{"id":"a9d99391-a6ab-4f7a-a0af-f7d93d1b4985","arxiv_id":"2412.03988","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Kinematic distances from the Numerical Action Method agree with HST TRGB distances to about 20 percent (scatter 1.57 Mpc) for Local Volume galaxies, except near Virgo and Coma I.","lead":"This study checks how well kinematic distances, computed from galaxy motions with the Numerical Action Method, match direct Hubble Space Telescope distance measurements for 418 nearby galaxies. It finds the kinematic method is accurate to about 20 percent over most of the sky, except near the Virgo cluster and the Coma I group.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 20% NAM accuracy claim may be in-sample: the ~400 TRGB distances used to construct the NAM model likely overlap the 418 TRGB benchmark galaxies, so an out-of-sample split is needed before the claim is established.","rationale":"The paper is a transparent empirical comparison, but its central claim rests on treating the NAM model and the TRGB benchmark as independent. The paper's own text says roughly 400 TRGB distances were used in constructing the NAM model, and the comparison uses 418 TRGB distances from the same catalog generation. Unless the authors show that the benchmark galaxies were not constraints, the reported scatter is an in-sample fit statistic rather than a predictive accuracy. This is not an internal inconsistency, but it is a correctness risk for the headline '20% accuracy over 90% of the sky.' The reader's conditional verdict already captures this concern, so I do not recommend moving the verdict. A straightforward cross-match and reanalysis on the disjoint subset would settle the issue.","tokens_in":5453,"tokens_out":3530,"duration_ms":36236,"concrete_test":"Cross-match the 430 LVGDB galaxies with the input distance catalog used in Kourkchi et al. (2020) / Cosmicflows-3, identifying which NAM distances were derived without using that galaxy's TRGB distance as a constraint. On the strictly out-of-sample subset (ideally >100 galaxies), recompute <D_NAM - D_TRGB>, sigma_delta, and relative scatter using the same outlier rejection as the paper. If the scatter is significantly larger, or the out-of-sample subset is too small to support the 20% claim, the central conclusion should be downgraded. A complementary check is to repeat the comparison using TRGB distances published after 2020 that were not present in CF3, and to publish the overlap table.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 1 states that in constructing the NAM distance calculator the authors of [8] used 'estimates of the distance of galaxies made by various methods. The main array consisted of high-precision measurements of distances by the TRGB ... with a total number of about 400.' The paper then validates NAM distances against 418 TRGB distances from the LVGDB. These numbers are suspiciously close, yet the paper never demonstrates that the benchmark galaxies were not among the constraints used to build the NAM model. If the same TRGB distances enter both sides, the comparison in Fig. 2 measures the model's in-sample reproduction of its own constraints, not its predictive accuracy for new galaxies. The quoted sigma_delta = 1.57 Mpc and 19% relative error would then be a lower bound, and the conclusion that NAM can substitute for TRGB distances would be overstated. The exclusion of Virgo/Coma I outliers only sharpens the issue: those are exactly regions where the model may have had little or no constraint, but the paper treats them as an external caveat rather than as untested regimes. The phrase 'according to the diagram [8]' (Section 3) is not evidence of out-of-sample status.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper compares kinematic distances from the Numerical Action Method (NAM) model of Cosmicflows-3 (Kourkchi et al. 2020) with high-precision TRGB distances from the Local Volume Galaxy Database for 430 galaxies, and also with group-membership and Tully-Fisher distances. After excluding a small number of outliers identified with the Virgo and Coma I regions, the authors report a mean difference <D_NAM - D_TRGB> = -0.30 ± 0.08 Mpc, a standard deviation of 1.57 Mpc, and a relative error of about 19% after subtracting a 5% TRGB error in quadrature. They conclude that NAM distances are accurate to ~20% for individual galaxies over ~90% of the