{"id":"afe2dfe4-e66f-4fa7-ade8-a99d733ab0b4","arxiv_id":"2608.12458","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"No 129 nearby galaxies show signs of Dyson-swarm waste heat; quiet galaxies allow covering-fraction limits below 0.3%, with a population bound under 2.6% at 25% starlight coverage.","lead":"This study models 129 nearby galaxies to search for the infrared waste heat that galaxy-spanning Dyson swarms would emit, and it finds no candidates. It sets the first per-galaxy upper limits, down to a median of 0.3% of starlight in quiet galaxies, and lays out a practical path for much larger surveys.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Injection-recovery calibrations are closed-loop, so the quoted detection thresholds and population bounds may not hold for real galaxy SEDs.","rationale":"The paper is transparent, releases code, and the null result on real photometry is a solid finding. However, the central sensitivity statements and the population bound are explicitly injection-calibrated, and the injection is generated from the same forward model used for fitting. This is a standard but real limitation: it validates internal consistency, not robustness to model misspecification. The reader identified the spectral-shape assumption as the weakest point; my concern is adjacent but distinct, namely that even within the assumed model family the calibration may be optimistic because the injected data are noiseless model spectra rather than observed photometry with realistic residuals. A concrete re-injection test on real photometry would settle whether the detection thresholds and f < 2.6% bound survive. Because the null result and per-galaxy upper limits are not what is primarily threatened, I would not reject, but the headline sensitivity claims should be conditional on this check.","tokens_in":43384,"tokens_out":8438,"duration_ms":85540,"concrete_test":"Repeat the injection-recovery of Sections 2.2 and 3.1 using the real observed 24-band photometry as the base data vector: take each galaxy's AGENT-off best-fit model, add the Equation 4 AGENT excess at alpha = 5, 10, and 25%, add realistic per-band noise and optionally the observed AGENT-off residual pattern, then refit with the identical pipeline and recompute the recovery slope and the Delta BIC > 8 detection fractions. If N_det or the alpha ~ 4-5% threshold changes by more than about 30%, the population bounds and sensitivity claims must be revised accordingly.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The load-bearing point is not the single-blackbody spectral assumption itself (which the paper scopes explicitly), but that the 1,419 injection-recovery tests used to calibrate the headline sensitivity are closed-loop. Appendix A states that injected photometry is generated with the same AGENT-enabled forward model used in the fits, as a self-consistent reprocessing of each galaxy's best-fit spectrum rather than an addition to the observed broadband points. The injection tests in Sections 2.2 and 3.1 therefore validate that the pipeline can recover an AGENT signal when the data were produced by the identical FSPS+AGN+dust model, with no model misspecification and no empirical residual structure. The quoted thresholds (alpha ~ 4-5% in quiescent galaxies, N_det = 19/28/112, and the resulting f < 14.6%/10.1%/2.6% population bounds) are calibrated on those tests. If real galaxies deviate from the assumed model at the few-percent level, e.g., through AGN torus diversity, unusual dust geometries, or SPS template errors, the true detection efficiency could be lower, and the population bound would weaken. The null result from the real photometry (no Delta BIC > 0) and the per-galaxy upper limits are less affected, but the abstract's central 'few-percent rarity' claim inherits the closed-loop calibration.