{"id":"56b2685e-45a9-4216-9c11-5c90ebc56a4f","arxiv_id":"1908.10669","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A survey of 37 exoplanet transmission spectra finds molecular features are typically muted to 33% of clear solar atmospheric model predictions.","lead":"The authors measured the strength of water absorption in 37 exoplanet atmospheres observed with the Hubble Space Telescope. They found the typical signal is only about one third as strong as a simple clear-sky model predicts, so future observations should be planned with weaker features in mind.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 33% median may be an artifact of low-SNR and non-detected water fits against a fixed clear-solar isothermal template, rather than a true population-level muting factor.","rationale":"The reader's weakest assumption is that the ATMO clear-solar template fitted by p0 measures true muting for all 37 planets. My concern overlaps with that but is more specific: the population statistic is generated from individual least-squares fits whose uncertainties are large for non-detections, and the paper does not provide enough information to assess how many fits are unconstrained. This is not a mere presentation issue: if many p0 values are consistent with zero at 1-2 sigma, the reported 33% median is not a robust property of the sample. A censored hierarchical analysis is the natural, concrete check. The reader's conditional verdict already emphasizes that the precise value depends on the template and missing per-planet data, so my concern does not change the overall verdict; it sharpens the condition under which the paper should be accepted. I do not regard the paper as fraudulent or internally inconsistent; the statistical framing is plausible, but the evidence as presented is insufficient to separate a real muted-amplitude population from a selection/fitting effect. Hence the verdict remains CONDITIONAL, equivalently UNCHANGED relative to the reader's assessment.","tokens_in":2821,"tokens_out":4122,"duration_ms":45529,"concrete_test":"Recover the 37 publicly available spectra and their fitted p0/sigma_p0 values, then recompute the population median using a hierarchical Bayesian model that treats non-detections as censored upper limits and reports the posterior median only from spectra with p0/sigma_p0 >= 3. If the resulting median rises above about 50%, the reported 33% muting is driven by non-detections and low-SNR fits rather than by the atmospheres themselves.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim depends entirely on the per-planet amplitude scale factor p0 in the fit S1 = (S0 * p0) + p1, where S0 is a 1D isothermal, solar-metallicity, solar-C/O ATMO clear model. For WFC3/G141 data at typical precision, many spectra do not show a significant water feature; a least-squares fit constrained to a positive template will return p0 near zero with a large uncertainty. The paper then samples each planet's p0 from a normal distribution with width sigma_p0, truncating at -60% to 150%, and reports the resulting median. Poorly constrained fits with large sigma_p0 are therefore heavily represented in the tails, which can drag the collective median downward even if the true feature amplitudes are larger. The manuscript does not report the target list, per-planet p0 values, sigma_p0 values, or how many of the 37 fits are detected at, say, >2 sigma, so the reader cannot distinguish a genuine population-level muting from a fitting artifact. A second, related issue is that the fixed template cannot absorb non-isothermal temperature structure, non-solar C/O, or wavelength-dependent cloud opacity; if those are present, p0 is not a clean measure of muting. The headline number 33 +/- 24% is thus load-bearing on an assumption that is not checked in the paper.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes 37 exoplanet transmission spectra observed with HST WFC3 G141. Each spectrum is fit with a 1D isothermal ATMO clear-solar, solar-C/O model that is scaled to the planet's parameters, using the relation S1 = (S0 x p0) + p1, where p0 is the model amplitude scale factor and p1 is a baseline offset. The authors then treat the per-planet p0 and its uncertainty as a normal distribution, sample 5000 points per planet, and report a global median amplitude of 33 ± 24% of the expected clear-solar model. They further state that clear-solar molecular features are measured in <7% of cases, and recommend that observers plan for muted transmission features rather than assuming maximum clear-solar amplitudes.","tokens_in":3067,"tokens_out":3605,"duration_ms":34990,"significance":"If the result holds, it provides a practically important population-level guidance for planning transmission spectroscopy observations: a typical close-in giant planet's 1.1–1.7 μm water feature is about one-third the strength predicted by a clear solar-metallicity model. The paper is commendably concise, defines its fitting model explicitly, and uses a publicly available model grid. The central weakness is that the headline number is an in-sample summary of fitted scale factors, and the manuscript does not provide the per-planet fit values, target list, or detection-significance information needed to separate a genuine population-level muting from a low-SNR fitting artifact. As a forecast, the result is also template-dependent, since the fixed isothermal clear-solar model may absorb non-isothermal structure, hazes, or non-solar C/O into the fitted