{"id":"dd229010-19a1-4fbd-9304-60c159e2e51a","arxiv_id":"1908.05993","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"IceCube's 10-year point-source search finds a 2.9 sigma excess from NGC 1068 and a 3.3 sigma population-level hint in the northern sky, both below discovery threshold.","lead":"Using 10 years of data, IceCube reports no single confirmed neutrino source, but finds a 2.9 sigma excess from the galaxy NGC 1068 and a 3.3 sigma combined hint from four northern sources. This is a promising clue, not a discovery, that active galaxies may accelerate cosmic rays.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"3.3σ population excess rests on unvalidated RA-scrambling null for misreconstructed Southern muon leakage.","rationale":"The reader's verdict is well calibrated. The analysis is carefully hedged: the 2.9σ and 3.3σ numbers are explicitly post-trial, the TXS-removal caveat is disclosed, and the likelihood-ratio machinery is standard for IceCube point-source searches. The weakest point is indeed the background ensemble. I sharpen the reader's concern: the issue is not simply whether the atmospheric muon model is accurate, but whether the residual Southern-muon leakage is RA-isotropic after reconstruction and selection. Because every pre-trial p-value feeding the population statistic is calibrated with RA-scrambled data, any RA-dependent leakage would directly and spuriously lower those p-values. The paper's own Data Selection text flags the leakage as non-negligible, and no validation is shown that it has no sidereal structure. The 3.3σ population result also depends on four specific sources, and the stated drop to 2.25σ upon removing TXS shows how sensitive the headline is to catalog and systematic choices. Still, these are validation gaps rather than demonstrated errors, so the claim is not refuted; the appropriate outcome remains CONDITIONAL pending a targeted closure test. My review therefore leaves the reader's verdict unchanged.","tokens_in":107,"tokens_out":10465,"duration_ms":127385,"concrete_test":"Run an ensemble test with the same 10-year Northern event selection and likelihood: generate background-only pseudo-experiments using (i) the paper's uniform-RA scrambling and (ii) the same scrambling augmented with simulated misreconstructed Southern atmospheric muons at the observed BDT leakage rate, imposing the known TeV cosmic-ray sidereal anisotropy (amplitude ~10^-3) on their arrival directions. For each ensemble, compute the Northern catalog pre-trial p-value distribution and the k=4 population post-trial p-value. If the augmented ensemble reproduces the observed four-source excess as often as ~5% of trials, or produces a 3.3σ-equivalent post-trial p-value comparable to 4.8e-4, the RA-scrambling null is inadequate; if the augmented ensemble remains uniform in RA and the observed excess stays significant, the concern is settled.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the 3.3σ post-trial Northern catalog population excess (Table 2), driven by the pre-trial p-values of NGC 1068, TXS 0506+056, PKS 1424+240, and GB6 J1542+6129. Those p-values are calibrated against background-only trials obtained by assigning each event a uniform RA while preserving all other properties. The Data Selection section explicitly states that poorly reconstructed Southern atmospheric muons still constitute a non-negligible background in the Northern hemisphere. RA scrambling is a valid null only if that residual leakage is RA-isotropic after reconstruction and selection. This is not established: TeV atmospheric muons have known sidereal anisotropies and local-azimuth asymmetries, and the BDT plus misreconstruction can imprint a declination- and energy-dependent RA pattern. If such structure overlaps the four candidate directions, the low pre-trial p-values and the k=4 binomial significance (3.3e-5) are biased downward, making the 3.3σ a statement about residual atmospheric background rather than neutrinos. The paper provides no closure test—neither a simulated-leakage check nor a demonstration that the remaining Northern catalog p-values are uniform—and the population statistic is not defined in enough detail to verify the post-trial correction. The claim is also fragile: removing TXS drops the significance to 2.25σ, a caveat the authors disclose but which intensifies the need for validation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This proceedings paper from the IceCube Collaboration analyzes roughly ten years of track-like muon-neutrino data to search for point-like astrophysical sources. Three searches are presented: an all-sky scan, a search over a 110-source catalog selected from Fermi and TeVCat, and a catalog population study. The most significant