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an SPS-based search for galaxy-spanning Dyson-sphere waste heat in 129 nearby galaxies from the Brown et al. (2014) atlas. The AGENT formalism is implemented inside FSPS at the level of simple stellar populations (Eq. 4), so reprocessed starlight self-consistently affects nebular and dust emission. Using Prospector, the authors run 1,419 injection-recovery tests at T_BB = 300 K, recover injected covering fractions with slope m = 0.92, and calibrate Bayesian model-selection thresholds (ΔBIC > 8). On real photometry, no galaxy prefers a Dyson-sphere component; the paper reports the first per-galaxy 95% upper limits on warm swarms, with median α95 < 0.3% for quiescent non-AGN hosts, and converts zero detections into a population bound f < 2.6% for α = 25% swarms at 95% confidence. The paper also develops a resolved-fitting demonstration on M77 with nuclear excision, identifies characteristic failure modes in AGN fraction and star-formation history when waste heat is unmodeled, and lays out a tiered follow-up strategy including Balmer-decrement and dynamical-mass diagnostics, PRIMA, and JWST/MIRI.","tokens_in":43582,"tokens_out":4068,"duration_ms":40518,"significance":"If the central null result and sensitivity calibration hold, this is a substantial advance: it replaces population-level WISE color cuts with per-galaxy SED fits, improves quiescent-galaxy sensitivity by more than an order of magnitude over Griffith et al. (2015), and provides the first individually calibrated upper limits on galaxy-scale waste heat. The paper is also strong on reproducibility: public forks of FSPS, python-fsps, and Prospector are provided; the injection-recovery campaign, nested-sampling cross-checks, and likelihood-refinement procedure are described in enough detail to reproduce. The falsifiable predictions, such as the Balmer-decrement/IR-excess plane and the hidden-mass offset, are genuinely useful for future searches. The main caveat is that the injection-recovery calibration is closed-loop: injections and fits use the same forward model, so the quoted detection efficiencies and population bounds measure internal consistency rather than robustness to SPS, AGN-torus, or dust-template misspecification.","major_comments":[{"comment":"The injection-recovery calibration that sets the headline sensitivity is closed-loop: Appendix A states that injected photometry is generated with the same AGENT-enabled forward model used in the fits, as a self-consistent reprocessing of each galaxy's best-fit spectrum. The m = 0.92 recovery and the N_det = 19/28/112 counts used to derive f < 14.6%/10.1%/2.6% therefore validate internal consistency, not robustness to the kinds of model misspecification that could affect real SEDs (e.g., imperfect AGN torus templates, unusual dust geometries, SPS template errors). I request an explicit statement that the quoted detection thresholds and population bounds are conditional on the FSPS+AGN+dust model being correct, and ideally an out-of-model test, such as injecting into residualized or perturbed photometry, to quantify how much efficiency could degrade.","section":"§3.1, §3.2, Appendix A"},{"comment":"The population bound f < 2.6% is calibrated only for 300 K injections, while Appendix A shows that for star-forming spirals and peculiar galaxies the 100 K and 600 K injections are nearly undetectable until α ≈ 50%. The abstract's language 'warm (T_BB ≳ 100 K) swarms' is broader than what the injection-calibrated detection counts support. Please either scope the population bound explicitly to T_BB ≈ 300 K, or provide a temperature-averaged sensitivity estimate before claiming a bound on warm swarms generally.","section":"§3.2 and Appendix A"},{"comment":"The claim that the 18-parameter model 'does not overfit' is supported by the absence of spurious detections at zero injection, but this test shares the closed-loop limitation of the injection suite: it shows that the pipeline does not overfit data generated by the same model. A stronger statement would require a null test on data with realistic residual structure, for example fits to the real photometry with randomized residual draws, or a report of the empirical distribution of ΔBIC under the actual observed residuals.","section":"§2.3.1, §3.2"}],"minor_comments":[{"comment":"In the paragraph on resolved fitting, 'we show blow that the swarm component will absorb its MIR excess' appears to contain a typo; 'blow' should be 'below'.","section":"§2.4"},{"comment":"In the model-selection paragraph, 'an AGNET-off pair' should read 'an AGENT-off pair'.","section":"§3.2"},{"comment":"The sentence 'Their error budget to was dominated by the fact that they had no star count for each those galaxies' contains typographical errors; it should read 'Their error budget was dominated by the fact that they had no star count for each of those galaxies.'","section":"§4.4"},{"comment":"The SPHEREx citation has a garbled accent ('O. Dor´ e'), and the name should be rendered as 'Doré'.","section":"§4.5"},{"comment":"The table caption states that no galaxy reaches ΔBIC > 8 and also that no galaxy attains ΔBIC > 0. The