sky, and to ~15% when group-averaged velocities are used, except near Virgo and Coma I.","tokens_in":5790,"tokens_out":5754,"duration_ms":53159,"significance":"If established, the claim is practically valuable: NAM distances, derived from coordinates and radial velocities alone, could substitute for much more expensive HST TRGB distances for many nearby galaxies, enabling large statistical samples in the Local Volume. The paper is concise, uses a large homogeneous TRGB comparison sample, provides a transparent list of outliers, and cross-checks with two independent distance indicators. The main result is quantitative and directly falsifiable, and the underlying analysis is straightforward. However, the significance hinges on whether the comparison is genuinely out-of-sample with respect to the NAM model construction.","major_comments":[{"comment":"The out-of-sample status of the TRGB comparison sample is not established. The Introduction states that the NAM model of Kourkchi et al. [8] was constructed using about 400 TRGB distances as the main constraint array, and the paper then validates that model against 418 TRGB distances from the LVGDB. The paper never demonstrates that these 418 galaxies are not the same objects used in constructing the model. If the two sets overlap, the scatter of 1.57 Mpc and the resulting 20% accuracy estimate are in-sample fit statistics, not predictive accuracies. The authors must quantify the overlap between the LVGDB TRGB sample and the distance catalog used by Kourkchi et al. [8], and if the overlap is non-empty, recompute the scatter for the subset of galaxies that were not used as NAM constraints, or otherwise show that excluding them does not change the conclusion.","section":"Section 1 and Section 3"},{"comment":"The headline scatter sigma_delta = 1.57 Mpc is derived after excluding 12 of 430 galaxies with |Delta| > 5 Mpc, and the claim of ~90% sky coverage depends on this exclusion. The threshold is post-hoc, and the excluded objects are concentrated in the Virgo infall zone and the Coma I group, which are then described as regions where NAM is inapplicable. To make the accuracy claim robust, the authors should report the scatter for the full sample as well as for the cleaned sample, and should justify the threshold or treat these regions as separate regimes rather than as outliers. As written, the analysis removes exactly the problematic parts of the sky before computing the statistic that is then generalized to 90% of the sky, which risks overstating the reliability.","section":"Section 3, Fig. 2, and Table 1"},{"comment":"The systematic offset <D_NAM - D_TRGB> = -0.30 ± 0.08 Mpc is statistically significant but is not incorporated into the stated accuracy. The paper quotes a relative error of ~19% based on the scatter after subtracting the 5% TRGB error in quadrature, but the offset represents a zero-point bias of about 4% at the mean distance of 8.0 Mpc. A complete accuracy statement should either combine random and systematic terms (e.g., root-mean-square about zero) or provide a zero-point correction and test its stability. Without this, the claim that NAM distances have an accuracy of 20% is not fully quantified.","section":"Section 5"}],"minor_comments":[{"comment":"The abstract says the TRGB comparison uses 418 distances, while Section 2 describes 430 galaxies and Section 3 excludes 12 outliers before the Gaussian fit. Please make explicit in the abstract that the 418 number refers to the cleaned sample after outlier removal.","section":"Abstract"},{"comment":"The phrase 'according to the diagram [8]' is vague; please specify how the NAM distances were obtained (e.g., the Cosmicflows-3 NAM distance calculator) so that the procedure is reproducible.","section":"Section 3"},{"comment":"The statement 'Assuming that the relative error of NAM distances is 15%, we find approximately the same value for the error of mem-distances, 15%' is circular: the 15% NAM error is itself estimated in Section 3, not independently measured. Please present the variance decomposition without assuming the NAM error, or state the assumption explicitly as a limitation.","section":"Section 4"},{"comment":"There are minor typographical issues: 'NG3115' should be 'NGC3115'; 'Cosmic-flow-3' should be 'Cosmicflows-3'; and the table heading 'LVGB' should be 'LVGDB'. Also, the text says '16 galaxies in the table with TRGB