{"id":"b7173a22-da14-4f02-84d6-78bee902d629","arxiv_id":"2501.07629","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Quasars with the top 20% most massive black holes are 2 to 3 times more likely to host strong radio jets, but only among the top 5% of radio-bright quasars.","lead":"This paper separates the radio emission of 222,782 quasars into star formation and black hole jet components using a Bayesian model, then compares the two components across black hole mass. It finds that only the most massive 20% of black holes, and only among the brightest 5% of radio sources, show a 2 to 3 times higher chance of hosting strong jets, which the authors say resolves earlier contradictory results.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim depends on fixed AGN power-law slope gamma=1.5; if gamma varies with BH mass or the luminosity function bends, the 0.4 dex boost in f could be a spectral-shape artifact rather than a real 2-3x increase in jet likelihood.","rationale":"The central claim of the paper is a ~0.4 dex increase in the AGN normalization f for the top 20% BH-mass bin, interpreted as a 2-3x higher probability of hosting strong radio jets. This interpretation requires that f cleanly measures jet abundance rather than spectral shape. Because f is the integral of a power law with fixed slope γ=1.5, any change in the true slope—either with BH mass or with luminosity—will be absorbed into f. The paper's justification for fixing γ is a visual inspection of fits, which is not quantitative, and the later KS tests are based on classifications derived from the same fixed-γ model, so they do not independently validate the claim. The most direct way to settle this is to fit γ as a free parameter and see if the f boost survives. This is the single most load-bearing concern because if it lands, the headline conclusion changes from a physical enhancement in jet production to a spectral-index effect; if it does not land, the conclusion is supported. The reader's CONDITIONAL verdict is therefore appropriate, and no change to the verdict is needed. The lack of error bars on Figure 5 is a related but secondary issue; it affects the significance statement, not the model interpretation. I agree with the reader's weakest_assumption.","tokens_in":30700,"tokens_out":8210,"duration_ms":83131,"concrete_test":"Re-run the model with γ as a free parameter (e.g., flat prior in 0.8-2.5) for each BH-mass quintile in each M_i-z grid, using the same likelihood, data, and fitting procedure as in Y24. Then compare the marginal posterior of log f between the top quintile and the lower quintiles, and also check whether the γ posteriors are consistent across quintiles at the 95% level. If the ~0.4 dex boost in log f persists after marginalizing over γ and γ is consistent with 1.5 in all quintiles, the fixed-slope assumption is not the cause; if the boost shrinks or γ varies with BH mass, the central claim is not robust to this model assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3.2 the AGN component is modeled as a single power law, P_AGN(L) dL ∝ phi L^{-γ} dL, with γ fixed to 1.5 based on the bright-end extrapolation in Y24. The parameter f, defined as the fraction of sources with AGN luminosity above 10^26 W/Hz, is an integral of this power law, so its inferred value is directly tied to the assumed slope. The observed radio flux density distribution is a convolution of the SF log-normal and this AGN power law; at the faint end, where most sources lie, the AGN power law and the tail of the SF distribution are degenerate. If the true AGN luminosity function steepens at low luminosities or if γ depends on BH mass, the fit can trade off γ against phi (and therefore f) while still matching the data visually. A flatter true γ for the most massive BHs would mimic a higher f even if the number of powerful jets is unchanged. The paper's only check is 'inspection of individual fits' (Section 3.2), which is not a quantitative test. The KS tests in Section 4 use the classifications Leq and Lpl that are themselves derived from the same fixed-γ model, so they do not provide independent confirmation. Hence the central claim that the top 20% BH-mass quasars are 2-3 times more likely to host strong jets is not yet separated from the assumed spectral shape.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper applies the two-component Bayesian model of Yue et al. (2024) to 222,782 SDSS DR16Q quasars with LOFAR LoTSS DR2 150 MHz measurements, binning the sample in M_i-z grids and, within each grid, by BH-mass quintile. The model separates a log-normal star-forming host component from a power-law AGN jet component. The main result is that the AGN normalization f is roughly constant across BH mass except for the top 20% quintile, where log(f/f0) increases by about 0.4 dex, implying a factor 2-3 higher probability of hosting a strong jet at fixed M_i and z. The paper then defines SF-dominated (L<Leq) and AGN-dominated (L>Lpl) classes from the model and uses BH-mass CDF comparisons to argue that traditional RL/RQ definitions mix populations, and it uses stacked Mg II spectra and CIV-distance distributions to argue that the result is not driven by outflow-induced BH-mass overestimates.","tokens_in":30987,"tokens_out":6526,"duration_ms":59448,"significance":"If the central claim is correct, the paper offers a quantitative resolution of the long-standing disagreement over BH mass and radio loudness: the dependence is confined to the most massive 20% of BHs