{"id":"2863aece-723c-461f-a2be-d2205fc720e3","arxiv_id":"2507.07840","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Kepler planets larger than about 3.5 Earth radii have higher average orbital eccentricities (0.20) than smaller planets (0.05), with the transition aligning with other demographic boundaries.","lead":"This paper measures the orbital shapes of 1,646 planets found by the Kepler telescope and finds that planets bigger than about 3.5 times Earth's radius have, on average, noticeably more elongated orbits than smaller planets. The size at which orbits become more eccentric matches known breakpoints in how common planets are and what kind of stars host them, suggesting two different ways planets form.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The size-eccentricity step could be a photoeccentric pipeline artifact; current injections recover an injected step but never test a flat-truth population, so a size-dependent duration/TTV systematic is not ruled out.","rationale":"I read the paper in good faith and find the analysis to be careful, large-sample, and mostly persuasive. The hierarchical model, the non-parametric eccentricity distribution, the radius-uncertainty weighting, and the injection/recovery tests are genuine strengths. The central claim is not internally inconsistent, and the alignment with occurrence and metallicity breaks is an interesting interpretive point, but that alignment is not independent evidence for the eccentricity measurement itself. My primary concern is narrower than the reader's: the reader flags TTV/detrending systematics as the weakest assumption, and I agree that those are the right class of systematics, but I would sharpen the issue. The current injection/recovery suite validates recovery of an injected step-like trend; it does not validate the null case of a flat eccentricity-size relation. If a size-dependent bias in transit duration, residual TTV smearing, or light-curve detrending exists, the pipeline could manufacture the reported step. A flat-truth injection spanning the full size range would settle this directly. Because the test is feasible and the paper already has the infrastructure to run it, I would not change the conditional verdict: accept the measurement as promising but require the flat-truth injection (and ideally a high-SNR split) before treating the step as established. My agreement is partial because the reader's concern is about possible systematics, whereas mine is specifically about the missing falsification test; the two are complementary rather than identical.","tokens_in":28805,"tokens_out":18779,"duration_ms":226851,"concrete_test":"Run the full ALDERAAN + hierarchical-eccentricity pipeline on an injection set with a single flat ground-truth eccentricity distribution for all planet sizes: draw e.g. e from half-normal with sigma=0.05 (or a single beta distribution with mean ~0.10) for simulated planets spanning R_p = 0.5-16 R_Earth and P = 1-100 days, with realistic TTVs, photometric noise, and the same stellar properties as the real sample; then recover <e> vs R_p exactly as in the paper. If the recovered curve shows a step of amplitude comparable to the reported ~0.15 increase near 3.5 R_Earth at >95% confidence, the headline trend is suspect. If the recovered curve is consistent with flat, the concern is resolved. A complementary check is to split the real sample by transit S/N and long-cadence smearing and confirm the size-eccentricity step persists in the high-SNR subsample.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that mean population eccentricity rises from ~0.05 below ~3.5 R_Earth to ~0.20 above it. This claim depends entirely on the individual eccentricity posteriors being unbiased as a function of planet size. The photoeccentric inversion (Methods: 'Extracting eccentricity constraints from the transit model'; Eq. 1 and SI Eq. 15) attributes any mismatch between the measured transit duration T_14 and the circular, center-crossing duration T_0 to eccentricity, after correcting for TTVs and stellar density. The paper itself warns (Methods, TTV paragraph) that 'unaccounted for TTVs will smear out transit ingress and egress and bias eccentricity measurements.' The injection/recovery tests in the SI are real evidence and show good recovery for the simulated populations, but they simulate a step-like ground truth: e_small from a half-normal with sigma=0.03 and e_large from a beta distribution with mean ~0.23, then verify recovery of that step. This demonstrates that the pipeline can recover an injected trend; it does not demonstrate that the pipeline would not create a spurious trend if the true eccentricity distribution were flat in planet size. A size-dependent systematic in residual TTV smearing, GP detrending, or long-cadence smearing could bias the recovered T_14 more for large or small planets, producing the observed step even if the intrinsic population is flat. The radius-valley injection test (SI Figs. 17-20) is not a substitute, since it only covers 1-4 R_Earth with a flat e=0.023 input and does not span the full 0.5-16 R_Earth range of the headline claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper measures eccentricities for 1646 Kepler planets with periods 1–100 days and radii 0.5–16 R⊕, using the photoeccentric effect: transit durations from the ALDERAAN pipeline are compared to duration predictions from Gaia stellar densities via importance sampling, producing per-planet posterior samples in (e, ω). The authors then infer population eccentricity distributions with an approximate hierarchical Bayesian model using a non-parametric regularized histogram, later compressed into a two-parameter empirical template. The central claim is that mean population eccentricity rises from ⟨e⟩ = 0.05 ± 0.01 for planets smaller than about 3.5 R⊕ to ⟨e⟩ = 0.20 ± 0.03 for larger planets, and that this transition aligns with breaks in occurrence rate and host-star metallicity, implying distinct formation channels. Additional findings include a tentative (2.1σ) elevation of eccentricity for planets in the radius valley, and a single-to-multi eccentricity ratio of about 2.5 with similar ⟨e⟩–R_p curves for both populations.","tokens_in":29222,"tokens_out":8686,"duration_ms":102941,"significance":"If the size–eccentricity step is real, it is an important demographic signature that connects planet occurrence, composition, and dynamical history, and it sharpens the evidence for distinct formation channels around 3.5 R⊕. The paper's strengths are substantial: the analysis pipeline is open source; the authors run a large injection/recovery campaign; they check robustness across distributional choices (empirical, beta, half-Gaussian), binning schemes, selection cuts, and detection-bias corrections; they forward-model duration ratios; and they honestly report the weak 2.1σ radius-valley signal. The per-planet eccentricity constraints are weak (σ(e) ≈ 0.3), so the population claims rest on the ensemble analysis being unbiased as a function of planet radius. The main weakness is that the injection/recovery tests validate recovery of a step-like ground truth but do not test the null hypothesis of a size-independent eccentricity distribution; this missing test is directly relevant to the central claim.","major_comments":[{"comment":"The validation simulations inject a step-like ground truth: small planets (Rp < 4 R⊕) drawn from a half-normal with σ = 0.03 (⟨e⟩ ≈ 0.023) and large planets drawn from a beta distribution with mean ≈ 0.23. The authors then show that the recovered ⟨e⟩ values are consistent with these injected values. This demonstrates that the pipeline can recover an injected step, but it does not test whether the pipeline could create a spurious step from a population whose eccentricity distribution is the same at all sizes. Because individual eccentricity posteriors have σ(e) ≈ 0.3, the reported rise from 0.05 to 0.20 is an ensemble-level statement; a size-dependent bias in measured T14 from, e.g., unaccounted TTV smearing, long-cadence smearing, or GP detrending could in principle produce the observed trend even if the intrinsic population is flat in planet size. The authors should inject a population with a single global f(e) (or size-independent template parameters) for all radii and show that the recovered ⟨e⟩–R_p relation is flat; analogous tests with a mild intrinsic slope would also constrain the pipeline response. This null test is the most important missing validation for the central claim.","section":"SI, Injection-and-recovery tests (Table 4, Figs. 15–20)"},{"comment":"The detection-bias correction in the hierarchical likelihood uses only the geometric transit probability, (1−e²)/(1+e sinω). However, Kepler's detection probability also depends on transit duration and depth, which are correlated with e and ω through the same photoeccentric relation. The injection/recovery tests do not simulate the full planet detection process: they inject synthetic transits into existing Kepler light curves and then fit them, so the recovered posteriors are conditioned on the transit being present and fitted, not on the original detection selection. The claim that Eq. (5) yields the intrinsic (rather than observed) eccentricity distribution is therefore not fully validated by the presented tests. This could affect the absolute scale of ⟨e⟩, and possibly the size-dependent trend if duration- and depth-dependent detectability vary with planet radius. I ask the authors to either add a duration-dependent detection term to Eq. (5) or demonstrate through a forward-modeled survey simulation (including the detection step) that the geometric correction alone is sufficient for the claimed precision.","section":"Methods, Eq. (5); SI, Eq. (22)"}],"minor_comments":[{"comment":"The sentence \"We first select a sample of 1209 single Sun-like stars hosing 1646 planets\" contains a typo: \"hosing\" should be \"hosting\".","section":"Methods, first paragraph"},{"comment":"The phrase \"has be remarkably successful at at explaining\" contains duplicated words; please correct to \"has been remarkably successful at explaining\".","section":"Introduction, paragraph 2"},{"comment":"The phrase \"Planets in the the radius valley\" has a duplicated \"the\".","section":"Conclusions, item 3"},{"comment":"The caption reads \"for each of our five plannet size