{"id":"400c8d43-d108-4f1e-83ae-1c1d89c21b32","arxiv_id":"2508.02792","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A survey of 23 mid-IR-selected obscured quasars at z=0.9-3.5 finds broad H-alpha emission in roughly one-third of them, linking them to JWST's little red dots.","lead":"Astronomers measured light from 23 very red, dusty quasars in the distant universe and found that about one-third of them unexpectedly show broad hydrogen emission lines, usually a sign of unobscured quasars. These objects look like the 'little red dots' discovered by the Webb telescope, so they may show how those mysterious sources fit into the growth of supermassive black holes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Broad H-alpha fraction is not robust: single-Gaussian fits ignore [N II] blending at R~1100, and the sample count is inconsistent (abstract 8/23, text 7/21, Table 1 lists 9).","rationale":"The reader's weakest assumption was the Lbol = 6.03 x L15um scaling used in the [O III]-Lbol analysis (Section 5, Eq. 2). That is a legitimate concern for the secondary claim about [O III] weakness, but it does not bear on the paper's central claim about broad H-alpha and the LRD analogy. The central claim rests on the reliability and consistency of the broad-line classification itself. I find that this classification is not robust for two concrete reasons. First, the H-alpha FWHM values come from single-Gaussian fits at R~1100 with no modeling of the unresolved [N II] doublet; narrow H-alpha plus strong [N II] can mimic a line with FWHM near or above the 2000 km/s threshold. Second, the paper reports mutually inconsistent numbers (abstract 8/23, Section 3.3.3 7/21, Table 1 nine entries above threshold, and Section 7 contradicts itself between five and three), and J0152-0024 is moved into the broad group for the composite despite a narrow measured FWHM. These issues do not necessarily falsify the conclusion: the objects with FWHM 3000-4800 km/s are almost certainly genuinely broad, and the fraction is roughly one-third under any of these counts. But before the quantitative headline is taken at face value, the fitting and counting need to be made consistent. The reader's CONDITIONAL verdict remains appropriate, so I recommend UNCHANGED. A reanalysis with [N II] included and a corrected count would settle whether the concern lands.","tokens_in":27757,"tokens_out":5470,"duration_ms":60869,"concrete_test":"Refit all 21 H-alpha detections with a physically motivated model: narrow H-alpha plus unresolved [N II] doublet (wavelength separation fixed, flux ratio 6548/6584 = 1/3) and an optional broad Gaussian component. Use an F-test or BIC to decide whether the broad component is required, then count how many sources have a broad-component FWHM > 2000 km/s. Repeat the count excluding J0152-0024 and with the Table 1 single-Gaussian values. If the fraction is still roughly one-third and consistent across abstract, text, and table, the concern is resolved; if not, the headline fraction must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that approximately one-third of the sample shows broad H-alpha emission (FWHM > 2000 km/s) despite Type-2 classification. This statistic depends entirely on FWHM measurements from single-Gaussian fits to H-alpha in Gemini/GNIRS spectra at R~1100. At this resolution, instrumental broadening is ~270 km/s and the [N II] 6548,6584 doublet is not resolved from H-alpha; the fits do not include [N II]. A single Gaussian over H-alpha plus [N II] can be broadened by several hundred km/s depending on the [N II]/H-alpha ratio, so objects near the 2000 km/s threshold (e.g., J2334+0031 at 2112, J0024-0012 at 2145, J0215+0042 at 2683) may be misclassified. The paper's own bookkeeping is inconsistent: the abstract says 8/23, Section 3.3.3 says 7/21, and Table 1 lists 9 objects with FWHM > 2000 km/s. Section 7 conclusion (i) says five broad H-alpha objects while conclusion (ii) says three. For the composite, J0152-0024 is added to the broad group (FWHM 1382 km/s) because its H-alpha falls at the edge of the GNIRS coverage, a post-hoc reclassification that further shows the broad/narrow boundary is subjective. Because the fraction of broad-line objects is the headline finding, its instability is the most load-bearing weakness.