{"id":"9b1e6386-3f73-4336-9ab6-b0f1c030690f","arxiv_id":"1908.04884","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"IceCube's next-generation realtime alert system adds two signalness-based channels (Gold and Bronze) and new event selections, with expected rates of about 10 and 20 alerts per year.","lead":"The IceCube neutrino observatory has upgraded its real-time alert system to flag astrophysical neutrino candidates for telescopes to follow up. The new two-tier 'Gold and Bronze' system aims to send about 30 alerts per year with clearer messages and better direction estimates.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Table 1's Bronze signal fraction is 16%, not the advertised 30–50%; this internal inconsistency undermines the Bronze purity claim.","rationale":"After reading the paper in good faith, the central claim is that the new two-channel system will provide roughly 10 Gold and 20 Bronze alerts per year with defined signalness and improved purity. The strongest support is the system description and effective-area plots; the selection logic (GFU, HESE, EHE) is clear and deployment details are concrete. However, the quantitative claim has a direct internal inconsistency: Table 1's Bronze rates imply 16% purity, not the advertised 30–50%. This is the most load-bearing weakness because it affects the headline numbers and the meaning of the Bronze stream. The issue is not merely an external flux/systematics caveat; it is a contradiction between the definition of signalness and the expected rates in the same paper. The reader's CONDITIONAL verdict is appropriate: the system seems real and the concern is fixable, but the paper must reconcile Table 1 with the Bronze definition before the advertised purities are used. I agree partially with the reader's weakest_assumption: their stated weakest assumption concerns flux and background dependence, but their rationale already notes the Bronze purity contradiction; my stress test promotes that internal contradiction to the primary concern. I would not change the verdict, hence UNCHANGED.","tokens_in":6139,"tokens_out":6427,"duration_ms":66219,"concrete_test":"Use the same Monte Carlo samples used for Table 1 to compute the distribution of per-event signalness (Eq. 2.1) for all simulated events passing the Bronze selection, and average the signalness over the expected rate. If the mean is ≈16% rather than lying in [30%,50%], the signalness is miscalibrated and the Bronze channel does not deliver what is advertised. Independently, recompute the integrated expected purity from the Table 1 entries: 2.8/(2.8+14.7)≈16%, and check whether any documented threshold setting can reconcile this with a 30% signalness cut.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline Bronze claim is contradicted by its own expected-rate table. Section 2 and the abstract define the Bronze channel as events with signalness between 30% and 50% (Eq. 2.1), so the expected astrophysical purity of Bronze alerts should be in that range. Table 1, however, lists expected Bronze signal plus background rates of 2.8 + 14.7 = 17.5 events/yr, giving an astrophysical fraction of 2.8/17.5 ≈ 16%. Under Eq. 2.1 the signalness is the cumulative fraction of signal events among all events above the energy proxy for that declination; if a cut is set where this fraction is 30%, the integrated fraction of signal in the selected sample cannot be below 30%, and a 30–50% window should be higher. A measured 16% implies either the Bronze threshold is not actually set at 30% signalness, the signalness values are not calibrated posterior probabilities, or Table 1 does not refer to the same Bronze selection described in the text. Any of these possibilities invalidates the advertised Bronze purity and the central rate/purity comparison. This defect is internal and independent of the external spectral-index uncertainty the authors acknowledge.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes the next-generation IceCube realtime alert system deployed in June 2019, which introduces two alert channels: a Gold channel for candidates at least 50% likely to be astrophysical and a Bronze channel for candidates between 30% and 50% likely to be astrophysical. It presents the three event selections (GFU, HESE, and EHE), the design of the alert message and GCN circular, and expected and observed alert rates computed from the best-fit diffuse astrophysical neutrino flux and simulated atmospheric backgrounds. The central claims are that the new system will deliver roughly 10 Gold and 20 Bronze alerts per year, with higher astrophysical purity and improved message content compared with the previous generation of alerts.","tokens_in":6318,"tokens_out":7933,"duration_ms":78091,"significance":"If the system performs as described, this is a useful operational advancement for