Northern all-sky hotspot is located 0.35 degrees from NGC 1068 with a pre-trial p-value of 3.5e-7 and a post-trial significance of 9.9%. The source-catalog search finds a 2.9 sigma post-trial excess at NGC 1068, and the population study reports a 3.3 sigma post-trial excess driven by four Northern catalog sources (NGC 1068, TXS 0506+056, PKS 1424+240, and GB6 J1542+6129). The population excess drops to 2.25 sigma when TXS 0506+056 is removed. The Southern sky shows no significant excess. The paper concludes that no individual source is firmly established but that some point-like sources may be emerging from the background.","tokens_in":6814,"tokens_out":4141,"duration_ms":39243,"significance":"If the population result holds, this is one of the strongest current hints of a population of point-like neutrino sources, with NGC 1068 as the leading single candidate. The analysis is a careful update of standard IceCube methods and documents an improved event selection with better angular resolution and sensitivity. The paper is transparent in reporting the drop to 2.25 sigma when TXS 0506+056 is removed and in discussing the offset between the all-sky hotspot and the catalog position of NGC 1068. However, the statistical evidence is modest, the central 3.3 sigma claim depends on an unvalidated background-scrambling assumption, and the population-study statistic is not specified in enough detail to be independently reproduced from the proceedings text.","major_comments":[{"comment":"The background-only p-values quoted in Table 2, including the 3.3 sigma population result, are calibrated by assigning each event a new right ascension uniformly on the sky while preserving all other properties. The paper states in §2 that poorly reconstructed Southern atmospheric muons still constitute a non-negligible Northern background. The validity of the RA-scrambling null therefore requires that this residual muon contamination be RA-isotropic after reconstruction and selection. This is not established in the manuscript; atmospheric muons are known to exhibit local-azimuth and sidereal anisotropies, and the BDT selection could imprint a declination- and energy-dependent RA pattern on the surviving events. I ask for a closure test, for example a simulated-leakage study verifying that scrambled trials reproduce the observed event properties, or a demonstration that the Northern catalog pre-trial p-values are uniform after excluding the flagged sources.","section":"§2 Data Selection and §3 All-Sky Scan"},{"comment":"The population-study statistic is not fully specified. The displayed equation for the cumulative binomial probability pbkg is missing from the text, and the procedure for scanning k and the threshold pk is described only verbally, so the post-trial correction of 4.8e-4 cannot be reproduced from the proceedings. Please provide the exact binomial formula, the range and step of the scan, whether pk is chosen a priori or from the data, and the number of background trials used for the post-trial correction.","section":"§3 Catalog Population Study"},{"comment":"The 3.3 sigma population excess is driven by four sources, one of which, TXS 0506+056, was already known from the IceCube-170922A flare. The authors disclose that removing TXS reduces the post-trial significance to 2.25 sigma, which is commendable, but this fragility means the evidence for a population of new sources is weak. The conclusion that 'some of the first point-like sources could potentially be emerging from the background' should be tempered accordingly, and a sentence should quantify how much of the 3.3 sigma is independent of the previously known TXS association.","section":"§3 Source Catalog Searches and Table 2"}],"minor_comments":[{"comment":"The best-fit values for the number of signal events ns and spectral index gamma are quoted without any uncertainties; please provide confidence intervals or state explicitly that uncertainties are not reported in this proceedings.","section":"§3 All-Sky Scan and Source Catalog Searches"},{"comment":"The date range for the IC86-2012-18 row contains a typo, reading '2012/04/26 1 2018/07/10'; it should be '2012/04/26 – 2018/07/10'.","section":"Table 1"},{"comment":"The text reads 'EightFermi-LAT sources' where 'Eight Fermi-LAT sources' is intended; please fix the missing space.","section":"§3 Source Catalog Searches"},{"comment":"Several references lack complete publication details, for example [10] and [20]; please add volume, page, and DOI information where available.","section":"References"},{"comment":"The conclusion states that there is 'no evidence for an astrophysical neutrino source' while the abstract and §3 report a 3.3 sigma inconsistency with background; please clarify that 