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a Gemini/GNIRS and Keck/LRIS+KCWI spectroscopic survey of 23 luminous WISE W4-selected obscured quasar candidates in SDSS Stripe 82 at z = 0.88–3.49, confirming 23 of 24 candidates as Type-2 or reddened Type-1 quasars. The authors measure Ly-alpha and H-alpha line widths from Gaussian fits, report that roughly one-third of the sample shows broad H-alpha (FWHM > 2000 km/s) despite the Type-2 classification, compare their [O III] luminosities with bolometric luminosities derived from a mid-IR scaling, and construct composite spectra for the full sample and for broad/narrow H-alpha subgroups. They argue that the broad-line obscured objects resemble JWST-discovered little red dots and that the sample bridges the gap between low-redshift obscured quasars and high-redshift JWST AGNs.","tokens_in":28055,"tokens_out":5944,"duration_ms":59181,"significance":"If the central claims hold, this sample is valuable: it is one of the few spectroscopic samples of luminous obscured quasars at cosmic noon, it demonstrates spectral diversity in a color-selected obscured population, and it directly connects z~2 obscured AGNs to the JWST little-red-dot phenomenon. The public release of reduced spectra is a concrete asset. The broad-H-alpha detections themselves are supported by the displayed spectra, and the composite spectra are a useful reference. However, the headline 'one-third broad H-alpha' fraction is not currently robust because of inconsistent bookkeeping across the abstract, text, and tables, and because the FWHM measurements near the 2000 km/s threshold are vulnerable to the ignored [N II] doublet and instrumental resolution. The [O III] comparison also rests on a single assumed mid-IR SED scaling with no propagated uncertainties. These are load-bearing issues for the paper's main conclusions.","major_comments":[{"comment":"The reported fraction of broad H-alpha emitters is inconsistent across the manuscript. The abstract states 8 of 23; Section 3.3.3 states '7/21 targets have broad H-alpha emission lines'; Table 1 lists 9 objects with FWHM > 2000 km/s (J0024-0012, J0130+0009, J0213+0024, J0215+0042, J0221+0050, J2243+0017, J2259-0009, J2329+0020, J2334+0031); and Section 7 conclusions (i) and (ii) give conflicting counts (four versus three objects with broad H-alpha and narrow Ly-alpha). Because the broad-line fraction is the paper's headline claim, the authors must reconcile these numbers, state which denominator (23, 21, or the 11 objects with both lines) is used, and correct the conclusion text.","section":"Abstract; Section 3.3.3; Table 1; Section 7"},{"comment":"The H-alpha FWHM measurements are derived from single-Gaussian fits at R~1100 that do not include the [N II] lambda-lambda 6548, 6584 doublet. The instrumental resolution is roughly 270 km/s at H-alpha, and the [N II] lines are separated from H-alpha by approximately -677 and +941 km/s, so they are unresolved at this resolution. A single Gaussian fit to H-alpha plus [N II] can be broadened by several hundred km/s depending on the [N II]/H-alpha ratio. Three objects lie within ~700 km/s of the 2000 km/s threshold (J2334+0031 at 2112 km/s, J0024-0012 at 2145 km/s, J0215+0042 at 2683 km/s), so their broad/narrow classification is not robust. The fits should include [N II] components with fixed wavelengths and atomic intensity ratios, and the sensitivity of the classification to the fitting model and to instrumental broadening should be quantified.","section":"Section 3.2; Section 2.2.1"},{"comment":"The 'broad H-alpha' composite spectrum in Section 6 includes J0152-0024, whose Table 1 FWHM is 1382 km/s, based on the argument that its H-alpha line falls at the edge of the GNIRS coverage and its measured FWHM may be underestimated. This is a post-hoc reclassification that is not applied consistently to the tabulated line properties, and it changes the composite 'broad' group from a rule-based FWHM>2000 km/s sample to a mixed sample. Since the broad composite is used to support the LRD analogy, the grouping should be defined by a transparent criterion (e.g., Table 1 FWHM) or supported by a re-fit of J0152-0024 with proper treatment of the line profile.","section":"Section 6; Table 1; Section 4.2"},{"comment":"The [O III] deficiency and the '100x brighter' comparison rest on the scalings Lbol = 6.03 x L15um and Lbol = 6.01 x L13.5um, which are derived from the authors' own composite SED and a specific torus template (Stalevski et al. 2016) and then assumed to apply to every target and to the Lacy et al. (2013a) and Zakamska et al. (2016) comparison samples. No uncertainties are propagated into Lbol, and Figure 6 shows no error bars on the comparison points or on the new measurements. The claim that [O III] is weak at fixed Lbol is only as strong as this scaling; the authors should show how the result changes if the mid-IR SED shape varies, or at least provide a quantitative systematic