and the most radio-bright 5% of quasars. The paper has several concrete strengths: a large, well-defined sample; an explicit physical decomposition rather than a threshold in radio loudness; careful exclusion of problematic CIV-based masses at z>2; a tabulated classification (Leq, Lpl, Table A1) that other studies can apply directly; and a dedicated stacked-spectrum test (Section 5) that addresses a plausible outflow-related bias. The main quantitative claim, however, is currently a readout of a model with a fixed power-law slope and is presented without posterior uncertainties, so its robustness is not yet established.","major_comments":[{"comment":"The AGN component is modeled as P_AGN(L)dL proportional to phi L^{-gamma} dL with gamma fixed to 1.5, and f is defined as the integral above 10^26 W/Hz. Since the fit is performed on the full flux density distribution, a change in gamma with BH mass, or a bend in the AGN luminosity function below the bright end, can be partially absorbed into the normalization phi and hence into f. The paper's only defense is 'inspection of individual fits' in Section 3.2, which is not a quantitative test. I request a free-gamma fit, or at least a systematic sensitivity test with gamma varied across quintiles, together with the joint posterior constraints on gamma and f, so that the 0.4-dex enhancement can be separated from a spectral-shape effect.","section":"Section 3.2, Eq. (4)"},{"comment":"The headline result, namely the 0.4 dex increase in log(f/f0) for the top BH-mass quintile and the corresponding factor 2-3 higher jet probability, is shown in Figure 5 without error bars, credible intervals, or a significance level. Without these, the reader cannot assess whether the top-quintile offset is significant relative to the fit uncertainties and the scatter across the M_i-z grid cells. Please report per-quintile posterior intervals and a combined significance statement, for example a posterior probability or a matched-pair test across the grid cells.","section":"Figure 5; Section 3.2"},{"comment":"The KS tests and CDF comparisons in Section 4 are not independent tests of the central claim, because the classification boundaries Leq and Lpl are derived from the same fixed-gamma best fits used to infer f. They demonstrate internal consistency, but they cannot break the degeneracy between f and gamma identified above. This limitation should be stated explicitly, or the classification should be validated with a procedure that does not rely on the fitted AGN power-law tail, for example a non-parametric definition of the radio-bright tail.","section":"Section 4, Figures 7-9"}],"minor_comments":[{"comment":"The sentence defining Lpl ends with 'so that 95% of .', which is incomplete; please complete the definition.","section":"Section 4"},{"comment":"The caption states 'where the RQ quasars have Rflux < 10 and the RL quasars have Rflux < 10'; the second condition should be Rflux > 10 for RL quasars.","section":"Figure B1 caption"},{"comment":"The quantities labeled '2 = 0.33' and '2 = 0.57' are not defined; please define them in the caption.","section":"Figure 10"},{"comment":"The phrase 'inspection of individual fits' is vague; please provide a quantitative goodness-of-fit statistic for the fixed-gamma model in place of, or in addition to, the visual statement.","section":"Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"The gamma-fixed assumption is the main technical risk; I would ask for a free-gamma or sensitivity analysis before acceptance. The paper otherwise fits the journal's scope, and the stacked-spectrum test in Section 5 is a genuine strength. The authors should also be asked to display posterior uncertainties for the central f-enhancement claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is worth taking seriously. It is not just another radio-loudness correlation: it applies the two-component Bayesian model from Y24 to separate SF and AGN contributions, and then shows that any BH mass dependence is confined to the top 20% of the BH mass distribution and the top 5% of radio-loud quasars. That localization is the genuinely new result, and it is a plausible reconciliation of the McLure & Jarvis / Whittam versus Gürkan / Arnaudova disagreement.\n\nWhat it does well: the model-driven classification using Leq and Lpl is a real step beyond arbitrary RL/RQ thresholds, and the comparison of BH mass CDFs under different definitions (Figure 8) gives a clear, intuitive picture of why the R_lum = 1 luminosity threshold dilutes the signal. The stacking analysis in Section 5 is an independent check that outflow biases are not driving the mass enhancement, and the CIV distance result actually points in the opposite direction: the massive, AGN-dominated quasars are less outflow-dominated, not more.\n\nThe soft spots are real but not fatal. The central number—about 0.4 dex in log f—is presented without posterior uncertainties in Figure 5, and it is a fitted parameter rather than an out-of-sample prediction. The KS tests that give p = 0.0054 use a classification derived from the same fixed-gamma model, so they are not fully independent. The fixed gamma = 1.5 is the main concern. The authors justify it only by visual inspection of the fits. If the AGN power-law slope varies with BH mass or bends at low luminosities, the fitted normalization f could be spectrally degenerate, and the claimed 2–3× boost could be an artifact. That is a legitimate worry because the AGN power-law tail and the faint-end SF component trade off in the convolution. The paper would be much stronger if gamma were left free in at least the extreme mass quintiles, or if the authors showed the boost persists for a range of fixed gamma values. The acknowledged 0.3–0.5 dex scatter in single-epoch masses cross-contaminates the quintile bins, which would weaken the trend, so that concern cuts in the authors' favor.