'no evidence' refers to individual sources and not to the collective population excess.","section":"§4 Conclusions"}],"recommendation":"major_revision","confidential_remarks":"This is a proceedings contribution, and the main numerical claims depend on details that are not fully present in the text. The missing population-study equation and the lack of a closure test for the RA-scrambling null are the main technical obstacles. If the journal can accommodate supplementary material or a companion analysis, these issues could be fixed; otherwise the central 3.3 sigma claim is not fully verifiable from the manuscript. The inclusion of the already-known TXS 0506+056 in the population result is disclosed, but it reduces the novelty of the claim and should be weighed in the editorial decision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere's the short version: this ten-year IceCube proceedings is a competent, incremental update that produces the first credible hint of a Seyfert neutrino source, but the stronger-sounding 3.3σ population claim is not as solid as it looks, and the null hypothesis it rests on is not tested in the paper.\n\nWhat is actually new: a unified event selection over the full 10-year sample, with a ~35% sensitivity improvement over the 7-year analysis, and post-trial p-values that do not appear in earlier work. NGC 1068 at 2.9σ post-trial is the main event, and the authors do a nice job showing that the 0.35° offset between the all-sky hotspot and the Fermi coordinates is plausibly a statistical fluctuation for a soft spectrum. They also disclose the TXS 0506+056 dependence head-on: remove it and the catalog population excess drops from 3.3σ to 2.25σ. That transparency is to their credit.\n\nThe soft spots are real but not disqualifying. The stress-test worry about RA scrambling is fair: the background-only trials are made by randomizing right ascension while preserving everything else, and that is a valid null only if the mis-reconstructed southern muon leakage into the Northern sky is RA-isotropic after selection. The paper acknowledges the leakage is non-negligible but never demonstrates its RA distribution is flat. I would not call this a fatal flaw—this scrambling procedure is standard in IceCube point-source searches and any residual anisotropy would have to line up with the four candidate directions to fully explain the 3.3σ—but a closure test, or at least a statement of the background model's systematics, should be in the journal version. The missing error bars on n_s and gamma are also a genuine omission, and the population statistic is described only in words; the equation is absent from the text. Minor, but it should be fixed.\n\nThe citation pattern looks fine; it points back to earlier IceCube methods papers properly. Nothing invented.\n\nBottom line: this is a paper for the neutrino and multimessenger community, and the NGC 1068 hint is worth taking seriously. As a conference proceedings it is what it is; the full analysis deserves a proper journal paper, and a serious referee should be able to get it into shape. I would not desk-reject it, but I would ask for background-model validation and error bars first.","headline":"Careful, honest proceedings with a new 10-year sample; the 3.3σ population signal is fragile and its scrambling null is unvalidated, but the NGC 1068 hint is real and deserves follow-up.","tokens_in":7391,"tokens_out":2494,"would_cite":false,"duration_ms":26608,"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":"Ten years of IceCube data produce a 3.3-sigma population excess toward four Northern neutrino source candidates, led by NGC 1068 at 2.9 sigma.","keywords":["neutrino point sources","IceCube","NGC 1068","TXS 0506+056","source catalog search","population study","maximum likelihood ratio","astrophysical neutrinos"],"falsifier":"Rerun the Northern catalog population analysis on the same ten-year sample after removing all events below 10 TeV or after applying a stricter track-quality cut; if the pre-trial p-value distribution at the four candidate directions becomes uniform, the 3.3-sigma excess would be attributable to low-energy or mis-reconstructed events rather than an astrophysical population. Conversely, if an independent event selection preserves or strengthens the excess, and a further few years of data push the combined significance past 5-sigma, the population claim would be confirmed.","tokens_in":6327,"feed_emoji":"🧊","tokens_out":14794,"duration_ms":123852,"temperature":0.7,"pith_summary":"These proceedings report a unified analysis of about ten years of IceCube data, from April 2008 to July 2018, searching for point-like neutrino sources among track-like events above roughly 1 TeV. No single source