{"id":"f85648fc-156b-405d-bf10-1b6bb2aa607f","arxiv_id":"2608.04078","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"The team confirmed 39 new high-redshift radio quasars and found that the number of X-ray-selected blazars at z>4 grows with redshift faster than a constant X-ray/radio ratio model, in broad agreement with the IC/CMB model.","lead":"Astronomers used X-ray, radio, and optical surveys to find new quasars from the early universe. They confirmed 39 distant quasars, including 14 with redshift above 4, and used them to test a popular model of how blazars' X-ray brightness evolves with cosmic time.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main rule-out is not threatened by follow-up incompleteness (unobserved candidates can only add counts); it rests on the absolute normalization of the constant-X/R prediction from the Mao et al. RLF extrapolated to z>4, whose uncertainty is not propagated in Fig. 6.","rationale":"The Reader's CONDITIONAL verdict is appropriate. I disagree, however, that the most load-bearing weakness is the transfer of completeness and blazar fraction from observed to unobserved candidates: that correction can only move the observed counts upward, which strengthens the rejection of constant X/R. The true weak point is that the comparison model's absolute normalization is treated as fixed. The paper is transparent about some model uncertainties (A0, photon index) and about the possibility of super-Eddington contamination, but it does not quantify the uncertainty in the Mao et al. RLF at z>4, which sets the constant-X/R expectation. The independent support is substantial: 39 confirmed quasars, 14 new at z>4, a 65% population increase, and classification consistency with Sbarrato et al. for 9 of 10 objects. The model conclusion should be presented as conditional on a propagated RLF uncertainty, so I keep the CONDITIONAL verdict rather than moving to ACCEPT or REJECT.","tokens_in":31327,"tokens_out":14408,"duration_ms":137422,"concrete_test":"Recompute the constant-X/R N(>z) prediction in Fig. 6 after perturbing the Mao et al. (2017) RLF parameters within their reported 1-sigma covariance, and repeat with an independent z>4 radio-quasar luminosity function (e.g., Gloudemans et al. 2022 or Caccianiga et al. 2024), keeping the X/R distribution and eRASS sensitivity map fixed. If N_const(>4) rises from about 5 to about 11 or higher, the observed 26 sources lie within about 3 sigma of the non-evolving model and the rule-out claim fails; if N_const(>4) stays below about 8 for all reasonable RLF choices, the conclusion is robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the rule-out of a non-evolving X/R ratio (Sec. 5.2, Fig. 6). The Reader's flagged incompleteness transfer is not the load-bearing issue for this claim: the 22 unobserved candidates can only add z>4 blazars, raising the cumulative observed counts, or leave them unchanged; they cannot reduce the observed 26. The load-bearing quantity is instead the absolute prediction of the constant-X/R model, N_const(>z), which is obtained in Fig. 6 by integrating the Mao et al. (2017) radio luminosity function to z>4 and combining it with the local X/R distribution (Eq. 1 with A0=0). The shaded model band varies only the photon index (1.3-1.7) and A0, not the RLF itself. If the Mao et al. normalization or faint-end slope is uncertain by a factor of about 2-3 at z>4, the claimed ~5x excess (observed 26 vs predicted ~5 at z=4-4.3) drops to below about 2-3 sigma. Without a propagated RLF uncertainty, the rule-out is not yet demonstrated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a sample of high-redshift radio quasars by cross-matching eRASS:1 X-ray sources with RACS-mid radio and DELVE optical/NIR data. Applying Ly-alpha dropout criteria to the X-ray/radio-selected catalog yields 68 quasar candidates; spectroscopic follow-up of 46 targets confirms 39 quasars at z>3.5, of which 14 are at z>4. Using the alpha_ox<1.355 criterion, the authors identify 26 z>4 blazars in the combined new and literature sample. They then compare the cumulative number of z>4 blazars with predictions from a constant X-ray-to-radio ratio model and a fractional IC/CMB evolution model, concluding that the constant scenario is ruled out while the IC/CMB model is broadly consistent with the data.","tokens_in":31559,"tokens_out":8124,"duration_ms":74323,"significance":"If the central model comparison holds, this is the largest X-ray-selected statistical sample of z>4 blazars to date and provides one of the cleanest tests of IC/CMB-driven X-ray evolution at high redshift. The paper's strengths are its careful sample construction, the extensive spectroscopic campaign with discovery spectra in Appendix A, the explicit discussion of incompleteness, and the new sample itself: the 14 new z>4 radio quasars represent an approximately 65% increase in the known population in the survey area. The main weakness is that the headline rule-out of the constant X/R scenario rests on an absolute model prediction whose dominant systematic uncertainties are not propagated into the comparison.","major_comments":[{"comment":"The constant