multi-messenger astronomy: the explicit two-channel design with signalness thresholds, the public alert message format, and the historical comparison using seven years of IceCube data are all strengths. The comparison with observed historical rates provides some empirical grounding for the rate predictions, which reduces circularity concerns despite the signal model being fitted to the same data. However, the advertised Bronze purity is not supported by the paper's own Table 1, and the absence of uncertainties on the rate table makes the central quantitative claims difficult to evaluate. The value of the paper is mainly as a system description and performance forecast, and the technical issues are correctable in revision.","major_comments":[{"comment":"The abstract and Section 2 define Bronze alerts as candidates with signalness between 30% and 50%, and Eq. (2.1) is described as the probability that an event is astrophysical. However, Table 1's expected Bronze rates give an astrophysical fraction of 2.8/(2.8+14.7), which is approximately 16%, well below the advertised range. This internal inconsistency undermines the central Bronze purity claim. Please either revise the Bronze threshold definition, clarify that Eq. (2.1) is a cumulative sample fraction rather than a per-event posterior (and adjust the text accordingly), or correct Table 1 so that the expected signal fraction matches the stated Bronze definition.","section":"Abstract and Section 2; Table 1"},{"comment":"The expected and observed alert rates in Table 1 are quoted as point values with no uncertainties. This matters because the expected signal rates and all signalness values are computed from the best-fit diffuse flux (spectral index -2.19, normalization 1.01e-18 GeV^-1 cm^-2 s^-1 sr^-1) and from simulated atmospheric backgrounds, while Section 2.2 acknowledges that signalness varies with spectral index but does not quantify the effect. The comparison between expected and observed rates (e.g., Gold 12.7/yr expected vs 9.9/yr observed; Bronze 17.5/yr expected vs 19.5/yr observed) is therefore unquantified. Please add statistical uncertainties from the seven-year historical sample and a statement of the systematic spread induced by the assumed spectral index and background model.","section":"Section 2.3; Table 1"}],"minor_comments":[{"comment":"Typo: 'over the Iridum satellite' should read 'over the Iridium satellite.'","section":"Section 2, first paragraph"},{"comment":"The statement that 'all of the information is generated during the reconstructions performed at South Pole' appears to conflict with Section 2, which says that signalness is assessed in the North after the event data are transferred; please clarify where each alert quantity is computed.","section":"Section 2.2, last paragraph"},{"comment":"Please specify whether Nsignal(E,δ) and Nbackground(E,δ) count events with energy proxy greater than E and whether these are rates per year or counts; currently 'above' is ambiguous.","section":"Equation (2.1)"},{"comment":"In the text, 'Figure4' is missing a space, and the figure caption uses 'Thru-going' where 'Through-going' is meant.","section":"Figure 4 and text"},{"comment":"The table would be easier to read if the Gold and Bronze columns separated signal and background totals more clearly and if the Gold overlap between GFU and EHE were denoted in the row layout rather than only in the caption.","section":"Table 1"}],"recommendation":"major_revision","confidential_remarks":"The main technical obstacle is the internal inconsistency between the advertised Bronze purity and Table 1; this is a load-bearing issue but appears fixable in revision by clarifying the definition of signalness and recomputing or rewording the rate/purity claims. The manuscript is otherwise a straightforward system description and I see no novelty or scope concerns."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi,\n\nQuick take: the paper is a straightforward operational report on IceCube's new two-channel realtime alert system, deployed June 2019. The Gold/Bronze split, the new GFU BDT track selection, and the revamped GCN message format are real and clearly described. If you use IceCube alerts for multi-messenger follow-up, this is the reference for interpreting the new streams.\n\nThe main problem is internal. The Bronze channel is defined as events with signalness between 30% and 50% (Eq. 2.1). Table 1, however, lists expected Bronze signal and background rates of 2.8 and 14.7 per year, which is a 16% astrophysical fraction. A sample where every event has signalness at least 30% cannot have a combined purity of 16%. The weighted average of the per-event signalness values would have to be below the threshold, which is impossible. So either the table is mislabeled, the signalness values used for classification are different from the rates, or the Bronze selection is not actually applying the 30% cut. As written, the advertised Bronze purity is not supported by the paper's own numbers.