{"id":"6c7e8968-1879-462c-9758-9c6bbc4a5419","arxiv_id":"2504.13248","paper_version":4,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A transient point source in COSMOS-Web, HZTDE-1, has a blackbody-like SED consistent with a tidal disruption event at z~5, though a superluminous supernova at z>3 remains equally plausible.","lead":"Using deep JWST images of the COSMOS field, the authors found a point-like infrared source that appears in 2024 data but is absent from images taken between 2005 and 2016. They argue it is a candidate for a star being torn apart by a black hole at redshift about 5, which would be the most distant such event ever seen.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Single-epoch SED cannot distinguish a z≈5 TDE from a z≈3.2 SLSN; the paper itself states these are equally likely, so the highest-redshift TDE claim is not yet substantiated.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing concern: the redshift and classification derive from a single blackbody model with a degeneracy between temperature and redshift, and a SLSN at z≈3.2 fits the same photometry. The paper itself is transparent about this, and the reader's CONDITIONAL verdict fairly reflects that the candidate merits follow-up but the headline claim is not yet substantiated. I find no additional internal inconsistency or more fundamental flaw. The transient nature of HZTDE-1 is convincingly established by the archival nondetections and the Appendix A aperture checks; the DLR analysis makes host association with the z≈1.75 galaxy unlikely; and the comparison to ordinary SN and AGN models is thorough. The sole decisive gap is the TDE-versus-SLSN classification, which only time-domain or spectroscopic follow-up can close. Therefore my stress-test does not change the reader's verdict: the paper is publishable as a candidate discovery with a clearly stated caveat, but the central claim of the highest-redshift TDE is conditional on future observations. The proposed second-epoch imaging test would settle the ambiguity. I agree with the reader's assessment and recommend no change to the verdict.","tokens_in":31408,"tokens_out":4931,"duration_ms":45653,"concrete_test":"Obtain a second epoch of NIRCam imaging in F150W and F277W, separated by 6–8 months in the observer frame (roughly 1.5–2 rest-frame months at z≈5, where time dilation is a factor of six). If HZTDE-1 fades by more than ~0.5 mag while its F150W−F277W color remains constant, the constant-temperature TDE interpretation is supported and the SLSN alternative (which cools and reddens after peak) is strongly disfavored. If the source instead reddens as it fades, the z≈3.2 SLSN interpretation becomes preferred and the z≈5 TDE claim is falsified. This test is explicitly recommended in Sections 2.5 and 5 of the paper as the distinguishing observation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that HZTDE-1 is a TDE at z=5.02—rests entirely on a single-epoch SED fit with a constant-temperature blackbody, using priors from local ZTF TDEs and a uniform redshift prior 3.5≤z≤7.5. Because a redshifted blackbody is itself a blackbody, temperature and redshift are degenerate; the same four-band photometry is fit by a 19,000 K superluminous supernova at z=3.2, as the authors explicitly show in Section 4.1 and Figure 8. The paper's own summary states: \"without further monitoring, the possibility of a lower-z SLSN is equally likely to a TDE.\" Therefore the headline \"highest-redshift TDE candidate\" is not yet justified. The lower redshift bound (z>3.5) also depends on the model assumption that a typical TDE host galaxy would be detected in the NIRCam short-wavelength filters; a low-mass, faint host at z≈2 could be undetected, allowing a cooler (≈10,000 K) TDE at lower redshift that fits the same photometry. The classification and the redshift are thus both contingent on the same unverified degeneracy between a high-z TDE and a lower-z SLSN. This is not an internal inconsistency, but it is the decisive weakness: the evidence presented does not uniquely select the TDE hypothesis over a plausible and physically motivated alternative.