distances' but the table appears to contain a different count after adding the four non-TRGB measurements; please reconcile the count.","section":"Table 2 and text"}],"recommendation":"major_revision","confidential_remarks":"The central claim is potentially sound but not yet established because of the likely overlap between the TRGB sample and the NAM model's training data. This is a fixable issue: the authors should identify the overlap using the Cosmicflows-3 input catalog and rerun the validation on a strictly out-of-sample subset, or carefully demonstrate non-overlap. The outlier handling and the unmodeled zero-point offset also need attention. Given the authors' access to the data, a major revision is appropriate rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Key point: the paper compares NAM kinematic distances with 418 HST TRGB distances, finds a 1.57 Mpc scatter, and declares 20% accuracy. But the NAM model in [8] was built using about 400 TRGB distances, and the paper never shows the 418 benchmark galaxies are not the same ones. So the quoted scatter is probably how well the model reproduces its own input distances, not how well it predicts distances for new galaxies. The 20% claim stands only after an out-of-sample test.\n\nCredit where due: the comparison itself is new, the writing is transparent, and the outlier list is useful. The authors correctly identify Virgo and Coma I as regions where NAM fails, and the group-membership and Tully-Fisher comparisons are sensible supporting checks even if none of them is independent of the possible overlap. The 5% TRGB error subtraction and the quoted offsets are clearly explained.\n\nSoft spots, in order. First, the circularity issue above is load-bearing. Section 1 says 'the main array consisted of high-precision TRGB distances ... about 400,' and Section 2 compares to 418 TRGB galaxies from the same LVGDB. Those numbers are too close to ignore. A short check of whether the 418 appear in the Cosmicflows-3 constraint set, or a split-sample validation, would settle it. Without that, the 20% is an upper bound on accuracy, not the claimed estimate.\n\nSecond, the 20% number comes after removing 12 of 430 galaxies (and 16 of 159 TF galaxies). Removing outliers is defensible if done transparently, and it is here, but the excluded galaxies cluster in Virgo and Coma I. Since those are also the regions where NAM is least constrained, treating them as an external caveat rather than part of the test makes the '90% of the sky' claim optimistic.\n\nThird, the systematic offset <DNAM - DTRGB> = -0.30 Mpc is reported but not modeled. At 8 Mpc mean distance it is a few percent, so minor for most purposes, but it hints at a scale problem in NAM that the paper does not address.\n\nFourth, the machine-readable table is only available on request. For a validation paper that is trying to convince people to use NAM distances, that is an unnecessary barrier.\n\nWho is this for? People who need quick distances for many Local Volume galaxies and cannot get HST time. They will want to read this, but they should not adopt the 20% figure until the overlap question is resolved. I would send it to a referee; the issue is important and fixable. If the authors demonstrate out-of-sample behavior, this becomes a solid reference. As it stands, treat the 20% claim as provisional.","headline":"Useful comparison, but the 20% accuracy claim is probably measuring the NAM model in-sample against the very TRGB distances used to build it.","tokens_in":6260,"tokens_out":2228,"would_cite":false,"duration_ms":20881,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["98.62.Py","98.65.-r","95.85.-e"],"model":"deepseek-v4-flash","headline":"Kinematic distance estimates for Local Volume galaxies hold up at 20% accuracy, except near Virgo and Coma I.","keywords":["Numerical Action Method","kinematic distances","Tip of the Red Giant Branch","Local Volume","peculiar velocity field","galaxy distances","Virgo cluster","Cosmicflows"],"falsifier":"Recompute the NAM model of Kourkchi et al. 2020 while withholding the 418 TRGB distances from its input constraints, then measure the scatter between those withheld TRGB distances and the new model's predicted distances; if the scatter exceeds ~20% (or the mean offset changes substantially), the paper's claimed predictive accuracy is