\n\nSmaller issues: the expected rates come from the collaboration's best-fit diffuse flux, so they are not independent predictions; the table has no uncertainties; and the signalness is a cumulative fraction, not a per-event posterior probability, which deserves clearer emphasis for the astronomy consumers of the alerts. The observed total historical rates roughly match the expectations, so the problem is specifically in the signal/background split.\n\nWho should read this: astronomers planning follow-up observations and anyone designing the next alert tier. The operational details are valuable. It deserves a serious referee. Before the quoted purities are used, the authors should reconcile Table 1 with Eq. 2.1 or explicitly state what the Bronze column represents. With that fix, the paper is solid.\n\nBest,","headline":"Useful description of IceCube's new Gold/Bronze alert system, but the Bronze purity claim is contradicted by the paper's own Table 1.","tokens_in":6879,"tokens_out":6573,"would_cite":true,"duration_ms":62572,"reading_group":"yes","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"IceCube's upgraded realtime alert system will issue about 10 Gold and 20 Bronze neutrino alerts per year, with the Gold channel reserved for events at least 50% likely to be astrophysical.","keywords":["IceCube","realtime alerts","neutrino astronomy","multi-messenger astronomy","signalness","Gold and Bronze channels","diffuse astrophysical neutrino flux","GCN notices"],"falsifier":"Compare the actual one-year alert stream after deployment with the Table 1 predictions: if the observed Gold rate (in total, not per selection) is far from ~10 per year, or if archival classification of the Bronze events shows that fewer than 30% are truly of astrophysical origin, the assumed flux or background model would be invalidated.","tokens_in":5901,"feed_emoji":"🔭","tokens_out":3222,"duration_ms":33102,"temperature":0.7,"pith_summary":"This paper describes the next generation of IceCube's public realtime neutrino alert system, deployed on June 17, 2019. The new system introduces two alert channels: a Gold channel for neutrino candidates with at least 50% signalness (expected ~10 alerts per year) and a Bronze channel for candidates with 30–50% signalness (expected ~20 alerts per year). The goal is to give the multi-messenger astronomy community more, cleaner, and better-characterized neutrino candidates for follow-up observations, building on the success of the TXS 0506+056 association. The paper argues that this will improve the chance of identifying astrophysical neutrino sources.","feed_headline":"IceCube's new alerts: ~10 Gold, ~20 Bronze neutrinos per year","feed_subtitle":"Upgraded realtime system flags neutrino candidates with 50% or 30% signalness for faster multi-messenger follow-up.","key_machinery":"The central object is the signalness formula, Signalness(E, δ) = N_signal(E, δ) / (N_signal(E, δ) + N_background(E, δ)), which assigns each candidate a probability of being astrophysical based on the assumed diffuse flux and simulated atmospheric backgrounds. Three event selections feed this formula: GFU uses boosted decision trees to pick well-reconstructed through-going muon tracks and applies declination-dependent energy thresholds; HESE selects starting tracks with an outgoing muon and a 200-meter track length; EHE requires 4000+ photoelectrons and a fit-quality cut. These selections define the effective area and the Gold/Bronze rate tables.","core_discovery":"The central claim is that IceCube can simultaneously increase the rate and the astrophysical purity of realtime neutrino alerts by splitting the alert stream into two tiers defined by a per-event signalness measure. Signalness is defined as the ratio of expected astrophysical signal events to expected total events at that declination and energy proxy, using the best-fit diffuse astrophysical neutrino flux with spectral index -2.19 and normalization 1.01e-18 $GeV^{-1}$ $cm^{-2}$ $s^{-1}$ $sr^{-1}$ at 100 TeV. The updated selection uses three event samples—GFU (boosted decision tree selected through-going tracks), HESE (starting tracks with a muon), and EHE (extreme high energy tracks)—and the Gold/Bronze cuts are set at 50% and 30% signalness respectively. The paper reports expected alert rates of about 10 Gold and 20 Bronze events per year, with higher purity and clearer message content than the previous system.","pith_inferences":["The quoted Gold/Bronze purities are conditional on the assumed astrophysical spectral index of -2.19; if the true diffuse flux is harder or softer, the same event could cross the 50% or 30% thresholds differently, so the real-time classification is