X/R prediction shown as the dashed blue curve in Fig. 6 is an absolute count obtained by integrating the Mao et al. (2017) radio luminosity function at z>4 and combining it with the local X/R distribution. The shaded band around the curve varies only the photon index (Gamma_X=1.3-1.7) and the A0 parameter; it does not include uncertainties in the Mao et al. RLF normalization, its faint-end slope, or the local X/R distribution parameters (log(X/R)=1.95, sigma=0.35). Because the claimed exclusion is based on comparing the observed 26 counts with a prediction of roughly 5 at z=4-4.3, an unpropagated factor-of-two-to-three systematic uncertainty in the z>4 RLF could reduce the significance of the rule-out substantially. Please propagate these uncertainties, for example by varying the Mao et al. parameters within their quoted errors or by adopting an empirical envelope from the local X/R distribution, and report the resulting significance for excluding the constant model.","section":"Section 5.2, Eq. (1), Fig. 6"},{"comment":"The only quantitative significance quoted in the model comparison, approximately 2.6 sigma, refers to the discrepancy between the data and the IC/CMB best-fit model, not to the exclusion of the constant X/R model. The statement 'we can rule out a scenario where the X/R ratios in blazars does not evolve as a function of redshift' is therefore not backed by a stated test statistic for the constant model. Please provide the formal significance of the exclusion, including both Poisson and propagated systematic uncertainties, and adjust the abstract and conclusions if that significance falls below the threshold implied by the word 'rule out'.","section":"Section 5.2, Sec. 6"}],"minor_comments":[{"comment":"The text reads 'we confirmed the high-z nature of 397, 14 of which are at z>4'; the number should be 39, not 397.","section":"Section 3"},{"comment":"'The high faction of confirmed candidates' should be 'the high fraction of confirmed candidates'.","section":"Section 3"},{"comment":"The phrase 'aoresecure classification' appears to be a typo for 'a more secure classification'.","section":"Section 5.1"},{"comment":"The caption states that the upper limit at z~5.5 is based on the two blazars in the Russian half of the sky, while the text identifies the z=5.7 upper limit with the non-confirmation of any i-dropout candidates and the Russian-half sources as a lower limit. Please make the caption consistent with the text.","section":"Fig. 6 caption"},{"comment":"The completeness correction in Fig. 6 uses dropout completeness fractions (90%, 76%, 58%) derived from all known z>4 quasars in DELVE, not specifically from the X-ray/radio-selected subset, and the approximately 30% identification efficiency is based on the 46 brighter observed targets. Please state explicitly how these corrections are applied to the literature-known sources, and discuss the systematic uncertainty this choice introduces into the corrected counts. Since unobserved candidates can only add counts, this does not threaten the exclusion of the constant model, but it does affect the claimed shape of N(>z) and the comparison at z>4.7.","section":"Sections 2.2 and 5.2"}],"recommendation":"major_revision","confidential_remarks":"The model parameters (log(X/R)=1.95, sigma=0.35, log(A0)=-2.74, sigma=0.66) and the blazar classification threshold (alpha_ox=1.355) both come from the authors' prior work. The eRASS sample is new data not used to fit those parameters, so this is not circular, but the manuscript would benefit from an independent calibration or at least a clear statement of the provenance of these inputs. The central issue is that the RLF normalization uncertainty is not propagated; if the authors can provide a systematic error budget for the constant-X/R prediction, the conclusion may be robust, but as written the strength of the rule-out claim is not yet demonstrated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper delivers something real — 39 newly confirmed z>3.5 radio-emitting quasars, 14 at z>4, from the eRASS/RACS/DELVE combination, with discovery spectra in the appendix. That is a substantial observational addition and a 65% increase in the known z>4 radio quasar population in the surveyed area. The statistical comparison to IC/CMB models is the weakest part, and the decisive problem is not the one the reader flagged.\n\nThe selection and spectroscopy are careful. Cross-match radii are derived from simulations, dropout completeness is measured against known z>4 quasars (90/76/58% for g/r/i drops), the four missed known sources are accounted for with specific explanations, and the SED screening against low-z ellipticals is sensible. The 85% confirmation rate is credible. The catalog with radio, optical, and X-ray luminosities and alpha_ox will be useful. Blazar classification via alpha_ox < 1.355 is a single threshold from earlier work, but the cross-check with radio loudness and with Sbarrato et al. for 10 overlapping sources makes it acceptable for a statistical sample.