reaches the $5\\sigma$ discovery threshold, but the Northern source catalog gives the Seyfert II galaxy NGC 1068 a post-trial excess of $2.9\\sigma$, and a catalog population study finds that four Northern sources together—NGC 1068, TXS 0506+056, PKS 1424+240, and GB6 J1542+6129—are inconsistent with a background-only sky at $3.3\\sigma$ post-trial. The Southern catalog and the all-sky scan remain consistent with background. If the population excess is real, it would be the first indication that a class of point-like astrophysical objects, not just one source, emits high-energy neutrinos, and it would support the hadronic origin of at least part of the cosmic-ray flux.","feed_headline":"Four candidate sources drive a 3.3-sigma neutrino excess","feed_subtitle":"The leading candidate, NGC 1068, stands at 2.9 sigma post-trial in ten years of IceCube data.","key_machinery":"The load-bearing machinery is the unbinned maximum-likelihood-ratio test statistic comparing a point-like signal plus diffuse background against a background-only null, maximizing the number of signal events $\\hat{n}_s$ and the spectral index $\\hat{\\gamma}$ at each trial direction. Background trials are produced by re-assigning each event a right ascension drawn uniformly over the sky while keeping all other event properties fixed; the pre-trial $p$-value at each position comes from the resulting test-statistic distribution, and post-trial significance comes from comparing the largest data fluctuation with the distribution of largest fluctuations in background trials. For the catalog population study, a cumulative binomial probability $p_{\\rm bkg}$ is computed for finding $k$ sources with pre-trial $p$-values below a threshold $p_k$, with $k$ scanned to maximize sensitivity and a trial factor applied for the scan. This machinery converts the raw clustering of events into every quoted $p$-value and significance in Table 2.","core_discovery":"The paper's central claim is that after ten years of IceCube data, the strongest evidence for point-like neutrino emission is collective rather than single-source. In the Northern catalog of 97 objects, NGC 1068 shows the most significant excess, with a pre-trial $p$-value of $1.8\\times10^{-5}$ ($4.1\\sigma$) and a post-trial significance of $2.9\\sigma$; its best-fit spectrum is soft, $\\hat{\\gamma}=3.2$, with $\\hat{n}_s=50.4$ signal events. The all-sky scan's hottest Northern spot lies $0.35^\\circ$ from NGC 1068, and injection simulations with an $E^{-3.2}$ spectrum show that such an offset is a plausible statistical fluctuation. The population study finds its strongest signal for $k=4$ sources with pre-trial $p=3.3\\times10^{-5}$ and post-trial $p=4.8\\times10^{-4}$ ($3.3\\sigma$), from excesses at NGC 1068, TXS 0506+056, PKS 1424+240, and GB6 J1542+6129; removing TXS 0506+056 leaves a $2.25\\sigma$ post-trial excess from the remaining three. The paper concludes that some point-like sources could be starting to emerge from the background, with more data or future detectors needed to confirm or rule them out.","pith_inferences":["Because the four excess sources belong to different classes of active galactic nuclei—one Seyfert II and three blazars—the population result, if real, would suggest that neutrino production is a common property of active galactic nuclei rather than a single subclass; the paper itself does not draw this conclusion.","A useful control experiment would be to rerun the population analysis with a source catalog built without gamma-ray flux weighting, or with a differently selected gamma-ray source list; if the $3.3\\sigma$ excess persists under a differently selected catalog, it would be harder to attribute to selection bias.","The population statistic counts sources below a p-value threshold but does not use the measured source fluxes simultaneously; a joint likelihood fit of the four candidate fluxes and a common spectral index could provide a stronger and more interpretable test of the same hypothesis.","The $0.35^\\circ$ hotspot offset suggests that future all-sky scans should include energy-dependent point-spread functions or extended-source templates, since a soft-spectrum point source can shift the reconstructed maximum noticeably."],"forward_implications":["If the $3.3\\sigma$ Northern population excess is real, the four candidate sources constitute the first resolved population of point-like neutrino emitters, meaning part of the diffuse astrophysical neutrino flux is produced by discrete objects rather than only by a truly diffuse mechanism.","The $2.9\\sigma$ excess at NGC 1068, combined with the simulated plausibility of the $0.35^\\circ$ offset, makes this Seyfert II galaxy the strongest single candidate to monitor as more IceCube