p0.","major_comments":[{"comment":"The manuscript does not report the list of 37 targets, the per-planet fitted amplitude p0, its uncertainty σp0, or the number of fits for which water absorption is detected at a given significance. Without these values the reader cannot distinguish a genuine population-level muting from a fitting artifact in which low-SNR, non-detected spectra return p0 ≈ 0 with large σp0 and then dominate the tails of the sampled distribution. Please provide a supplemental table or appendix with target identifications, p0, σp0, mean data uncertainty, and detection significance for each of the 37 fits.","section":"Methods (paragraph defining S1 = (S0*p0) + p1) and Fig. 1"},{"comment":"The headline 'forecast' is the median of p0 values fitted to the same 37 spectra that define the sample; it is a summary statistic of the fitting results, not an independent prediction for future observations. The text should either explicitly frame the result as an in-sample population statistic or demonstrate predictive validity through a holdout analysis or comparison with an external sample; otherwise the wording 'forecast' overstates the evidence.","section":"Methods and Summary (global distribution sampling)"},{"comment":"The assumed template is a 1D isothermal, solar-metallicity, solar-C/O, clear, no-scattering ATMO model. Any departure from these assumptions in a real atmosphere—non-isothermal temperature structure, haze/cloud opacity, or non-solar C/O—will be absorbed into the fitted multiplier p0, so p0 is not a clean measure of feature muting unless the template shape is validated. The paper should quantify the sensitivity of p0 to plausible template variations (e.g., metallicity, C/O, temperature structure, or added uniform opacity) to show that the 33% median is not an artifact of the chosen template.","section":"Methods (ATMO model template)"},{"comment":"The criterion behind the statement that 'clear solar molecular features are measured in <7% of cases' is only introduced later as '≥70% or ≥2H', and the detection threshold appears arbitrary. Please define the detection criterion up front, state how many of the 37 fits qualify as constrained, unconstrained, and detected H2O (as labeled in Fig. 1), and report the uncertainty on the <7% fraction.","section":"Abstract and Summary ('<7% of cases')"}],"minor_comments":[{"comment":"The median amplitude is quoted as '33 ± 24%' in the text and '33 ± 25%' in the figure's internal label; please make the numbers consistent.","section":"Figure 1 caption and text"},{"comment":"The TEPCat database is credited in a footnote but is not in the reference list; please add a formal citation. Also give the full Goyal et al. reference if accessed before publication.","section":"References"},{"comment":"Figure 1 contains several duplicated labels and panels; please replot with a cleaner layout and larger font so the example fit, histogram, and probability-density panels are legible.","section":"Figure 1 layout"}],"recommendation":"major_revision","confidential_remarks":"The central claim is plausible and practically useful, but the manuscript as submitted cannot be evaluated quantitatively because the per-planet fit results are not reported. I recommend requiring a supplemental data table as a condition of acceptance. The paper is a short Letter-style report; the missing data and detection definitions are the main blockers."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a short research note that gives observers a useful prior—transmission features are typically smaller than clear-model predictions—but the exact 33% figure is less solid than it seems. The paper's real value is the compilation: 37 WFC3/G141 transmission spectra, fit with a consistent ATMO grid, and a simple linear scaling model. That is worth having. The median amplitude of 0.89H is in line with the more conservative 1.4H from Fu et al. 2017, and it's a sensible number to put in exposure time calculators. The soft spots are concentrated around the statistics. First, the per-planet p0 and sigma_p0 values are not given anywhere. The entire result is a histogram made from those fits, but the reader can't check the fits or the sample. The public spectra alone don't reproduce the analysis without significant effort. For a population-level claim, that's a real reproducibility gap. Second, the quoted 33±24% is the 16th–84th percentile of the combined sampled distribution—the spread of the population, not the error on the median. With 37 planets, the uncertainty on the median is roughly 24/sqrt(37) ≈ 4%, so conflating the two is misleading. Third, and most importantly, the analysis treats non-detections as measurements near zero. A fit to a noisy spectrum with no detectable water will return p0 ≈ 0 with a large sigma; the paper then samples from that normal and includes it in the collective distribution. That biases the median downward. The paper has a figure with categories like 'Detected H2O' and 'Unconstrained' but the text doesn't say how many fall in each bin, nor how many p0 values are inconsistent with zero at 2 sigma. Without that, the headline number could be an artifact of averaging in a lot of upper limits. The template shape is a secondary concern: a 1D isothermal clear-solar model is a convenient reference, but non-isothermal structure or non-solar C/O would change the shape and make p0 a less clean measure of muting. None of this kills the qualitative conclusion—there is plenty of independent evidence that many giant planet