inherently model-dependent.","A natural test of the system's calibration is to accumulate one year of archival alerts and compare the empirically measured fraction of events that later show multi-messenger counterparts with the claimed signalness values.","The Bronze channel effectively creates a public low-threshold sample that could be used to search for time-clustered neutrino flares (like the 2014-2015 TXS 0506+056 episode) even when no single event is high-purity.","Because the EHE selection is held unchanged from the previous system, the main improvement in alert rate comes from adding GFU and loosening HESE cuts; this suggests future gains may come from even lower-energy track selections or from improving the southern-sky charge-based energy estimator."],"forward_implications":["If the flux and background assumptions hold, the astronomy community will receive roughly three times as many realtime neutrino alerts per year as before, with each alert carrying a quantified astrophysical signalness.","The Gold/Bronze classification gives follow-up observers a principled way to prioritize targets: Gold alerts warrant immediate multi-wavelength observation, while Bronze alerts still merit attention as moderate-purity candidates.","The reported angular error, energy, and signalness in each GCN Notice should allow more efficient and homogeneous follow-up campaigns compared with the previous alert stream.","The expected declination dependence means most alerts will be concentrated in the Northern sky, guiding the pointing strategies of optical, X-ray, and gamma-ray telescopes."],"supporting_citations":[{"why":"Establishes the high-energy astrophysical neutrino flux that motivates the alert program.","marker":"[1]"},{"why":"Defines the High-Energy-Starting-Event (HESE) selection that the updated HESE track selection is built on.","marker":"[2]"},{"why":"Describes the previous realtime alert system whose infrastructure and lessons are extended here.","marker":"[7]"},{"why":"The TXS 0506+056 association that demonstrated the scientific payoff of realtime neutrino alerts.","marker":"[8]"},{"why":"The EHE event selection from the first PeV neutrino analysis, kept unchanged for the new alert system.","marker":"[11]"},{"why":"Supplies the best-fit diffuse astrophysical neutrino flux (spectral index -2.19) used in the signalness calculation and rate predictions.","marker":"[12]"},{"why":"IceCube point-source analyses whose track selections and background estimates inform the new GFU selection.","marker":"[4, 10]"}],"fun_headline_variants":["IceCube splits neutrino alerts into Gold and Bronze tiers","New IceCube alerts: ~10 Gold, ~20 Bronze per year","IceCube's two-tier alerts: Gold and Bronze for faster follow-up","IceCube's new alert tiers: higher purity, more neutrinos","Gold and Bronze: IceCube's faster, richer neutrino alerts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The signalness values and alert rates rely on the assumption that the true diffuse astrophysical neutrino flux follows a power law with spectral index -2.19 and normalization 1.01e-18 $GeV^{-1}$ $cm^{-2}$ $s^{-1}$ $sr^{-1}$ at 100 TeV, and that the simulated atmospheric neutrino and muon backgrounds accurately match the real detector; if those are wrong, the Gold/Bronze classifications and quoted purities shift.","fun_headline_variants_meta":{"raw":{"variants":["IceCube splits neutrino alerts into Gold and Bronze tiers","New IceCube alerts: ~10 Gold, ~20 Bronze per year","IceCube's two-tier alerts: Gold and Bronze for faster follow-up","IceCube's new alert tiers: higher purity, more neutrinos","Gold and Bronze: IceCube's faster, richer neutrino alerts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000665,"raw_usage":{"total_tokens":3063,"prompt_tokens":998,"completion_tokens":2065,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":614,"completion_tokens_details":{"reasoning_tokens":1990}},"tokens_in":614,"tokens_out":2065,"duration_ms":13980,"temperature":1.0,"reasoning_tokens":1990,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:31:05.831857+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the actual one-year alert stream after deployment with the Table 1 predictions: if the observed Gold rate (in total, not per selection) is far from ~10 per year, or if archival classification of the Bronze events shows that fewer than 30% are truly of astrophysical origin, the assumed flux or background model would be invalidated.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The EHE event selection from the first PeV neutrino analysis, kept unchanged for the new alert system."},{"cited_title":"Haack and C","cited_arxiv_id":null,"evidence_quote":"Supplies the best-fit diffuse astrophysical neutrino flux (spectral index -2.19) used in the signalness calculation and rate predictions."}],"review_version":1}