classes\"; \"plannet\" should be \"planet\".","section":"SI, Fig. 11 caption"},{"comment":"The clause \"a narrow band of low occurrence rate density which separates rocky ``super-Earths'' (1.0–1.5 Earth-radii) from gas-rich ``sub-Neptunes'' (2.0–3.0 Earth-radii\" is missing a closing parenthesis after \"Earth-radii\".","section":"Abstract"},{"comment":"The regularization term added to the log-likelihood, ln L = −(ν−h)², is ad hoc and not derived from the data; please justify its form and strength, or show that results are unchanged when this term is omitted (Figure 14 partially addresses this via alternative parametric models, but not for the template fit itself).","section":"SI, after Eq. (26)"},{"comment":"The paper does not report the number of planets per size bin; given that the largest bins (R0 ≈ 14.6 R⊕) contain very few planets, the sigmoid fit parameters and their uncertainties would be more interpretable with effective sample sizes listed.","section":"Methods, Table 1"},{"comment":"The combination of tail probabilities P(A_singles ≤ 0) and P(A_multis ≤ 0) treats the two measurements as independent, but they share the same empirical template and are drawn from the same parent population according to the paper's own conclusion; please discuss the sensitivity of the 2.1σ figure to this independence assumption.","section":"SI, Estimating the significance of the eccentricity peak in the radius valley"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong demographic study with a transparent pipeline and extensive internal validation, but the headline claim—a sharp rise in mean eccentricity at ~3.5 R⊕—rests on the absence of a spurious-trend test. The injection/recovery campaign recovers an injected step but never tests a flat-truth population, which is the natural null for a claim of a step. This is fixable within the manuscript's scope (additional simulations) and should be requested before acceptance. The radius-valley 2.1σ feature is appropriately hedged and should not block publication. The selection-function concern in Eq. (5) is more involved but can be addressed by a duration-dependent treatment or a forward-modeled detection simulation; if the authors can show the size-eccentricity step survives such a treatment, the paper would be suitable for a high-visibility journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline claim—mean eccentricity rising from ~0.05 below 3.5 Earth radii to ~0.20 above—is probably real, but the paper doesn't fully close the door on a pipeline artifact. The injection/recovery tests are thorough, but they only simulate the step they're trying to detect. They never inject a flat eccentricity distribution across all sizes and ask whether the pipeline would create a step on its own. The stress-test note is right: a size-dependent bias in TTV smearing or long-cadence smearing could masquerade as a trend in recovered durations, and this null scenario is simply not tested.\n\nWhat is genuinely new and good: this is the largest full-transit-profiling eccentricity sample by an order of magnitude—1646 Kepler planets—and the resulting quantitative <e>–Rp relation with a fitted transition at 3.3–4.2 Earth radii is new. The ALDERAAN pipeline is described in enough detail to be trusted, and the authors did careful work: iterative TTV correction, GP detrending with transit exclusion windows, per-planet visual inspection, and a battery of robustness checks across distributional forms, binning schemes, and geometric detection biases. They also honestly flag the radius-valley eccentricity excess as only 2.1-sigma, and they provide code and a machine-readable empirical distribution table. The consistency with prior single/multi eccentricity offsets and the alignment of the transition with independent occurrence and metallicity breaks give me some confidence the step is astrophysical rather than an artifact.\n\nThe soft spots, in order: first, the missing flat-truth injection test is a real gap and should be addressed before publication—it's a straightforward experiment. Second, the paper never reports a significance or model comparison for the step itself; they fit a sigmoid but don't compare to a constant <e> model. Third, individual eccentricity constraints are weak (sigma_e ~ 0.3), so the population inference leans heavily on the hierarchical model; that's a known limitation of the method, and the paper handles it about as well as possible. Fourth, the data products are not pinned with a version or DOI, making exact reproduction harder than it should be.