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a method for identifying high-redshift tidal disruption event (TDE) candidates in deep JWST near-infrared imaging using color-color, color-magnitude, and point-source morphology cuts, and applies it to the COSMOS-Web survey. The selection yields one transient point source, HZTDE-1, which is detected in NIRCam (2024) but not in earlier HST, HSC, or UltraVISTA imaging. The authors fit a constant-temperature blackbody TDE model and report a photometric redshift of z=5.02(+1.32/-1.11), an absolute magnitude of M_g=-21.15, and log(T_BB)=4.31, making it a candidate highest-redshift TDE. They compare against supernova and AGN models and find that a superluminous supernova at z~3.2 can equally well explain the SED, so they stop short of claiming a confirmed TDE. The paper also discusses future confirmation strategies and the implications for high-redshift TDE rates.","tokens_in":31708,"tokens_out":4089,"duration_ms":40754,"significance":"If the TDE interpretation were confirmed, this would be a scientifically important result: the highest-redshift TDE candidate to date, with implications for SMBH demographics and TDE rates at early cosmic times. The methodological contribution is also potentially useful for Roman and other deep infrared surveys. Strengths of the paper include the careful verification of the nondetection in archival imaging (Appendix A), the use of directional light radius to assess host association, and the unusually candid admission of the SLSN degeneracy. However, as presented, the central classification is not established: the paper's own analysis states that a lower-redshift SLSN is equally likely, and the photometric redshift is strongly prior-dependent. The significance of the discovery therefore hinges on follow-up data or on a substantial reframing of the claim.","major_comments":[{"comment":"The central claim that HZTDE-1 is a high-redshift TDE is not uniquely supported by the data, and the authors themselves state that a 19,000 K SLSN at z=3.2 fits the same SED and that 'the possibility of a lower-z SLSN is equally likely to a TDE.' Because the title and abstract present HZTDE-1 as a high-redshift TDE candidate, the paper must either provide a quantitative model comparison that includes prior rates and demonstrates preference for the TDE interpretation, or reframe the manuscript as a search-method paper plus an unclassified UV-bright transient candidate. As written, the headline claim overstates the evidence.","section":"§4, Fig. 8"},{"comment":"The photometric redshift z=5.02(+1.32/-1.11) is a posterior conditioned on a uniform prior 3.5<=z<=7.5 and on the assumption that TDE host galaxies drop out below z~3.5. The paper explicitly notes that a 10^4 K TDE at z~2 can fit the same photometry if the host is undetected. Thus the redshift is not an independent measurement but a model-dependent inference; the lower bound should be presented as conditional on the host-dropout assumption, and that assumption should be tested against explicit faint-host scenarios or stacked upper limits.","section":"§4, Fig. 9"},{"comment":"The selection region (Eq. 2, Table 1, and the magnitude-color cut of Eq. 3) is derived from the same constant-temperature blackbody TDE model that is later used to fit HZTDE-1. The object was chosen by those cuts and then fit with that model, so the apparent consistency of the SED with a TDE is partly circular. The paper should quantify the expected contamination rate of the selection by injecting SLSN, cool dwarf, and compact-galaxy templates and reporting the fraction of selected sources that would be TDEs under stated rate assumptions, or explicitly label the candidate as selected by a TDE-simulation-based box rather than as an independently validated TDE.","section":"§2.3 and §3.2 vs. §4"}],"minor_comments":[{"comment":"The text defines 'bright' sources as having MF115W > 27.8 AB mag, but a magnitude greater than the detection limit corresponds to a fainter source; the inequality or the labels should be corrected.","section":"§3.2"},{"comment":"The caption contains a typo: 'anr parameterized' should be 'and parameterized'.","section":"Table 1 caption"},{"comment":"The predicted factor-of-ten TDE rate enhancement relies on Karmen & et al. (in prep.) and should be clearly marked as unpublished/speculative in the main text; currently it is used to argue that the candidate is plausible.","section":"§5.1"},{"comment":"The statement that SNe II 'would need rest-frame b-band absolute magnitudes ≲−20, which is not observed in SNe