\n\nWho this is for: anyone working on quasar radio loudness, AGN feedback, or the BH mass–radio connection. The paper is well structured, uses a large sample, and gives a physically motivated framework that explains prior contradictory results. The quantitative claim is not yet nailed because of the gamma issue, but the qualitative picture—effect only in the most massive and most radio-loud quasars—may well survive.\n\nRecommendation: send it to peer review. Ask the authors to add error bars to the f evolution, test gamma as a free parameter, and show how Leq/Lpl shift under gamma uncertainty. With those additions, this could become a standard reference for the field.","headline":"A credible unification of the RL/RQ–BH mass debate, but the headline 0.4 dex boost rests on a fixed power-law slope and fitted parameters without error bars; referee it, and ask for a free-gamma test.","tokens_in":31635,"tokens_out":4092,"would_cite":true,"duration_ms":44163,"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":"The paper claims that quasar radio emission links to black hole mass only for the top 20% most massive black holes, where AGN jets are 2 to 3 times more likely to be bright at fixed redshift and luminosity.","keywords":["quasar radio emission","black hole mass","AGN jets","radio loudness","star formation","Bayesian two-component model","LOFAR","SDSS quasars"],"falsifier":"Refit the LoTSS-SDSS radio flux density distributions with gamma as a free parameter in each black-hole-mass quintile; if the best-fit gamma shifts systematically with mass, for instance by more than about 0.1 dex between the lowest and highest quintiles, or if a broken power law fits better at the faint end, the fixed-slope assumption fails and the f boost is not a jet-likelihood effect.","tokens_in":30481,"feed_emoji":"📡","tokens_out":5899,"duration_ms":51338,"temperature":0.7,"pith_summary":"This paper tries to settle a long-running dispute: does the mass of a quasar's supermassive black hole control how bright it is in the radio? The authors apply a two-component Bayesian model that separates each quasar's radio light into star formation in the host galaxy and AGN jet activity, fitting the observed 150 MHz flux density distribution of 222,782 SDSS quasars observed by LOFAR. They find that black hole mass makes no difference to host-galaxy star formation at fixed redshift and bolometric luminosity, and no difference to AGN jet activity for most black hole masses. The exception is the top 20% most massive black holes, whose quasars are 2 to 3 times more likely to be radio-bright, and this excess shows up only in the roughly 5% most radio-loud quasars. If right, the result explains why previous studies disagreed: radio-loud and radio-quiet definitions that mix in star-formation-dominated sources hide the mass dependence, and a physically motivated SF/AGN classification recovers it.","feed_headline":"Quasar radio jets tied to black hole mass only in the top 20 percent","feed_subtitle":"Two-component model of 222,782 LOFAR-observed quasars isolates the effect to the 5% brightest radio sources.","key_machinery":"The carrying mechanism is the two-component Bayesian model of the radio flux density distribution introduced in the companion paper (Y24): the star-formation component is a log-Gaussian centered on the 150 MHz luminosity corresponding to a mean SFR, with scatter, and the AGN component is a power-law luminosity function with fixed slope gamma = 1.5 and normalization phi, expressed as the radio-loud fraction f. Fitting this model inside M_i-z grids and then within black-hole-mass quintiles separates mass-dependent changes in star formation from mass-dependent changes in AGN jet production. The classification thresholds L_eq, where the SF and AGN probability densities cross, and L_pl, where the AGN power law provides 95% of the total PDF, turn the fitted components into a physically motivated radio-quasar taxonomy.","core_discovery":"The paper's central claim is that supermassive black hole mass is not a general driver of quasar radio emission; instead, a specific subpopulation, quasars hosting the top 20% of black hole masses at given redshift and bolometric luminosity, shows an enhanced AGN contribution, with the fitted jet normalization f increased by about 0.4 dex, meaning they are 2 to 3 times more likely to be radio-bright at fixed optical luminosity. The same excess appears when quasars are classified by the physical origin of their radio emission: only AGN-dominated sources, roughly the top 5% in radio loudness, host systematically more massive black holes, while SF-dominated quasars show no mass dependence. The paper further claims that traditional radio-loud definitions contaminate the radio-loud sample with SF-dominated and intermediate sources, diluting or erasing the mass signal, and that a classification based on model-derived thresholds L_eq and L_pl reconciles previously contradictory results.","pith_inferences":["If the fixed-slope power law for the AGN component is correct, the 0.4 dex boost in f implies that jet launching efficiency itself increases