overstated.","tokens_in":5230,"feed_emoji":"🌌","tokens_out":3086,"duration_ms":22631,"temperature":0.7,"pith_summary":"This paper asks whether the Numerical Action Method (NAM), which derives galaxy distances from measured radial velocities plus a model of local gravity, can replace the expensive Tip of the Red Giant Branch (TRGB) technique for nearby galaxies. Comparing 418 TRGB distances from the Local Volume catalog with the NAM kinematic distances, the authors find a small scale offset of -0.30 Mpc and a scatter of 1.57 Mpc, implying about 20% accuracy per galaxy and 15% for group members. They conclude that NAM distances are dependable for roughly 90% of the sky, with the notable exceptions being the Virgo cluster infall zone and the Coma I group. If correct, this makes accurate distance and luminosity estimates for thousands of nearby galaxies cheap and fast, limited mainly by the availability of radial velocity measurements.","feed_headline":"NAM distances for nearby galaxies reach 20% accuracy","feed_subtitle":"The cheap kinematic method beats Tully-Fisher and works over 90% of the sky, except Virgo and Coma I.","key_machinery":"The Numerical Action Method (NAM) is the central mechanism: it reconstructs galaxy orbits and distances by modeling the local peculiar velocity field, using the positions and masses of nearby galaxy groups as attractors, including the influence of the Virgo cluster and the Local Void. The comparison benchmark is the Tip of the Red Giant Branch (TRGB) distance scale, whose ~5% per-galaxy accuracy comes from HST photometry. The paper relies on the published NAM model of Kourkchi et al. 2020, reading off NAM distances from its diagram/calculator, and compares them statistically (mean offset, standard deviation, Gaussian fit) against the TRGB distances in the Local Volume catalog.","core_discovery":"Using 418 galaxies with TRGB distances and excluding 12 strong outliers, the paper finds that NAM-derived distances agree with TRGB distances with a mean difference of -0.30 ± 0.08 Mpc and a standard deviation of 1.57 Mpc. After quadratically subtracting the 5% TRGB error, the NAM distance error is about 19% for individual galaxies, improving to about 15% when the average is taken over the members of a populated group. The systematic offset is consistent with comparisons against group-membership distances and Tully-Fisher distances, and the main failures are confined to the Virgo cluster infall region and the Coma I group, covering about 10% of the sky.","pith_inferences":["A testable extension would be to split the comparison by projected angular distance from the Virgo cluster center and from the Coma I group, to map exactly where NAM scatter exceeds its nominal 20% and where it remains valid at the group-average 15% level.","The claimed accuracy naturally extrapolates to using 21-cm HI velocity surveys (e.g., future wide-area radio surveys) as a cheap pipeline for distances and luminosities of dwarf galaxies, but only outside the two anomalous regions.","One could directly test the scale offset by comparing NAM distances against a larger sample of Cepheid or SN Ia distances that were not used in constructing the NAM model.","The method's demonstrated failure in the Coma I group and the Virgo infall zone suggests that any kinematic reconstruction of local velocities must explicitly model infall onto Virgo and the anomalous negative-velocity flow north of it before its distances can be trusted there."],"forward_implications":["For most galaxies in the Local Volume, distances can be estimated from radial velocities and the NAM model with roughly 20% accuracy, without costly HST observations.","Within galaxy groups containing several measured radial velocities, averaging NAM distances improves the typical error to about 15%.","NAM distances are competitive with or better than Tully-Fisher distances for the Local Volume population, where the paper estimates a 31% Tully-Fisher error.","The known local velocity anomalies (Virgo infall and the Coma I group with its anomalous peculiar velocities, plus the Leo Spur region) must be avoided for NAM distance work, constraining the method's reliable sky footprint to about 90%.","The small systematic offset of -0.30 Mpc implies a mild difference in the distance scale between the NAM model and the TRGB scale, which could affect volume-limited samples of nearby galaxies."],"supporting_citations":[{"why":"Kourkchi et al. 