latter is the stronger and more important statement, so consider stating it explicitly in the main text near the ranking discussion rather than only in the table caption.","section":"Table 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid contribution to the technosignature literature and the comparison with Huang et al. (2026) is fair and clearly argued. The main issue is that the sensitivity calibration is closed-loop, which is a caveat rather than a fatal flaw; the authors can address it in revision with a clear scope statement and, desirably, a robustness test. I would be comfortable with acceptance after such a revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is the first Dyson-sphere search that puts the AGENT reprocessing inside FSPS at the SSP level, so the swarm dims the UV-optical continuum and re-emits as a blackbody before the light reaches gas and dust. That is a real step beyond the WISE color cuts that dominated earlier work, and it changes the inference qualitatively: per-galaxy posterior limits on alpha, not just sample-level flux caps. The injection-recovery campaign is well executed (1,419 fits, slope m=0.92, >95% within 1 sigma), the null result is clean, and the median alpha<0.3% upper limit for quiescent non-AGN hosts is respectable. They also ship public code forks and a machine-readable ranking table; that is more than most papers in this area do.\n\nThe stress-test concern is real. Appendix A says explicitly that injected photometry is generated with the same AGENT-enabled forward model used in the fits, so the quoted detection thresholds (alpha~4-5% for quiescent galaxies, and the f<2.6% population bound at alpha=25%) calibrate internal consistency, not external realism. If real SEDs deviate from the model at the few-percent level, through AGN torus diversity, unusual dust geometry, or SPS template errors, those numbers would move. The paper's own M77 test shows how severe the integrated-light AGN degeneracy can be: the log-uniform AGN prior lets the fitter hand the MIR excess to the Dyson sphere component. Resolved excision brings the outer-disk alpha back to 0.10 with uncertainty that includes zero, which is reassuring but only demonstrated for one galaxy.\n\nThat said, the central claim survives. The null result does not depend on the injection calibration, and the per-galaxy upper limits are stated within the model assumptions. The paper scopes itself to warm (T_BB >= 100 K) single-blackbody swarms and explicitly flags the cold branch as unfalsifiable. The closed loop is a real limitation, not a fatal one.\n\nThe citation pattern is fine: the series, Wright et al., Lacki, Huang, and the rest are the relevant literature, and the code forks are acknowledged. This paper deserves a serious referee. The method is novel in a way that matters for future survey-scale searches, and the code release makes it checkable. My main requested revision would be to say more plainly in the abstract and the population-bound sections that the detection thresholds and f<2.6% bound are calibrated by closed-loop injection and should be read as best-case numbers under the model. A skimmer could over-read them otherwise. I would bring it to reading group and I would cite it in my own work on galaxy-scale technosignature limits.","headline":"A well-executed, properly scoped null result; the closed-loop injection calibration means the few-percent sensitivity numbers are best-case, but the core search and per-galaxy limits stand.","tokens_in":44244,"tokens_out":3919,"would_cite":true,"duration_ms":38000,"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":"None of 129 nearby galaxies shows the warm infrared signature of a galaxy-spanning Dyson swarm; new per-galaxy limits place median covering below 0.3% in quiescent hosts and cap the population at 2.6%.","keywords":["technosignatures","Dyson spheres","extragalactic SETI","waste heat","stellar population synthesis","galaxy SED fitting","mid-infrared excess","quiescent galaxies"],"falsifier":"Take a quiescent elliptical from the atlas, inject an $\\alpha = 5\\%$, 300 K swarm into its 24-band photometry, and refit with the same pair of models: the BIC $> 8$ test must flag it in roughly half of trials; then take an archival optical spectrum of any of the ten most anomalous candidates and look for high infrared excess together with H$\\alpha$/H$\\beta$ near the Case B value of about 2.86, since a galaxy showing both would be a genuine waste-heat candidate that this null result