IIP' should clarify that this refers specifically to normal SNe IIP and not to superluminous supernovae, which are discussed separately.","section":"§2.5"},{"comment":"The paragraph describing the cross-match with previous COSMOS data would benefit from a clearer statement of the decision tree: which candidates are checked in COSMOS2020, which are manually inspected, and how the F115W brightness threshold is used.","section":"§3.1"}],"recommendation":"major_revision","confidential_remarks":"The authors are appropriately cautious in places, but the title and abstract framing outpace the evidence. I would support publication after either obtaining follow-up data or substantially reframing the paper as a methodology paper with an unclassified transient candidate. There are no ethical concerns about the conduct of the research itself."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHZTDE-1 is a real find, but not for the reason the title emphasizes. The genuinely new thing is the search method: using NIRCam point-source morphology plus color-color cuts calibrated on simulated high-z TDEs, then checking variability against 20 years of COSMOS archival data, to dig TDE candidates out of a single-epoch deep survey. That is a sensible and useful route for Roman. The candidate itself is solidly a transient: the Appendix A aperture checks convincingly rule out a faint counterpart in UltraVISTA and HSC, and the DLR analysis makes association with the nearby z~1.75 galaxy unlikely.\n\nThe paper is also unusually honest. Section 4.1 states plainly that a 19,000 K superluminous supernova at z=3.2 fits the same photometry, and the conclusion repeats that without monitoring a lower-z SLSN is equally likely. That is the right scientific posture, and it should be credited.\n\nThe soft spot is the headline. \"Highest-redshift TDE candidate\" is not supported by a single-epoch four-band SED. The redshift and the classification come from the same blackbody model that defined the selection region, with priors drawn from local ZTF TDEs and a uniform redshift prior 3.5–7.5. Because a redshifted blackbody is still a blackbody, temperature and redshift are degenerate; the paper itself shows the alternative. The lower redshift bound z>3.5 rests on the assumption that a typical TDE host would be detected in F115W/F150W—but a low-mass host at z~2 could easily drop out, making a cooler ~10,000 K TDE at lower z fit as well. So the candidate is interesting, but the specific claim of z~5 is model-dependent, not measured.\n\nOne more thing: the rate-enhancement discussion leans on a companion paper (Karmen et al., in prep.) that is not available. That is fine for context, but it should not be load-bearing in the interpretation.\n\nBottom line: this deserves a serious referee. It is a thorough, well-documented search paper with a candidate worth follow-up. I would recommend softening the abstract to say \"TDE candidate possibly at z~5, SLSN not ruled out\"—which is already close to what the body says—and avoiding the highest-redshift phrasing until NIRSpec or photometric monitoring settles it. As a methodology paper, it will be cited.","headline":"The search method is the contribution; the z~5 TDE claim is not yet substantiated—the paper's own SLSN fit shows the SED alone can't tell them apart.","tokens_in":32453,"tokens_out":2262,"would_cite":true,"duration_ms":20775,"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":"A transient point source in JWST's COSMOS-Web field is best explained as a tidal disruption event at $z\\approx5$, the most distant such candidate found to date.","keywords":["tidal disruption events","high-redshift transients","JWST NIRCam","COSMOS-Web survey","photometric redshift","superluminous supernovae","supermassive black holes"],"falsifier":"Two observations would settle it. A second NIRCam epoch 6-8 observer-frame months after the first: if HZTDE-1 faded at constant color, the TDE interpretation survives; if it cooled and reddened, it is a superluminous supernova. A NIRSpec spectrum would be decisive: broad H$\\alpha$ or He II at $z\\approx5$ confirms a TDE, while a Lyman break and blue continuum near $z\\approx3.2$ identifies a SLSN.","tokens_in":31177,"feed_emoji":"🔭","tokens_out":8020,"duration_ms":67052,"temperature":0.7,"pith_summary":"This paper reports HZTDE-1, an unresolved point source that