uncertainty on the bolometric correction.","section":"Section 5; Eq. (2); Eq. (4); Figure 6"}],"minor_comments":[{"comment":"The number of objects with both narrow H-alpha and narrow Ly-alpha is given as five in Section 3.3.1 and in the Figure 5 caption, but the text in Section 3.3 says six; please make these counts consistent.","section":"Section 3.3.1; Figure 5 caption"},{"comment":"The conclusion text refers to 'J2334+0041', but the corresponding table and figures list this target as J2334+0031.","section":"Section 7, conclusion (i)"},{"comment":"The reference entry beginning 'S., Hamann F., et al. 2019' is missing the first author's name; this is likely a formatting error that should be corrected.","section":"References"},{"comment":"The sentence 'Since our targets don't have H-beta line detection, we couldn't conduct the extinction correction using .' is incomplete and should be finished or removed.","section":"Section 5"},{"comment":"The instrumental setup description contains unclear notation, including '1.0.′′ long-slit' and '560 dichroic'; please clarify the slit width and dichroic wavelength.","section":"Section 2.2.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript contains a promising sample and genuinely interesting spectra, but the internal inconsistencies in the broad-line fraction are more than typographical: the abstract, table, and conclusions disagree on the headline number, and the FWHM fitting procedure is not adequate at the 2000 km/s threshold. The [O III] section similarly depends on an assumed bolometric scaling without error bars. These issues can likely be fixed within the scope of the paper, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Ben, you should know about this one. The companion SED paper did the selection; this paper adds the spectra. The genuinely new things are: 23 confirmed luminous Type-2 quasars at z=0.88-3.49, the systematic line-width measurements, the finding that a non-negligible fraction show broad H-alpha despite Type-2 selection, the composite spectra split by broad/narrow H-alpha, and the suggested SED analogy to JWST little red dots. The data are public, the redshifts are solid, and the spectral diversity is real. I believe the core finding - that some DOG-selected obscured quasars have FWHM(H-alpha) > 2000 km/s - survives scrutiny.\n\nWhere it gets soft: the headline count is a mess. The abstract says 8/23, Section 3.3.3 says 7/21, Table 1 lists nine objects with FWHM > 2000, and the conclusions contradict themselves (five vs three broad-H-alpha with narrow Ly-alpha). The fitting is single Gaussian with no [N II] doublet at R~1100. At that resolution, [N II] contamination can inflate FWHM by a few hundred km/s; three objects sit within ~700 km/s of the 2000 km/s threshold, so the exact fraction is not robust. The fix is easy - refit with a [N II]+H-alpha model and quote a range - but until they do, I would not take 'one-third' as a precise number.\n\nThe [O III]-Lbol analysis is the weakest section. Lbol comes from Lbol = 6.03 L15um, a scaling from their own Wang et al. (2025) composite and a Stalevski torus template, applied to every target and to the Lacy et al. comparison points. No error bars are propagated. The conclusion that [O III] is weak for their Lbol is interesting but could shift by factors of a few if individual SEDs deviate. This is separate from the broad-H-alpha result, which does not depend on that scaling.\n\nThe LRD analogy is suggestive, not proven. They do not show compactness or variability, but they do note the SED similarity and stop short of overclaiming. Given the small sample, this is a pointer for follow-up rather than a settled link.\n\nBottom line: this deserves a serious referee. The data are real, novel, and mostly well presented; the flaws are tightening, not fatal. I would send it out with a request to fix the count inconsistency, re-fit H-alpha with [N II], and temper the [O III] claims or add uncertainties. The broad-H-alpha population and the LRD connection will be useful to the community even after revision.","headline":"Valuable spectroscopic sample of obscured quasars at cosmic noon, with a suggestive but currently imprecise broad-H-alpha fraction and a model-dependent [O III] comparison.","tokens_in":28649,"tokens_out":4104,"would_cite":true,"duration_ms":44139,"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":"One-third of bright mid-infrared-selected quasars that appear to be Type-2 nonetheless show broad H$\\alpha$ lines, identifying them as lower-redshift analogues of JWST's little red dots.","keywords":["obscured quasars","Type-2 AGN","mid-infrared