\n\nThe model comparison in Sec. 5.2 is where I would push back. The claim to rule out a constant X/R ratio rests on the ~5x excess at z=4–4.3. The reader's concern about transferring the ~30% blazar fraction to unobserved candidates is real but not fatal for that claim: unobserved candidates can only add blazars, so the observed counts would go up, not down. The load-bearing issue is the absolute normalization of the prediction. In Fig. 6, the prediction integrates the Mao et al. radio luminosity function and assumes it holds at z>4, where it is an extrapolation. The shaded band varies photon index and A0, but not the RLF normalization or faint-end slope. If those carry a factor-of-two or three uncertainty — not an unlikely amount for an extrapolated LF — the discrepancy drops to below 2–3 sigma. That uncertainty needs to be propagated before the word \"rule out\" is warranted.\n\nA related caveat: the model parameters come from the same group's prior papers. That is not circularity in the strong sense, because the eRASS sample is new data used for comparison rather than fitting. Still, the sentence that the discrepancy reduces when model uncertainties are included is doing a lot of work, and those model uncertainties are not fully mapped.\n\nBottom line: this deserves serious refereeing. The observational core is strong, the catalog is a contribution, and the model section is fixable by a more honest treatment of the prediction band. The paper is for the high-redshift blazar/AGN population crowd, and the referee should focus on Sec. 5.2 and the RLF error budget. If that is addressed, I would be happy to see it in MNRAS.","headline":"The spectroscopic sample is a real contribution; the claimed rule-out of a non-evolving X-ray-to-radio ratio is plausible but under-supported because the prediction rests on an extrapolated radio luminosity function with unpropagated uncertainty.","tokens_in":32199,"tokens_out":2647,"would_cite":true,"duration_ms":24499,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A new eRASS-based census finds five times more X-ray-bright blazars at z>4 than a non-evolving jet model predicts, supporting the IC/CMB mechanism.","keywords":["high-redshift quasars","blazars","eROSITA","X-ray surveys","radio-loud AGN","IC/CMB","relativistic jets","Lyman dropout"],"falsifier":"A complete spectroscopic census of the 22 unobserved candidates: if fewer than about 4-5 of them turn out to be z>4 blazars, or if they concentrate at z<4, the fivefold excess at z=4-4.3 shrinks and the claimed rejection of the constant X-ray-to-radio model weakens.","tokens_in":31104,"feed_emoji":"🔭","tokens_out":5123,"duration_ms":40637,"temperature":0.7,"pith_summary":"Combining X-ray detections from the first eROSITA all-sky survey with radio and optical/NIR photometry, the paper builds a sample of 68 high-redshift quasar candidates, spectroscopically confirms 39 at z>3.5, and adds 14 new radio quasars at z>4, a 65% increase in the known population in the surveyed area. The central claim is that the completeness-corrected number counts of the 26 X-ray-bright blazars in this sample rule out a scenario in which the X-ray-to-radio luminosity ratio of blazars does not evolve with redshift: counts at z=4-4.3 are about five times larger than the constant-ratio model predicts. The counts are broadly consistent with the fractional IC/CMB model, in which cosmic microwave background photons are inverse-Compton scattered by jet electrons, with an energy density scaling as (1+z)^4. If correct, this confirms that the apparent X-ray brightening of high-redshift blazars is a real population-level effect rather than a selection artifact, and it strengthens the idea that jets at z>4 efficiently upscatter CMB photons.","feed_headline":"26 blazars at z>4 rule out non-evolving X-ray jets","feed_subtitle":"A 14,000-square-degree census finds five times more X-ray-bright blazars than a constant X-ray-to-radio model predicts.","key_machinery":"The load-bearing machinery is the comparison between the observed cumulative number of X-ray-selected blazars, N_{\\rm blazar}(>z), and the prediction of a fractional IC/CMB model. In the model, the X-ray-to-radio luminosity ratio evolves as L_X/L_R(z) = (L_X/L_R(0)) [(1-A_0) + A_0(1+z)^4], where A_0 is the fraction of X-ray emission produced by inverse-Compton scattering of CMB photons at z=0; the mock blazar population is generated by integrating the radio luminosity function of blazars and then detecting it with the eRASS:1 sky-dependent sensitivity. The observed counts come from the roughly 30% blazar fraction among spectroscopically confirmed targets, applied to the candidates not yet observed.","core_discovery":"Using the eRASS:1 X-ray catalogue matched to RACS-mid radio and DELVE optical/NIR data, the authors select 68 z>4 quasar candidates via Lyman-dropout colour criteria and obtain spectra for 46 of them. Of these, 39 are confirmed at z>3.5, 14 at z>4. Using the X-ray-to-UV intensity ratio \\tilde{\\$\\alpha$}_{\\rm ox}<1.355 to identify blazars, they find 26 blazars at z>4. Comparing the redshift distribution of these blazars, corrected for selection completeness (90%, 76%, and 58% dropout recovery for the g-, r-, and i-dropouts), with the predictions of a fractional IC/CMB