excludes.","tokens_in":2132,"feed_emoji":"🔭","tokens_out":2761,"duration_ms":90508,"temperature":0.7,"pith_summary":"This paper tests whether any of 129 nearby galaxies radiates the waste heat a galaxy-spanning civilization would leave behind, by fitting each galaxy's full ultraviolet-to-mid-infrared spectrum with a model that includes an optional Dyson-swarm component. The answer is no: no galaxy prefers that component, and the fits place the first per-galaxy 95% upper limits on warm swarms, with a median covering fraction below 0.3% for quiescent hosts without a dominant active nucleus. Injection-recovery tests show the method would catch swarms intercepting about 4–5% of starlight in quiescent galaxies, an order of magnitude deeper than earlier color-cut searches, and the null result bounds the population: fewer than 2.6% of atlas-like galaxies can hide 25%-covering swarms at 95% confidence. The search matters because it converts galaxy-scale SETI from crude color cuts into a calibrated measurement, and it identifies quiescent galaxy outskirts as the best place to look next.","feed_headline":"No Dyson-sphere glow found in 129 galaxies","feed_subtitle":"Per-galaxy SED fitting sets the first upper limits, catching swarms that cover just 4–5% of starlight.","key_machinery":"The load-bearing object is the AGENT Dyson-sphere parameterization injected at the simple-stellar-population level, before light propagates into the galaxy's gas and dust. In its two-parameter form ($\\alpha$ and $T_{\\rm BB}$, with non-thermal losses set to zero), each stellar population is replaced by $$$L^{{\\rm SSP,DS}}$_\\nu(t,Z)=(1-\\$\\alpha$)$L^{{\\rm SSP}}$_\\nu(t,Z)+\\left[\\frac{\\$\\alpha$\\,\\pi B_\\nu(T_{\\rm BB})}{\\sigma_{\\rm SB}T_{\\rm BB}^4}\\right]\\int $L^{{\\rm SSP}}$_\\nu(t,Z)\\,d\\nu,$$ so a fraction $\\alpha$ of starlight is dimmed in the UV–optical while a Planck bump at $T_{\\rm BB}$ is added in the infrared. Because the reprocessing happens inside the stellar population, nebular emission and dust emission respond self-consistently, giving the model an energy-conserving signature that separates waste heat from starbursts and AGN rather than relying on population-level color cuts.","core_discovery":"The paper's central claim is a null result with quantified reach: fitting 129 nearby galaxies with a stellar-population model that can absorb a fraction $\\alpha$ of starlight and re-emit it as a Planck spectrum finds no galaxy that prefers such a Dyson sphere component. On the real photometry, the AGENT-on fits return the first per-galaxy 95% upper limits on warm ($T_{\\rm BB} \\gtrsim 100$ K) swarms, with median $\\alpha_{95} < 0.3\\%$ across quiescent hosts without a dominant AGN. The accompanying 1,419 injection-recovery fits recover injected covering fractions with slope 0.92 and reach Bayesian detection at $\\alpha \\approx 4$–$5\\%$ in quiescent galaxies, while star-forming spirals need $\\alpha \\approx 20\\%$; converting the zero detections through injection-calibrated detection rates gives population bounds of $<14.6\\%$, $<10.1\\%$, and $<2.6\\%$ of atlas-like galaxies hosting 300 K swarms at $\\alpha = 5\\%$, $10\\%$, and $25\\%$, respectively, at 95% confidence. In short, warm, galaxy-scale waste heat appears rare in the local universe at the few-percent level.","pith_inferences":["The IR-excess versus Balmer-decrement plane and the stellar-to-dynamical-mass offset, calibrated in the paper's appendices, can be run on archival spectra for millions of galaxies as an inexpensive pre-screen; the paper does not itself apply them at scale.","The two failure modes of the waste-heat-free model are a general diagnostic: any large SED-fitting catalog showing unexpected AGN or star-formation spikes may be flagging unmodeled mid-infrared excess, not necessarily technosignatures.","If fast amortized inference is extended to the roughly $10^7$ galaxies with archival multiwavelength photometry, the population bound of a few percent should tighten by orders of magnitude, and the same injection tests would validate it.","The single-temperature Planck assumption is the main blind spot: a civilization that radiates waste heat across many temperatures, in non-thermal channels, or anisotropically would evade both detection and the quoted covering-fraction limits."],"forward_implications":["With the full 24-band atlas, quiescent galaxies are detectable hosts at $\\alpha \\approx 4$–$5\\%$, while star-forming spirals require $\\alpha \\approx 20\\%$, so future