appears in JWST NIRCam imaging of the COSMOS-Web field but is absent from Hubble, Subaru, and VISTA images taken between 2005 and 2016. The authors argue that its four-band infrared SED is best described by a redshifted constant-temperature blackbody with $\\log T=4.31\\pm0.09$ and $M_g=-21.15$, which for a tidal disruption event (TDE) implies a photometric redshift of $z=5.02^{+1.32}_{-1.11}$. That would make it the highest-redshift TDE candidate known, roughly four times farther than the previous record. They also show that the SED cannot be reproduced by ordinary Type Ia, Ib/c, IIP, or IIn supernovae or by AGN at lower redshift, although a superluminous supernova at $z\\gtrsim3$ remains a viable alternative. The larger claim is methodological: in deep infrared surveys, high-redshift TDEs should appear as hostless point sources with a recognizable color track, so a single epoch of JWST or Roman imaging can identify them and open a new window on early black hole demographics.","feed_headline":"JWST spots a tidal disruption candidate at z≈5","feed_subtitle":"A point source absent from two decades of imaging fits a stellar-death flare at z≈5, the most distant candidate yet.","key_machinery":"The load-bearing object is HZTDE-1 itself, selected by a color-and-morphology pipeline built for hostless high-redshift transients. The paper simulates TDEs as redshifted constant-temperature blackbodies with temperatures and luminosities drawn from the local ZTF TDE sample, and adds host-galaxy spectra from 30 local TDE hosts to predict when the host drops out of NIRCam detection ($z\\gtrsim4$). Because a redshifted blackbody is still a blackbody, all idealized TDEs fall on a nearly one-dimensional curve in NIRCam color-color space; the paper fits parabolas to those curves, adds 0.2 mag of scatter measured from real UV TDE templates, and combines the color cuts with a point-source aperture-flux-ratio cut and a nondetection check in the COSMOS2020 archive. That sequence reduces more than 700,000 sources to 117 candidates, and then to HZTDE-1 as the only one that is point-like and genuinely absent from all previous imaging.","core_discovery":"The central discovery is a single transient: HZTDE-1. In the COSMOS-Web NIRCam data it is point-like in F115W and F150W, detected at about 25 mag in F150W, and undetected to $1\\sigma$ in the deeper archival UltraVISTA stacks; aperture photometry confirms the nondetection at its exact position. It sits more than five directional light radii from the nearest plausible host galaxy, and its SED is poorly fit by galaxy, star, and AGN templates. Under the assumption that the source is a TDE with a constant-temperature blackbody spectrum drawn from the local ZTF TDE population, the MCMC fit gives $z=5.02^{+1.32}_{-1.11}$, $\\log T=4.31\\pm0.09$, and $M_g=-21.15^{+0.21}_{-0.13}$. Supernova models at the nearby galaxy's redshift $z\\approx1.75$ fail: the source is too red for SNe Ia, too bright for SNe Ib/c, and only an unusually bright IIn with an infrared excess could come close. A superluminous supernova at $z\\gtrsim3$ with a 19,000 K blackbody fits the photometry, so the paper leaves that as an open alternative while noting that the source would then be the highest-redshift SLSN known.","pith_inferences":["Because the classification rests on a single epoch, the relative volumetric rates of TDEs and SLSNe at $z>3$ determine the prior odds; if the high-redshift TDE rate enhancement argued in the paper is real, HZTDE-1 being a TDE becomes more probable than the photometry alone implies.","A cheap testable extension is to require two epochs separated by roughly six to eight observer-frame months: TDEs are expected to fade without changing color, while SLSNe cool and redden, which would break the main degeneracy without spectroscopy.","If rest-frame UV TDE spectra deviate from a blackbody via Bowen fluorescence or disk reprocessing, the inferred redshift and temperature could shift systematically; comparing high-redshift candidates against the HST UV TDE templates used for calibration would quantify that bias.","For a confirmed TDE, late-time infrared follow-up searching for a dust echo could measure the surrounding nuclear environment and help estimate the black hole mass independently of the flare photometry."],"forward_implications":["If HZTDE-1 is confirmed as a TDE, it would