model built on the radio luminosity function of blazars, they conclude that a constant X-ray-to-radio ratio is ruled out (by a factor of about five at z=4-4.3) and that the data are broadly consistent with the IC/CMB model, although with a residual about 2.6 $\\sigma$ excess of z>4 blazars and an overprediction at z>4.7.","pith_inferences":["If the IC/CMB enhancement is real, deeper eRASS scans and wider-area surveys should find even more X-ray-bright blazars at z>5.5, where the (1+z)^4 boost becomes extreme; the current upper limits at z>5.7 are still based on only two sources.","The residual excess of z>4 blazars over the IC/CMB best fit, if it survives completeness corrections, could point to additional X-ray production from super-Eddington accretion or from a redshift-dependent jet composition, alternatives the paper explicitly leaves open.","The ~65% increase in known z>4 radio quasars implies that simple extrapolations from the current census underestimate the true space density; a full-spectroscopic census of the remaining 22 candidates would test whether the corrected counts at z~4 are indeed as high as reported.","The blazar classification relies entirely on the \\tilde{\\alpha}_{\\rm ox} threshold; adding VLBI brightness-temperature measurements to a subsample would test whether the X-ray-selected excess really corresponds to relativistic boosting rather than to a softer spectral component."],"forward_implications":["The 14 new z>4 radio quasars raise the known population in the eRASS:1/DELVE/RACS area by about 65%, showing that the high-redshift radio-quasar census is far from complete.","A constant X-ray-to-radio ratio for blazars is ruled out: completeness-corrected counts at z=4-4.3 are about five times higher than that model predicts.","The observed counts are broadly consistent with the fractional IC/CMB model, supporting inverse-Compton scattering of CMB photons as the dominant source of the extra X-ray emission at high redshift.","The model overpredicts the number of blazars at z>4.7, suggesting that the redshift evolution of the X-ray enhancement may not follow the simple (1+z)^4 scaling at the highest redshifts.","The 26 blazars at z>4 form a sample for future VLBI and X-ray follow-up to constrain jet orientation and accretion properties in the early Universe."],"supporting_citations":[{"why":"Provides the fractional IC/CMB model parameter distributions (log X/R, A0) and the mock-detection method used for the model comparison.","marker":"Ighina et al. (2021)"},{"why":"Supplies the radio luminosity function of blazars from which the mock blazar population is built.","marker":"Mao et al. (2017)"},{"why":"The eRASS:1 X-ray catalogue and its sky-dependent sensitivity, used for both source selection and detection simulation.","marker":"Merloni et al. (2024)"},{"why":"DELVE optical/NIR photometry and the point-like EXTENDED_CLASS parameter used to define the Lyman-dropout selection.","marker":"Drlica-Wagner et al. (2022)"},{"why":"RACS-mid radio catalogue at 1.37 GHz used for the X-ray-to-radio cross-match and radio-loudness measurements.","marker":"Duchesne et al. (2024)"},{"why":"Defines the \\tilde{\\alpha}_{\\rm ox}<1.355 classification threshold used to identify blazars in the sample.","marker":"Ighina et al. (2019)"},{"why":"Provides the comparison quasar sample for the \\alpha_{\\rm ox}-L_{2500} plane, showing where the selected sources sit relative to the general quasar population.","marker":"Lusso et al. (2020)"}],"fun_headline_variants":["26 z>4 blazars rule out constant X-ray jets","eRASS census: 26 blazars at z>4, static X-ray ratio fails","High-z blazar sample of 26 rejects constant X-ray-to-radio","X-ray-bright blazars at z>4: IC/CMB model wins over static"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The completeness corrections and the roughly 30% blazar fraction measured on the spectroscopically confirmed targets are assumed to transfer to the 22 candidates that were not observed and to the dropout-incomplete redshift bins; if those unobserved candidates have different redshifts or a different blazar fraction, the completeness-corrected number counts change.","fun_headline_variants_meta":{"raw":{"variants":["26 z>4 blazars rule out constant X-ray jets","eRASS census: 26 blazars at z>4, static X-ray ratio fails","High-z blazar sample of 26 rejects constant X-ray-to-radio","X-ray-bright blazars at z>4: IC/CMB model wins over static"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00037,"raw_usage":{"total_tokens":2024,"prompt_tokens":1026,"completion_tokens":998,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":907}},"tokens_in":642,"tokens_out":998,"duration_ms":8926,"temperature":1.0,"reasoning_tokens":907,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:43:25.562442+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A complete spectroscopic census of the 22 unobserved candidates: if fewer than about 4-5 of them turn out to be z>4 blazars, or if they concentrate at z<4, the fivefold excess at z=4-4.3 shrinks and the claimed rejection of the constant X-ray-to-radio model weakens.","supporting_citations":[],"review_version":1}