surveys should prioritize the outskirts of quiescent galaxies.","The zero detections plus injection-calibrated counts imply that fewer than 14.6% ($\\alpha = 5\\%$), 10.1% ($\\alpha = 10\\%$), and 2.6% ($\\alpha = 25\\%$) of atlas-like galaxies can host 300 K swarms at 95% confidence.","An unmodeled swarm leaves two falsifiable artifacts: inferred AGN fractions inflate by up to three orders of magnitude, and recent star formation in quiescent galaxies inflates by 1.4–1.8 dex, giving cheap red-flag diagnostics.","Resolved fitting with nuclear excision cuts the recovered covering fraction for the extreme AGN host M77 from 0.182 to 0.100, demonstrating that AGN contamination can be separated spatially.","Adding wide-field far-infrared photometry plus mid-infrared imaging would turn an ambiguous AGN-versus-swarm case into a decisive test, raising an example from 1.2$\\sigma$ to 3.7–4.5$\\sigma$ and pushing the model-selection statistic well past the detection threshold."],"supporting_citations":[{"why":"Supplies the AGENT formalism relating $\\alpha$, $\\gamma$, $\\epsilon$, $\\nu'$, and $T_{\\rm BB}$ that the paper embeds in stellar population synthesis.","marker":"J. T. Wright et al. 2014b"},{"why":"Provides the 129-galaxy, 24-band matched-aperture atlas that is the entire observed sample and the injection-recovery target.","marker":"M. J. I. Brown et al. 2014"},{"why":"Supplies the Bayesian SED-fitting model architecture and priors that the paper extends with the AGENT parameters.","marker":"J. Leja et al. 2017"},{"why":"Provides the base stellar population synthesis code that generates the simple stellar populations to which AGENT reprocessing is applied.","marker":"C. Conroy et al. 2009"},{"why":"Sets the previous WISE color-based limits at roughly 50–80% covering fractions that this work claims to surpass by an order of magnitude.","marker":"R. L. Griffith et al. 2015"},{"why":"Offers the contemporaneous WISE/CatWISE flux caps that the paper contrasts as non-per-galaxy and unable to detect a waste-heat component.","marker":"B.-L. Huang et al. 2026"},{"why":"Provides the theoretical motivation for partial cloaking, where selectively covering less luminous stars leaves detectable photometric signatures.","marker":"B. C. Lacki 2019"}],"fun_headline_variants":["No Dyson-sphere glow in 129 galaxies, limits set","129 galaxies show no sign of alien drain on starlight","Dyson swarm search leaves no galaxy glowing with industry","Null result: no Dyson spheres in 129 galaxies, but limits on waste heat","First per-galaxy upper limits on Dyson swarm coverage, all null"],"cache_read_input_tokens":46208,"weakest_assumption_plain":"The limits presume every galaxy-spanning swarm absorbs and re-emits starlight as a single isotropic Planck spectrum at one temperature; a civilization radiating waste heat across broad temperatures, in non-thermal channels, or anisotropically would not be recovered and would not be constrained by the quoted covering fractions.","fun_headline_variants_meta":{"raw":{"variants":["No Dyson-sphere glow in 129 galaxies, limits set","129 galaxies show no sign of alien drain on starlight","Dyson swarm search leaves no galaxy glowing with industry","Null result: no Dyson spheres in 129 galaxies, but limits on waste heat","First per-galaxy upper limits on Dyson swarm coverage, all null"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000714,"raw_usage":{"total_tokens":3336,"prompt_tokens":1193,"completion_tokens":2143,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":809,"completion_tokens_details":{"reasoning_tokens":2051}},"tokens_in":809,"tokens_out":2143,"duration_ms":15262,"temperature":1.0,"reasoning_tokens":2051,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:09:02.806636+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a quiescent elliptical from the atlas, inject an $\\alpha = 5\\%$, 300 K swarm into its 24-band photometry, and refit with the same pair of models: the BIC $> 8$ test must flag it in roughly half of trials; then take an archival optical spectrum of any of the ten most anomalous candidates and look for high infrared excess together with H$\\alpha$/H$\\beta$ near the Case B value of about 2.86, since a galaxy showing both would be a genuine waste-heat candidate that this null result excludes.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the theoretical motivation for partial cloaking, where selectively covering less luminous stars leaves detectable photometric signatures."}],"review_version":1}