be the highest-redshift tidal disruption event found to date, at $z\\approx5$, and would show that such flares are detectable in a single deep JWST epoch.","A confirmed high-redshift TDE population would support an enhanced TDE rate in the early universe, driven by compact nuclear stellar clusters and merger activity, rather than a rate that simply declines with the low-redshift black hole mass function.","The color-color and morphology selection can be carried over to other wide-field infrared surveys, most directly the Roman High Latitude Wide Area Survey, to find dozens or hundreds of $z>4$ TDEs and similar UV-bright transients.","If the transient is instead a superluminous supernova, it would be the highest-redshift SLSN known and would constrain the rate and host environments of massive-star explosions at early cosmic times.","High-redshift TDEs give a way to weigh supermassive black holes below the AGN-selected mass range and to probe how seed black holes formed and grew."],"supporting_citations":[{"why":"Supplies the TDE light-curve shape and the typical blackbody temperatures that define the simulated high-redshift TDE population.","marker":"van Velzen et al. 2021"},{"why":"Provides the ZTF TDE temperature and luminosity distributions used as priors in the MCMC fit of HZTDE-1's redshift and luminosity.","marker":"Yao et al. 2023"},{"why":"Supplies the 30 local TDE host galaxies whose spectra and sizes are used to predict that high-redshift TDEs appear hostless.","marker":"Hammerstein et al. 2021"},{"why":"Describes the COSMOS-Web survey whose NIRCam imaging contains the detection.","marker":"Casey et al. 2022"},{"why":"Provides the COSMOS-Web catalog used for color, morphology, and forced-photometry selection.","marker":"Shuntov et al. 2025"},{"why":"Defines the directional light radius statistic used to show HZTDE-1 is hostless and unlikely to be a supernova in the nearby galaxy.","marker":"Gupta et al. 2016"},{"why":"Supplies the superluminous supernova templates and luminosity ranges used to test the remaining SLSN explanation.","marker":"Gomez et al. 2024"},{"why":"Provides an alternative UV-bright TDE disk model used to check that the color selection is not an artifact of blackbody extrapolation.","marker":"Dai et al. 2018"},{"why":"Gives the previous TDE rate framework that the paper extends to high redshift when interpreting the candidate's significance.","marker":"Kochanek 2016"}],"fun_headline_variants":["JWST finds possible star-shredding event at z≈5","Most distant tidal disruption event candidate yet found by JWST","JWST spots possible TDE at z≈5 but SLSN alternative remains","New JWST method reveals distant black hole meal candidate","Record-holding TDE candidate at z≈5 from JWST data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole classification depends on the idea that a high-redshift TDE looks like a single constant-temperature blackbody drawn from the local TDE population; because redshift and temperature are degenerate, a cooler TDE at $z\\sim2$ or a hot superluminous supernova at $z\\sim3$ can mimic the same four-band colors.","fun_headline_variants_meta":{"raw":{"variants":["JWST finds possible star-shredding event at z≈5","Most distant tidal disruption event candidate yet found by JWST","JWST spots possible TDE at z≈5 but SLSN alternative remains","New JWST method reveals distant black hole meal candidate","Record-holding TDE candidate at z≈5 from JWST data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001491,"raw_usage":{"total_tokens":6102,"prompt_tokens":1177,"completion_tokens":4925,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":793,"completion_tokens_details":{"reasoning_tokens":4836}},"tokens_in":793,"tokens_out":4925,"duration_ms":31266,"temperature":1.0,"reasoning_tokens":4836,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T12:13:08.599619+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Two observations would settle it. A second NIRCam epoch 6-8 observer-frame months after the first: if HZTDE-1 faded at constant color, the TDE interpretation survives; if it cooled and reddened, it is a superluminous supernova. A NIRSpec spectrum would be decisive: broad H$\\alpha$ or He II at $z\\approx5$ confirms a TDE, while a Lyman break and blue continuum near $z\\approx3.2$ identifies a SLSN.","supporting_citations":[],"review_version":1}