\n\nNone of these are fatal. The paper is a serious, honest measurement that deserves a referee. I would ask for the null injection test and a formal step-significance claim, but I'd expect the main result to survive. This is useful for anyone building planet formation models around the small-planet/large-planet dichotomy, and it will get cited.","headline":"A careful, large-sample photoeccentric eccentricity measurement that suggests a real size-eccentricity step, but the missing flat-truth injection test leaves a nagging systematic worry.","tokens_in":29725,"tokens_out":3815,"would_cite":true,"duration_ms":46180,"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":"This paper establishes that the mean orbital eccentricity of close-in planets rises from about 0.05 to 0.20 at a radius near 3.5 Earth radii, matching breaks in occurrence rate and host-star metallicity and pointing to distinct formation…","keywords":["exoplanets","orbital eccentricity","photoeccentric effect","Kepler mission","hierarchical Bayesian inference","radius valley","planet formation","transit timing variations"],"falsifier":"Re-measure eccentricities for a sample spanning the 3-4 Earth-radius transition using an independent technique, such as radial velocities or asteroseismic stellar densities, and compare the inferred $\\langle e \\rangle$ values; if the jump disappears or shifts to another radius, the photoeccentric result would be an artifact of the transit-duration or stellar-density assumptions.","tokens_in":28604,"feed_emoji":"🪐","tokens_out":7083,"duration_ms":67958,"temperature":0.7,"pith_summary":"This paper makes the case that the average orbital eccentricity of close-in planets changes sharply with planet size: the population mean rises from $\\langle e \\rangle = 0.05 \\pm 0.01$ for planets below about 3.5 Earth radii to $\\langle e \\rangle = 0.20 \\pm 0.03$ for larger planets. The evidence comes from eccentricity constraints for 1646 Kepler planets, 92% of them smaller than Neptune, derived by comparing transit light-curve shapes with Gaia-based stellar densities. The authors point out that this same size threshold marks abrupt changes in planet occurrence rate and host-star metallicity, which together suggest separate formation channels for small and large close-in planets. They also find that the eccentricity distribution peaks at zero for every planet size and report a tentative peak in eccentricity among radius-valley planets at $2.1\\sigma$ significance.","feed_headline":"Eccentricity jumps from 0.05 to 0.20 above 3.5 Earth radii","feed_subtitle":"Kepler data on 1646 planets tie the jump to breaks in occurrence and metallicity, pointing to two formation channels.","key_machinery":"The paper uses the photoeccentric effect: a planet's transit duration relative to a circular, center-crossing transit depends on eccentricity $e$ and argument of pericenter $\\omega$ through the factor $(1 + e \\sin\\omega)/\\sqrt{1-e^2}$, so a duration anomaly can be converted into a per-planet eccentricity constraint once the stellar density is known. It couples full light-curve fitting, including Gaussian-process detrending and iterative transit-timing-variation correction, to an importance-sampling scheme that combines transit posteriors with Gaia stellar-density priors, producing joint posterior samples of $\\{e,\\omega\\}$ for each planet. A hierarchical Bayesian model with a flexible, regularized histogram and an empirical template then turns these noisy per-planet constraints into intrinsic population eccentricity distributions while accounting for geometric detection biases.","core_discovery":"The central discovery is a size-dependent population eccentricity: small planets below roughly 3.5 Earth radii orbit on nearly circular paths with $\\langle e \\rangle \\approx 0.05$, while larger planets have $\\langle e \\rangle \\approx 0.20$. The transition appears in both single- and multi-transiting systems, with singles uniformly about 2.4 times more eccentric than multis. Because the same radius threshold also separates a common, metal-insensitive population of small planets from a rare, metal-rich population of large planets, the paper interprets the eccentricity jump as evidence that the two size groups formed through distinct dynamical pathways. The eccentricity distribution is self-similar in shape across size classes, peaking at $e = 0$ and declining monotonically, so the size effect is primarily a change in the spread of the distribution rather than its functional form.","pith_inferences":["A natural extension is to test whether the 3.5 Earth-radius threshold shifts with stellar mass or orbital period, which would sharpen the formation-channel interpretation by separating intrinsic size effects from environmental ones.","The same hierarchical photoeccentric machinery could be applied to TESS light curves, where shorter baselines and higher noise may require combining many systems to recover the same size-eccentricity signal.","The paper's eccentricity-dilution argument implies that any small planet found with a confidently high eccentricity may be a marker for an unseen outer giant companion, a prediction that radial-velocity surveys could directly test.","If the radius-valley eccentricity peak is confirmed, it would motivate targeted searches for collision remnants or stripped cores in that radius range using atmospheric and density measurements."],"forward_implications":["Large and small close-in planets must be explained by separate formation channels, one that preserves low eccentricities for sub-Neptune-sized bodies and one that excites eccentricities for larger planets.","Models of planet occurrence must reproduce a coordinated break at roughly 3.5 Earth radii in eccentricity, occurrence rate, and host-star metallicity simultaneously.","If the radius-valley eccentricity