selection","broad emission lines","little red dots","cosmic noon","composite spectra"],"falsifier":"Measure each of the 13 [O III]-detected objects' bolometric luminosity independently, using X-ray luminosity or per-object torus SED fits, and re-plot $L_{\\rm [O\\,III]}$ against $L_{\\rm bol}$; if the points fall onto the Lacy et al. relation, the reported [O III] weakness is an artifact of the shared SED scaling.","tokens_in":27535,"feed_emoji":"🔭","tokens_out":11840,"duration_ms":125388,"temperature":0.7,"pith_summary":"This paper reports a spectroscopic survey of 23 luminous mid-infrared-selected Type-2 quasar candidates in SDSS Stripe 82 at $z=0.88$ to $3.49$, selected by bright WISE W4 detections and red $r-W4$ colors. It confirms 23 of 24 candidates as genuine obscured quasars and finds a wide range of emission-line widths: roughly one third show broad H$\\alpha$ emission (FWHM $>2000$ km s$^{-1}$) despite the Type-2 classification, while the rest show only narrow lines. These broad-line obscured quasars have spectral energy distributions similar to those of JWST's \"little red dots,\" so the paper argues they are lower-redshift analogues of the heavily obscured broad-line AGNs JWST reveals at higher redshift. It also finds that [O III] is relatively weak for their bolometric luminosities and constructs new composite spectra for the sample. If correct, the paper bridges the gap between well-studied low-redshift obscured quasars and the JWST-discovered obscured AGN population, bearing on whether the obscured fraction of supermassive black hole growth evolves through cosmic noon.","feed_headline":"One-third of mid-IR-selected hidden quasars show broad H-alpha lines","feed_subtitle":"These cosmic-noon objects resemble JWST's little red dots, closing the gap in obscured black-hole growth.","key_machinery":"The load-bearing machinery is the selection-and-confirmation chain. First, a single color cut in the SDSS Stripe 82 field, $r-W4>8.38$ AB with a bright WISE 22 $\\mu$m detection ($f_{W4}>5$ mJy), isolates candidates whose optical light is extremely faint or red. Then deep rest-frame UV and optical spectroscopy confirms the objects as quasars and measures line profiles: Gaussian fits to Ly$\\alpha$ and H$\\alpha$ yield the FWHM that decides \"broad\" versus \"narrow\" at 2000 km s$^{-1}$. For the [O III] analysis, the paper converts W4/15 $\\mu$m luminosities to bolometric luminosity with $L_{\\rm bol}=6.03\\,L_{15\\,\\mu\\mathrm{m}}$, based on a composite SED with a torus template, and this conversion is what places the sample on the $L_{\\rm [O\\,III]}$--$L_{\\rm bol}$ diagram. The combination of the color cut, the FWHM criterion, and the SED conversion carries the entire argument.","core_discovery":"The paper confirms 23 of 24 very red, mid-infrared-bright quasar candidates in SDSS Stripe 82 as genuine obscured (Type-2) quasars at $z=0.88$ to $3.49$, with 12 objects above $z=2$. Its central spectroscopic discovery is that the sample is not a single class: 8 of 23 objects show broad H$\\alpha$ emission (FWHM $>2000$ km s$^{-1}$) even though they were selected to be optically faint and red, while most of the rest show only narrow lines; one object has broad lines in both Ly$\\alpha$ and H$\\alpha$. The broad-H$\\alpha$ objects have rest-frame UV/optical SEDs similar to those of the JWST \"little red dots,\" so the paper argues they are lower-redshift analogues of the heavily obscured broad-line AGNs uncovered at $z \\simeq 4$--$9$. At the same time, the sample's [O III] $\\lambda5007$ luminosities fall below the $L_{\\rm [O\\,III]}$--$L_{\\rm bol}$ trend of lower-redshift Type-2 quasars even before extinction correction, and the coadded spectra show weak optical lines such as H$\\beta$, [O II], and [S II] that are usable as future follow-up targets.","pith_inferences":["If the broad-H$\\alpha$ objects are true little-red-dot analogues, the Stripe 82 counts imply that such objects are not rare at $z\\sim2$, and the same selection applied to full-sky WISE catalogs could yield many nearby analogues for detailed study.","A decisive way to separate a physically unobscured broad-line region from scattered or transmitted light would be spectropolarimetry: dust-scattered broad lines should be strongly polarized, and measuring that would settle whether the Type-2 classification is purely a line-of-sight effect or a partially obscured BLR.","The broad-to-narrow H$\\alpha$ fraction could vary with luminosity or redshift; splitting a larger sample by $L_{\\rm bol}$ and $z$ would show whether the broad-line fraction is a property of the most luminous obscured quasars or