data accumulate.","Because the population excess remains at $2.25\\sigma$ after removing TXS 0506+056, the signal is not carried by a single flaring source, but its significance depends on several modest excesses and could disappear if any one of them is a fluctuation.","The Southern catalog and the all-sky scan show no comparable signal, so any emerging point-source population is currently evident only in the Northern hemisphere, where IceCube's track sensitivity is highest.","The paper's sensitivity projection implies that a future detector with an order-of-magnitude better discovery potential would be able to confirm or exclude neutrino fluxes at the level of the best-fit fluxes of these four sources."],"supporting_citations":[{"why":"Supplies the maximum-likelihood-ratio search method and the earliest IceCube sample (IC40) included in the ten-year data set.","marker":"[6]"},{"why":"The seven-year all-sky point-source search whose event selection, sensitivity, and source catalog this analysis extends and compares against.","marker":"[9]"},{"why":"The eight-year Northern-sky track analysis whose event selection and population-study approach are unified into this work.","marker":"[10]"},{"why":"Sets out the unbinned likelihood-ratio formalism for point-source searches used to compute all test statistics.","marker":"[15]"},{"why":"The gamma-ray source catalog used to select and flux-weight the extragalactic members of the 110-source candidate catalog.","marker":"[16]"},{"why":"The very-high-energy gamma-ray source catalog from which Galactic candidates are selected by converted hadronic flux estimates.","marker":"[19]"},{"why":"The report of neutrino emission from TXS 0506+056 that motivates the a-posteriori removal test in the population study.","marker":"[20]"},{"why":"The measurement identifying NGC 1068 as the brightest Seyfert II galaxy at gamma-ray energies, used to interpret the leading excess.","marker":"[21]"}],"fun_headline_variants":["Four sources hint at cosmic neutrino origins","NGC 1068 tops neutrino point-source search","Collective neutrino excess from four sources at 3.3 sigma","Ten-year IceCube scan hints at neutrino point sources"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"All quoted p-values rest on the assumption that randomly re-assigning each event a new right ascension produces a faithful background-only sky, and that the atmospheric muon and neutrino backgrounds, including mis-reconstructed Southern muons leaking into the North, are accurately modeled.","fun_headline_variants_meta":{"raw":{"variants":["Four sources hint at cosmic neutrino origins","NGC 1068 tops neutrino point-source search","Collective neutrino excess from four sources at 3.3 sigma","Ten-year IceCube scan hints at neutrino point sources"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000867,"raw_usage":{"total_tokens":3812,"prompt_tokens":1057,"completion_tokens":2755,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":2693}},"tokens_in":673,"tokens_out":2755,"duration_ms":16998,"temperature":1.0,"reasoning_tokens":2693,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:58:35.398363+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Rerun the Northern catalog population analysis on the same ten-year sample after removing all events below 10 TeV or after applying a stricter track-quality cut; if the pre-trial p-value distribution at the four candidate directions becomes uniform, the 3.3-sigma excess would be attributable to low-energy or mis-reconstructed events rather than an astrophysical population. Conversely, if an independent event selection preserves or strengthens the excess, and a further few years of data push the combined significance past 5-sigma, the population claim would be confirmed.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The eight-year Northern-sky track analysis whose event selection and population-study approach are unified into this work."},{"cited_title":"Braun, J","cited_arxiv_id":null,"evidence_quote":"Sets out the unbinned likelihood-ratio formalism for point-source searches used to compute all test statistics."},{"cited_title":"Acero et al., Astrophys","cited_arxiv_id":null,"evidence_quote":"The gamma-ray source catalog used to select and flux-weight the extragalactic members of the 110-source candidate catalog."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The very-high-energy gamma-ray source catalog from which Galactic candidates are selected by converted hadronic flux estimates."},{"cited_title":"Ackermann et al., Astrophys","cited_arxiv_id":null,"evidence_quote":"The measurement identifying NGC 1068 as the brightest Seyfert II galaxy at gamma-ray energies, used to interpret the leading excess."}],"review_version":1}