradiation, pointing to missing atmospheric physics in cooling models.","key_machinery":"The central comparison is R_pred versus R_obs for a carefully selected sample, fit with linmix, a regression that accounts for uncertainties in both axes, with pivots at the error-weighted means. The fractional residual R_obs/R_pred is then tested for trends against six system parameters, and a joint-Gaussian likelihood with an intrinsic-dispersion term separates physical scatter from measurement error. The interpretation hinges on the cooling-age logic: at fixed mass, a brown dwarf's radius is set by how fast it radiates interior entropy, so any boundary-condition change that throttles the emergent flux slows cooling and leaves the object larger at a given age.","core_discovery":"On the paper's own terms: comparing the measured radii of 31 transiting brown dwarfs (23 for the irradiated grid) against the Sonora Diamondback, Sonora Bobcat, ATMO 2020, and Mukherjee et al. (2026) predictions, every grid shows the same signature—a fit of predicted versus observed radius with slope 0.3–0.4, shallower than unity at 3.7–8σ. The mean radius excess is 5–14% depending on the grid, rising to ~20% for the largest radii, and the residual scatter contains an intrinsic dispersion of 8–12%, detected at 8–14σ. No significant trend appears with mass, equilibrium temperature, orbital separation, tidal heating rate, or age; only a 1.7σ metallicity hint. The paper concludes the inflation","pith_inferences":["A uniform reanalysis of all known transiting brown dwarfs with a single pipeline—especially one that lets each brown dwarf's age and metallicity float rather than assuming they match the host star—would separate a real boundary-condition effect from a host-star transfer artifact.","The marginal metallicity trend could be sharpened with homogeneous stellar metallicities; if it holds, it would point to metal-dependent cloud or opacity physics in the atmospheres.","JWST secondary-eclipse spectroscopy of a matched pair of inflated and non-inflated brown dwarfs could test the flux-throttling picture directly: inflated objects should show muted molecular features, lower brightness temperatures, or cloud signatures.","If the same flux-throttling mechanism operates in hot Jupiters or directly imaged planets, radius-dependent model biases of this kind could also affect transiting giant-planet radius comparisons, though the paper does not make that claim."],"forward_implications":["All four grids consistently underpredict the radii of the largest field-age transiting brown dwarfs, so the discrepancy is not an artifact of one model.","The 8–12% intrinsic dispersion means current grids cannot reproduce the spread of radii at fixed mass, age, and metallicity; future models should be evaluated against this dispersion.","The lack of an irradiation trend rules out stellar heating as the dominant inflation mechanism, redirecting theory toward atmospheric boundary conditions and flux throttling.","The amplitude of inflation matches M-dwarf radius inflation, suggesting a common missing ingredient in models of fully convective low-mass objects.","If the trend continues below the deuterium-burning limit, luminosity-age masses of directly imaged planets would be overestimated by several-to-ten percent, while spectroscopic-gravity masses would need upward revision by ~20–45%."],"fun_headline_variants":["Brown dwarfs bigger than predicted, irradiation not to blame","Brown dwarf radii up to 20% larger than models for largest objects","Brown dwarf models need a fix: predicted radii 5-14% off","Population of 50+ brown dwarfs reveals systematic radius inflation","No irradiation effect: brown dwarf inflation points to missing physics"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The analysis assumes each brown dwarf's age and metallicity equal its host star's; if a subset of brown dwarfs are younger (e.g., via tidal-heating reset) or more metal-rich than their hosts, the predicted radii move in exactly the direction needed to create the apparent inflation and scatter.","fun_headline_variants_meta":{"raw":{"variants":["Brown dwarfs bigger than predicted, irradiation not to blame","Brown dwarf radii up to 20% larger than models for largest objects","Brown dwarf models need a fix: predicted radii 5-14% off","Population of 50+ brown dwarfs reveals systematic radius inflation","No irradiation effect: brown dwarf inflation points to missing physics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000638,"raw_usage":{"total_tokens":2876,"prompt_tokens":946,"completion_tokens":1930,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":690,"completion_tokens_details":{"reasoning_tokens":1851}},"tokens_in":690,"tokens_out":1930,"duration_ms":13565,"temperature":1.0,"reasoning_tokens":1851,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T08:59:08.827057+00:00","model_set":{"reader":"deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.3.0-daily-deepseek","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"Repeat the comparison allowing each brown dwarf's age and metallicity to be free parameters fitted to its measured radius, or independently constrain ages for a subsample; if the R_pred-versus-R_obs slope and the 8–12% intrinsic dispersion survive, the trend is physical, and if they collapse toward zero, it is an artifact of the host-star assumption.","supporting_citations":[],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}