atmospheres have muted features—but it means the 33% is a rough prior, not a measured population statistic. If this lands in a refereed journal, I'd send it out but require the target list, the fitted p0/sigma_p0 table, and a treatment of censored non-detections before accepting. It's still worth a read for anyone planning JWST transmission spectroscopy.","headline":"Useful rule-of-thumb paper—features are typically muted—but the 33% number is built on best-fit amplitudes that include non-detections, and the per-planet fits are not published.","tokens_in":3615,"tokens_out":6113,"would_cite":true,"duration_ms":58131,"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":"Transmission spectra of 37 exoplanets show molecular features are muted to 33 ± 24% of clear-solar model predictions.","keywords":["exoplanet atmospheres","transmission spectroscopy","molecular absorption","water features","clear solar metallicity","clouds and hazes","signal-to-noise ratio","population statistics"],"falsifier":"A concrete test would be to re-fit the same 37 spectra with a retrieval that allows non-isothermal temperature structure, clouds and hazes, and non-solar C/O, and compare the recovered water feature amplitudes to the clear-template $p_0$ values; if the flexible retrievals return amplitudes near 100% for a large fraction of the sample, the 33% median is a template artifact. A simpler observational check is to measure the 1.4 µm water amplitude directly in a subset of these planets at higher signal-to-noise (e.g., with a larger-aperture infrared telescope) and see whether the detected amplitudes scatter around 33% or instead cluster near the clear-model prediction.","tokens_in":2598,"feed_emoji":"🪐","tokens_out":10595,"duration_ms":90221,"temperature":0.7,"pith_summary":"This paper asks how much molecular absorption in exoplanet transmission spectra actually departs from the optimistic 'clear solar metallicity' model that many observing proposals assume. The authors fit 37 publicly available transmission spectra in the 1.1–1.7 µm range with a one-dimensional isothermal, cloud-free, solar-composition ATMO model, allowing only an amplitude scale factor and a baseline offset to vary. The central result is that the fitted amplitudes cluster at a median of 33 ± 24% of the clear model's prediction, and features reaching at least 70% of the clear model are measured in fewer than 7% of the planets. If this is right, observers planning exposure times around clear solar feature strengths will routinely overestimate the signal-to-noise ratio and underestimate the observing time needed by roughly a factor that the 33% median implies. The paper's practical prescription is to design future observations, including space-based transit spectroscopy, around muted molecular features from the outset.","feed_headline":"Exoplanet water features are 33% as strong as clear-model forecasts","feed_subtitle":"A 37-planet infrared survey finds clear-sky features in under 7% of cases: plan for fainter signals.","key_machinery":"The central mechanism is the amplitude scale factor $p_0$ defined by the linear fit $S_1 = (S_0 \\times p_0) + p_1$, where $S_0$ is the fixed clear-solar ATMO model spectrum, $p_0$ multiplies the molecular feature amplitude, and $p_1$ is a wavelength-flat offset. The ATMO grid (Goyal et al. 2019) supplies one-dimensional isothermal, solar-metallicity, solar-C/O, no-scattering, no-uniform-opacity models as the 'clear' baseline, chosen by equilibrium temperature and surface gravity and then rescaled to each planet's stellar and planetary radius. The statistical payload is the population distribution of $p_0$: each planet contributes a Gaussian with mean $p_0$ and width $\\sigma_{p_0}$, 5000 Monte Carlo draws per planet build the sample distribution, and the 16th/50th/84th percentiles give the 33 ± 24% result. The scale-height correlation (median 0.89 ± 0.77 H) translates the amplitude into a physically interpretable atmospheric unit, connecting to the 1.4 H value from Fu et al. (2017).","core_discovery":"The paper's discovery, stated on its own terms, is that molecular absorption features in the 1.1–1.7 µm transmission spectra of 37 close-in exoplanets observed with HST/WFC3 G141 are muted to 33 ± 24% of the amplitude predicted by clear solar metallicity model atmospheres. To reach this, each spectrum is fit with $S_1 = (S_0 \\times p_0) + p_1$, where $S_0$ is a fixed ATMO 1D isothermal cloud-free solar-metallicity model scaled to the planet's radius, gravity, and equilibrium temperature, $p_0$ is the free amplitude, and $p_1$ is a baseline offset. The distribution of $p_0$ across the sample is built by sampling each planet's Gaussian uncertainty, giving a population median of 33 ± 24% and a scale-height-equivalent median of 0.89 ± 0.77 H. The authors also report that only ~7% of the sample shows features at or above 70% of the clear model, while ~30% of the time the observed feature is below 20% or 0.5 H. The paper frames this as a statistical forecast: clear-solar-assumption models overestimate achievable SNR about 93% of the time.","pith_inferences":["If the muting is largely due to clouds and hazes, the same template-scaling analysis applied to emission spectra or phase curves could reveal a different (likely larger) muting factor, since emission probes different pressure levels; a similar 33% figure in emission would be an independent test of the mechanism.","The correlation between $p_0$ and scale-height units suggests that a simple empirical 