2020 supplies the NAM kinematic distance model from Cosmicflows-3 that the paper reads distances off of.","marker":"[8]"},{"why":"The Local Volume catalog defining the ~1500 candidate galaxies within 11 Mpc supplies the TRGB distances and radial velocities compared here.","marker":"[3]"},{"why":"The online Local Volume database is the actual data source for the TRGB distances, velocities, and group membership used in the comparison.","marker":"[4]"},{"why":"Tully et al. 2008 supplies the local velocity field model with Virgo infall and Local Void expansion that motivates the NAM comparison and explains the anisotropy.","marker":"[5]"},{"why":"Shaya et al. 2017 describes the NAM model evolution/peculiar velocity field used to generate kinematic distances.","marker":"[7]"},{"why":"This work identifies the Coma I group around NGC4278, whose anomalous negative peculiar velocities the paper uses to explain the outlier behavior.","marker":"[14]"},{"why":"Tully-Fisher 1977 supplies the HI linewidth distance estimator used as an independent comparison sample and baseline for the method's error estimate.","marker":"[17]"}],"fun_headline_variants":["Kinematic distances hit 20% precision for Local Volume galaxies","Cheap NAM method maps Local Volume to 20% accuracy","NAM distances for Local Volume galaxies reach 20% accuracy","Local Volume distances via NAM: 20% error, 90% sky coverage","20% distance accuracy for galaxies using NAM, 90% of sky"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 418 TRGB distances used as the benchmark were not used as constraints when the NAM model of Kourkchi et al. 2020 was constructed, so the measured 20% scatter is a genuinely independent test rather than an in-sample fit.","fun_headline_variants_meta":{"raw":{"variants":["Kinematic distances hit 20% precision for Local Volume galaxies","Cheap NAM method maps Local Volume to 20% accuracy","NAM distances for Local Volume galaxies reach 20% accuracy","Local Volume distances via NAM: 20% error, 90% sky coverage","20% distance accuracy for galaxies using NAM, 90% of sky"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000443,"raw_usage":{"total_tokens":2191,"prompt_tokens":843,"completion_tokens":1348,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":1254}},"tokens_in":459,"tokens_out":1348,"duration_ms":11983,"temperature":1.0,"reasoning_tokens":1254,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:52:27.349579+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the NAM model of Kourkchi et al. 2020 while withholding the 418 TRGB distances from its input constraints, then measure the scatter between those withheld TRGB distances and the new model's predicted distances; if the scatter exceeds ~20% (or the mean offset changes substantially), the paper's claimed predictive accuracy is overstated.","supporting_citations":[{"cited_title":"J., 159, 67, 2020","cited_arxiv_id":null,"evidence_quote":"Kourkchi et al. 2020 supplies the NAM kinematic distance model from Cosmicflows-3 that the paper reads distances off of."},{"cited_title":"J., 145, 101, 2013","cited_arxiv_id":null,"evidence_quote":"The Local Volume catalog defining the ~1500 candidate galaxies within 11 Mpc supplies the TRGB distances and radial velocities compared here."},{"cited_title":"Bull., 67, 115, 2012","cited_arxiv_id":null,"evidence_quote":"The online Local Volume database is the actual data source for the TRGB distances, velocities, and group membership used in the comparison."},{"cited_title":"J., 676, 184, 2008","cited_arxiv_id":null,"evidence_quote":"Tully et al. 2008 supplies the local velocity field model with Virgo infall and Local Void expansion that motivates the NAM comparison and explains the anisotropy."},{"cited_title":"J., 850, 207, 2017","cited_arxiv_id":null,"evidence_quote":"Shaya et al. 2017 describes the NAM model evolution/peculiar velocity field used to generate kinematic distances."},{"cited_title":"J., 743, 123, 2011","cited_arxiv_id":null,"evidence_quote":"This work identifies the Coma I group around NGC4278, whose anomalous negative peculiar velocities the paper uses to explain the outlier behavior."},{"cited_title":"and Astrophys., 54, 661, 1977","cited_arxiv_id":null,"evidence_quote":"Tully-Fisher 1977 supplies the HI linewidth distance estimator used as an independent comparison sample and baseline for the method's error estimate."}],"review_version":1}