of the selection itself.","If the [O III] weakness is confirmed with per-object SEDs, it would strengthen the analogy with little red dots, whose emission lines are often weak relative to their red continua."],"forward_implications":["The high confirmation rate (23 of 24) means the $r-W4>8.38$ selection can produce large, clean samples of obscured quasars at $z\\sim1.5$--$3.5$ from wide-area WISE data alone.","The presence of broad H$\\alpha$ in about one third of a red-selected obscured sample shows that optical/UV classification as Type-2 does not rule out a broad-line region, so unification models must accommodate a population with broad Balmer lines seen through red SEDs.","The broad-H$\\alpha$ objects provide a lower-redshift benchmark for the JWST little red dots, enabling comparisons of line widths, SED shapes, and host properties that are difficult for the high-redshift JWST objects.","The weak [O III] relative to bolometric luminosity, if real, implies that the narrow-line region in these luminous obscured quasars is either heavily extinguished or intrinsically different, and any model of obscured growth at cosmic noon will need to explain it.","The new composite spectra provide a template for identifying similar objects in future surveys and for planning JWST observations aimed at detecting weak rest-optical lines."],"supporting_citations":[{"why":"Selects the 24 candidates, builds the composite SED, and supplies the torus model and $L_{\\rm bol}=6.03 L_{15}$ conversion that the [O III] analysis uses.","marker":"Wang et al. (2025)"},{"why":"Provides the early Gemini/GNIRS identifications and redshifts that this paper's sample and line measurements build on.","marker":"Ishikawa et al. (2023)"},{"why":"Supplies the lower-redshift Type-2 QSO sample and the [O III]--mid-IR luminosity correlation used as the comparison baseline for the [O III] deficiency.","marker":"Lacy et al. (2013a)"},{"why":"Defines the 2000 km s$^{-1}$ FWHM threshold used to separate broad and narrow emission lines and hence the Type-2 classification convention.","marker":"Zakamska et al. (2003)"},{"why":"Provides the torus SED template used to convert 15 $\\mu$m luminosity to bolometric luminosity.","marker":"Stalevski et al. (2016)"},{"why":"Introduces the little red dot population that the broad-H$\\alpha$ objects are compared with.","marker":"Matthee et al. (2023)"},{"why":"Example of JWST little red dots/AGNs showing broad emission lines, the high-redshift counterparts invoked for the analogy.","marker":"Kokorev et al. (2023)"},{"why":"Another JWST sample with broad-line AGNs among red compact sources, used to support the little-red-dot analogy.","marker":"Harikane et al. (2023)"},{"why":"Quantifies the extreme obscured-to-unobscured ratio inferred for JWST AGNs, motivating the need for a $z\\sim2$ bridge population.","marker":"Pizzati et al. (2024)"}],"fun_headline_variants":["One-third of obscured quasars reveal broad H-alpha lines","Hidden quasars show broad H-alpha, bridging gap to JWST red dots","A third of 'obscured' quasars aren't so obscured: broad H-alpha found","Broad H-alpha in obscured quasars: cosmic noon cousins of little red dots"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that [O III] is weak for the bolometric luminosity rests on a single SED conversion, $L_{\\rm bol}=6.03 L_{15\\,\\mu\\mathrm{m}}$, applied to every target and to the comparison sample; if individual mid-IR SED shapes differ, the deficiency could disappear.","fun_headline_variants_meta":{"raw":{"variants":["One-third of obscured quasars reveal broad H-alpha lines","Hidden quasars show broad H-alpha, bridging gap to JWST red dots","A third of 'obscured' quasars aren't so obscured: broad H-alpha found","Broad H-alpha in obscured quasars: cosmic noon cousins of little red dots"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000911,"raw_usage":{"total_tokens":4065,"prompt_tokens":1249,"completion_tokens":2816,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":865,"completion_tokens_details":{"reasoning_tokens":2733}},"tokens_in":865,"tokens_out":2816,"duration_ms":19287,"temperature":1.0,"reasoning_tokens":2733,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:37:08.558917+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure each of the 13 [O III]-detected objects' bolometric luminosity independently, using X-ray luminosity or per-object torus SED fits, and re-plot $L_{\\rm [O\\,III]}$ against $L_{\\rm bol}$; if the points fall onto the Lacy et al. relation, the reported [O III] weakness is an artifact of the shared SED scaling.","supporting_citations":[],"review_version":2}