{"claim": "Cost and validated accuracy are produced by separate communities and never jointly modeled across a population of radiometers: the NASA Instrument Cost Model (NICM) establishes that instrument development cost is a structured, predictable function of design attributes (its dependent variable is cost), so the cost-accuracy join the dissertation proposes is a genuine, currently-unfilled gap, not an artifact of a missed study. This grounds assurance-case subclaim S1 (the problem is real).", "evidence": [{"source": "Habib-Agahi, Mrozinski & Fox, 'Latest NASA Instrument Cost Model (NICM): Version VI', NASA JPL, NTRS 20160008251 (candidate ref [42])", "doi_or_url": "https://ntrs.nasa.gov/citations/20160008251", "grade": "C"}, {"source": "Stahl, 'Survey of cost models for space telescopes', Optical Engineering (2010) (candidate ref [46])", "doi_or_url": "https://doi.org/10.1117/1.3430603", "grade": "A"}], "facet": "empirics", "chapter": "ch3_literature_review", "subclaim": "real"}
{"claim": "The accuracy side of the join is a real, independently measured, uncertainty-quantified quantity defined product family by product family in the calibration/validation literature (SST against in-situ references, MODIS aerosol against AERONET expected-error envelopes), but that literature validates one product at a time and never relates accuracy to instrument cost or to a population frontier. This evidences the accuracy half of subclaim S1 (the problem is real).", "evidence": [{"source": "Merchant, Embury & Bulgin, 'Half a century of satellite remote sensing of sea-surface temperature', Remote Sensing of Environment (2019) (candidate ref [73])", "doi_or_url": "https://doi.org/10.1016/j.rse.2019.111366", "grade": "A"}, {"source": "Remer, Kaufman & Tanre, 'The MODIS Aerosol Algorithm, Products, and Validation', J. Atmospheric Sciences (2005) (candidate ref [85])", "doi_or_url": "https://doi.org/10.1175/jas3385.1", "grade": "A"}], "facet": "measurement", "chapter": "ch3_literature_review", "subclaim": "real"}
{"claim": "The problem is material because per-instrument capability choices are being made now, under fixed budget toplines, with a cost tool (NICM-class cost-estimating relationships and distributed-mission cost surveys) and no corresponding accuracy tool, exactly the value-centric tradespace decision the over-specification rule would inform. This grounds assurance-case subclaim S2 (the problem is material).", "evidence": [{"source": "Maier, Collopy & Hollingsworth, 'Value-Driven Design', J. Spacecraft and Rockets / AIAA (2012) (candidate ref [17])", "doi_or_url": "https://doi.org/10.2514/1.54033", "grade": "A"}, {"source": "Foreman, Le Moigne & de Weck, 'A Survey of Cost Estimating Methodologies for Distributed Spacecraft Missions', AIAA SPACE 2016 (candidate ref [40])", "doi_or_url": "https://doi.org/10.2514/6.2016-5245", "grade": "A"}, {"source": "Ross, Hastings, Warmkessel & Diller, 'Multi-Attribute Tradespace Exploration with Concurrent Design', AIAA (2003) (candidate ref [23])", "doi_or_url": "https://doi.org/10.2514/6.2003-1328", "grade": "A"}], "facet": "empirics", "chapter": "ch3_literature_review", "subclaim": "material"}
{"claim": "The cost regressor's reliability is itself a substantive empirical question that conditions the design: cost overruns are common, large, and fat-tailed rather than symmetric noise, so a single high-leverage tail-cost instrument can bend a flexibly fit g(cost) toward apparent curvature; this is why the decision rule must privilege out-of-sample (cross-validated-beats-linear-null) performance over in-sample shape and report leverage diagnostics. This grounds residual-risk subclaim S5 on the cost axis.", "evidence": [{"source": "Flyvbjerg, 'Five Things You Should Know About Cost Overrun', Transportation Research Part A (2018) (candidate ref [38])", "doi_or_url": "https://doi.org/10.1016/j.tra.2018.07.013", "grade": "A"}], "facet": "empirics", "chapter": "ch3_literature_review", "subclaim": "residual_risk"}