'muting factor' could be incorporated into time-allocation models as a function of equilibrium temperature or gravity, turning the population statistic into a predictive tool for individual targets.","A testable extension would be to compare $p_0$ for the same planets across different wavelength bands (e.g., optical vs 1.1–1.7 µm); if the muting factor is wavelength-dependent, it points to Rayleigh scattering or gray cloud decks, whereas a uniform factor favors high-altitude aerosols."],"forward_implications":["Observers using standard clear-solar amplitude assumptions will overestimate the molecular signal-to-noise ratio; exposure-time estimates for transmission spectroscopy should be recomputed for muted features, roughly tripling the needed precision for a given feature detection.","A large majority of hot-Jupiter transmission spectra will appear flat or weakly featured at 1.1–1.7 µm, so null detections of water in individual planets should not be interpreted as absence of water without a population-level muting prior.","The 33 ± 24% distribution provides a quantitative prior for atmospheric retrievals: retrieved abundances and cloud properties that assume clear templates will be biased, and future retrievals should marginalize over a muting scale factor.","The method's scale-height equivalent (0.89 ± 0.77 H) offers a simple forecasting rule: plan for molecular features of roughly one pressure scale height or less, not the two or more scale heights of a clear solar atmosphere."],"supporting_citations":[{"why":"Supplies the ATMO 1D isothermal clear-solar model grid used as the baseline template $S_0$ for all 37 fits.","marker":"Goyal et al. (2019)"},{"why":"Contributes a major subset of the WFC3 G141 transmission spectra and the detection of water features in several planets.","marker":"Sing et al. (2016)"},{"why":"Defines the reduction and spectral extraction methods used for unpublished datasets in the sample.","marker":"Stevenson et al. (2014)"},{"why":"Sets out the analysis methods applied to the public WFC3 G141 datasets.","marker":"Wakeford et al. (2016)"},{"why":"Supplies the example transmission spectrum and data points used in the model-fit illustration (black points in Fig. 1).","marker":"Wakeford et al. (2018)"},{"why":"Gives the prior estimate of 1.4 scale heights for feature amplitude that the measured 0.89 ± 0.77 H median is compared against.","marker":"Fu et al. (2017)"}],"fun_headline_variants":["Exoplanet spectral features are typically a third of clear-sky predictions","Muted exoplanet atmospheres: features are only 33% of clear-model strength","Most exoplanet spectra show 33% of predicted clear-sky feature amplitude","Expect exoplanet transmission features at 33% of clear-model strength","Survey finds exoplanet atmospheric features muted to one-third of predictions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a single clear-solar, isothermal, cloud-free template with solar carbon-to-oxygen ratio accurately represents the spectral shape of every planet in the sample, so that a fitted amplitude scale factor really measures muting of molecular features rather than errors introduced by an inappropriate template.","fun_headline_variants_meta":{"raw":{"variants":["Exoplanet spectral features are typically a third of clear-sky predictions","Muted exoplanet atmospheres: features are only 33% of clear-model strength","Most exoplanet spectra show 33% of predicted clear-sky feature amplitude","Expect exoplanet transmission features at 33% of clear-model strength","Survey finds exoplanet atmospheric features muted to one-third of predictions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001146,"raw_usage":{"total_tokens":4717,"prompt_tokens":873,"completion_tokens":3844,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":489,"completion_tokens_details":{"reasoning_tokens":3743}},"tokens_in":489,"tokens_out":3844,"duration_ms":26719,"temperature":1.0,"reasoning_tokens":3743,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:15:52.776959+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete test would be to re-fit the same 37 spectra with a retrieval that allows non-isothermal temperature structure, clouds and hazes, and non-solar C/O, and compare the recovered water feature amplitudes to the clear-template $p_0$ values; if the flexible retrievals return amplitudes near 100% for a large fraction of the sample, the 33% median is a template artifact. A simpler observational check is to measure the 1.4 µm water amplitude directly in a subset of these planets at higher signal-to-noise (e.g., with a larger-aperture infrared telescope) and see whether the detected amplitudes scatter around 33% or instead cluster near the clear-model prediction.","supporting_citations":[{"cited_title":"2019, , 482, 4503","cited_arxiv_id":null,"evidence_quote":"Supplies the ATMO 1D isothermal clear-solar model grid used as the baseline template $S_0$ for all 37 fits."},{"cited_title":"2014, , 147, 161","cited_arxiv_id":null,"evidence_quote":"Defines the reduction and spectral extraction methods used for unpublished datasets in the sample."},{"cited_title":"2016, The Astrophysical Journal, 819, 10","cited_arxiv_id":null,"evidence_quote":"Sets out the analysis methods applied to the public WFC3 G141 datasets."},{"cited_title":"2018, , 155, 29","cited_arxiv_id":null,"evidence_quote":"Supplies the example transmission spectrum and data points used in the model-fit illustration (black points in Fig. 1)."}],"review_version":1}