excess is real, gap planets likely formed through collisions or atmospheric stripping, a prediction testable with radial-velocity follow-up or JWST secondary-eclipse measurements.","Single- and multi-transiting planets share one parent population, with singles representing the high-eccentricity, high-inclination tail, which bears on explanations of the Kepler dichotomy.","For planets above 4 Earth radii, the positive eccentricity-metallicity and eccentricity-period correlations constrain when and how giant planet eccentricities are excited relative to disk dispersal."],"supporting_citations":[{"why":"Supplies the importance-sampling procedure that converts transit shape posteriors and Gaia stellar-density priors into joint eccentricity samples.","marker":"44"},{"why":"Provides the hierarchical Bayesian formalism used to infer population eccentricity distributions from per-planet posteriors.","marker":"48"},{"why":"Supplies the uniform transit-timing-variation catalog used to correct transit times, which the paper argues is essential to avoid smearing ingress and egress.","marker":"31"},{"why":"Provides the homogeneous Gaia-Kepler stellar properties catalog, including the stellar densities used as priors in the photoeccentric inversion.","marker":"26"},{"why":"Defines the radius-gap location and period dependence used to compute adjusted radii for the radius-valley eccentricity analysis.","marker":"29"},{"why":"Documents the sharp drop in planet occurrence rate above about 3.5 Earth radii that the paper aligns with the eccentricity transition.","marker":"57"},{"why":"Establishes the metallicity preference of large planets versus small planets that the paper aligns with the eccentricity break.","marker":"58"},{"why":"Previous full transit-profiling eccentricity measurements for small planet systems provide the methodological baseline and comparison sample.","marker":"11"}],"fun_headline_variants":["Kepler finds planet eccentricity jumps at 3.5 Earth radii","Eccentric planets are big: size threshold at 3.5 Earth radii","Two formation channels revealed by planet eccentricity jump","Small planets circular, large planets elliptical: 3.5 Earth radii","Eccentricity spike marks planets above 3.5 Earth radii"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that, after Gaussian-process detrending and transit-timing-variation correction, the measured transit durations are unbiased and the Gaia-based stellar densities are accurate, so any leftover duration anomaly is a real eccentricity signal rather than a size-dependent systematic error.","fun_headline_variants_meta":{"raw":{"variants":["Kepler finds planet eccentricity jumps at 3.5 Earth radii","Eccentric planets are big: size threshold at 3.5 Earth radii","Two formation channels revealed by planet eccentricity jump","Small planets circular, large planets elliptical: 3.5 Earth radii","Eccentricity spike marks planets above 3.5 Earth radii"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000514,"raw_usage":{"total_tokens":2553,"prompt_tokens":1055,"completion_tokens":1498,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":671,"completion_tokens_details":{"reasoning_tokens":1405}},"tokens_in":671,"tokens_out":1498,"duration_ms":11963,"temperature":1.0,"reasoning_tokens":1405,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:32:11.242208+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-measure eccentricities for a sample spanning the 3-4 Earth-radius transition using an independent technique, such as radial velocities or asteroseismic stellar densities, and compare the inferred $\\langle e \\rangle$ values; if the jump disappears or shifts to another radius, the photoeccentric result would be an artifact of the transit-duration or stellar-density assumptions.","supporting_citations":[{"cited_title":"AJ 166, 61 (2023)","cited_arxiv_id":null,"evidence_quote":"Supplies the importance-sampling procedure that converts transit shape posteriors and Gaia stellar-density priors into joint eccentricity samples."},{"cited_title":"ApJ 725, 2166–2175 (2010)","cited_arxiv_id":null,"evidence_quote":"Provides the hierarchical Bayesian formalism used to infer population eccentricity distributions from per-planet posteriors."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the uniform transit-timing-variation catalog used to correct transit times, which the paper argues is essential to avoid smearing ingress and egress."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the homogeneous Gaia-Kepler stellar properties catalog, including the stellar densities used as priors in the photoeccentric inversion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the radius-gap location and period dependence used to compute adjusted radii for the radius-valley eccentricity analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the sharp drop in planet occurrence rate above about 3.5 Earth radii that the paper aligns with the eccentricity transition."},{"cited_title":"Nature 486, 375–377 (2012)","cited_arxiv_id":null,"evidence_quote":"Establishes the metallicity preference of large planets versus small planets that the paper aligns with the eccentricity break."}],"review_version":1}