with black hole mass at the high-mass end, a testable prediction for very-long-baseline observations of jet cores in this population.","The classification scheme could be applied to radio-selected quasar samples at other frequencies or redshifts to check whether the top-20% mass effect is universal or specific to the LoTSS-selected population.","The lower CIV distance and Eddington ratios of the massive AGN-dominated quasars hint at a distinct accretion state rather than an outflow-driven artifact, so X-ray or polarimetric follow-up of this quadrant could discriminate among accretion-mode scenarios.","Because gamma is fixed, letting the slope vary with black hole mass would provide a direct stress test: if gamma steepens at high mass, part of the claimed boost would migrate from normalization to spectral shape."],"forward_implications":["Quasar host-galaxy star formation is independent of black hole mass at fixed redshift and bolometric luminosity, so radio emission from star formation cannot serve as a black-hole-mass indicator.","AGN jet activity is also mass-independent across most of the black hole mass range, with the mass signal confined to the most massive 20% of black holes.","The radio excess in the most massive quasars affects only the roughly 5% most radio-loud quasars at a given redshift and luminosity, namely the AGN-dominated tail.","Traditional radio-loud and radio-quiet definitions, whether based on flux ratios or luminosity ratios, mix in SF-dominated and intermediate sources, which explains why some studies find no black-hole-mass dependence.","The SF-dominated versus AGN-dominated classification reproduces earlier positive results and erases the mass difference if the top 20% of black hole masses are removed, unifying previously divergent findings."],"supporting_citations":[{"why":"Supplies the fully Bayesian two-component model that this paper extends to black-hole-mass bins.","marker":"Yue et al. 2024 (Y24)"},{"why":"Introduces the two-component SF-plus-AGN decomposition and the radio-loud fraction f that the model fits.","marker":"Macfarlane et al. (2021)"},{"why":"Provides the stacking and PDF methodology for analysing the flux density distribution of individually faint sources.","marker":"Roseboom & Best (2014)"},{"why":"Provides the LoTSS DR2 150 MHz catalogue and mosaics that yield the radio flux density data.","marker":"Shimwell et al. (2022)"},{"why":"Provides the SDSS DR16Q spectroscopic sample with single-epoch black hole masses used for the quintile binning.","marker":"Wu & Shen (2022)"},{"why":"Calibrates the radio luminosity to star-formation-rate conversion that turns the fitted log-Gaussian into an SFR.","marker":"Smith et al. (2021)"},{"why":"The positive earlier result that radio-loud quasars host massive black holes, which the new classification reproduces.","marker":"McLure & Jarvis (2004)"},{"why":"The null result on black hole mass versus radio loudness that the paper explains as contamination by SF-dominated sources.","marker":"Arnaudova et al. (2024)"},{"why":"The LOFAR-based correlation study that found no mass-radio-loudness link and is reconciled by the new SF/AGN classification.","marker":"Gürkan et al. (2019)"},{"why":"Provides the CIV distance definition used to test whether outflows bias the black hole masses.","marker":"Richards et al. (2021)"}],"fun_headline_variants":["Black hole mass drives quasar jets only in the most massive 20%","Quasar radio jets: mass matters only for top 20% black holes","Bayesian model links quasar jets to mass only in top 20%","Top 20% black holes 2-3x more likely to host strong quasar jets","Quasar jet excess limited to most massive black holes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the AGN radio luminosity function is a single power law with slope gamma = 1.5 at all black hole masses and luminosities probed here; if the slope bends at low luminosities or changes with black hole mass, the fitted normalization f would absorb that change and the claimed 2 to 3 times boost would not cleanly measure jet likelihood.","fun_headline_variants_meta":{"raw":{"variants":["Black hole mass drives quasar jets only in the most massive 20%","Quasar radio jets: mass matters only for top 20% black holes","Bayesian model links quasar jets to mass only in top 20%","Top 20% black holes 2-3x more likely to host strong quasar jets","Quasar jet excess limited to most massive black holes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000993,"raw_usage":{"total_tokens":4279,"prompt_tokens":1089,"completion_tokens":3190,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":705,"completion_tokens_details":{"reasoning_tokens":3090}},"tokens_in":705,"tokens_out":3190,"duration_ms":19989,"temperature":1.0,"reasoning_tokens":3090,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:38:03.953669+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Refit the LoTSS-SDSS radio flux density distributions with gamma as a free parameter in each black-hole-mass quintile; if the best-fit gamma shifts systematically with mass, for instance by more than about 0.1 dex between the lowest and highest quintiles, or if a broken power law fits better at the faint end, the fixed-slope assumption fails and the f boost is not a jet-likelihood effect.","supporting_citations":[{"cited_title":"G., Best P","cited_arxiv_id":null,"evidence_quote":"Provides the stacking and PDF methodology for analysing the flux density distribution of individually faint sources."}],"review_version":1}