{"claim": "The balance-and-support apparatus the identification demands is specified but unexecuted: entropy balancing achieves exact covariate-moment balance across cost terciles by construction (the method Section 6.5 operationalizes), and the synthetic-control convex-hull principle converts extrapolation into interpolation only inside common support (the trimming Table 6.2 logs). Balancing neutralizes only observed confounders, so an unobserved driver raising both cost and accuracy survives all three steps; the pre/post SMDs and survivor count remain unpopulated by design and gated on the unresolved NICM instrument-level cost-table dependency.", "evidence": [{"source": "Hainmueller, 'Entropy Balancing for Causal Effects', Political Analysis (2012) (candidate ref [101])", "doi_or_url": "https://doi.org/10.1093/pan/mpr025", "grade": "A"}, {"source": "Abadie, 'Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects', J. Economic Literature (2021) (candidate ref [94])", "doi_or_url": "https://doi.org/10.1257/jel.20191450", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "residual_risk"}
{"claim": "The falsification standard for an observational cost effect is the program-evaluation conditional-independence discipline: a cost effect is credible only where confounders are held fixed (within difficulty-by-epoch cells) rather than read off the pooled cross-section. The candidate pre-commits to within-product-family re-estimation and difficulty-plus-epoch controls, but treats non-accuracy cost drivers (calibration-subsystem, contractor, build-site) only as measurement error to drop, not as placebo instruments, and no within-cell estimate has been produced.", "evidence": [{"source": "Imbens & Wooldridge, 'Recent Developments in the Econometrics of Program Evaluation', J. Economic Literature (2009) (candidate ref [105])", "doi_or_url": "https://doi.org/10.1257/jel.47.1.5", "grade": "A"}], "facet": "empirics", "chapter": "ch5_research_design", "subclaim": "mechanism"}
{"claim": "With few effectively-independent units, classical large-sample inference fails and credibility comes from design-based placebo/permutation inference, asking whether the realized statistic sits in the tail of a distribution built by re-running the method across donor units and reassigned labels. The candidate pre-commits to instrument-clustered errors, a whole-instrument cluster-bootstrap, and leave-one-instrument-out for aggregate curvature, but specifies no permutation distribution of the over-specification channel-count edge itself nor a randomized-cost-ordering reassignment test.", "evidence": [{"source": "Abadie, 'Using Synthetic Controls...', J. Economic Literature (2021) (candidate ref [94])", "doi_or_url": "https://doi.org/10.1257/jel.20191450", "grade": "A"}, {"source": "Abadie, Diamond & Hainmueller, 'Synth: An R Package for Synthetic Control Methods', J. Statistical Software (2011) (candidate ref [103])", "doi_or_url": "https://doi.org/10.18637/jss.v042.i13", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "mechanism"}
{"claim": "A cost effect free of technology vintage is identified only inside the common-support / overlap region of the cost-by-vintage plane, where instruments of different cost coexist at the same epoch; absent overlap, fitted curvature is interpolation across vintages and is not separable from a secular time trend. The candidate's own cited telescope cost models document a negative technology-epoch term (cost falls and capability rises with year), so cost (the regressor) and epoch (a control) co-move by the candidate's own account, which is the exact collinearity the identification must defeat through common-support trimming.", "evidence": [{"source": "Stahl et al., 'Multivariable parametric cost model for ground optical telescope assembly', Optical Engineering (2005) / Stahl & Henrichs SPIE (2016) (candidate ref family [41]/[46])", "doi_or_url": "https://doi.org/10.1117/1.2031216", "grade": "A"}, {"source": "Abadie, 'Using Synthetic Controls...', J. Economic Literature (2021) (candidate ref [94]); overlap/common-support necessary condition", "doi_or_url": "https://doi.org/10.1257/jel.20191450", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "The within-vintage-cell placebo (hold launch-year / reprocessing-generation fixed in a cell and re-test concavity) is the correct design-based test and coarsened exact matching is the established apparatus that operationalizes holding vintage fixed in a cell, but the candidate's robustness battery holds PRODUCT FAMILY fixed rather than VINTAGE fixed, names no within-vintage fake-cost-shock placebo, and pre-commits to no specific cell-level falsification threshold, so the within-vintage estimate that would distinguish a frontier from a secular technology trend does not exist.", "evidence": [{"source": "Iacus, King & Porro, 'Causal Inference without Balance Checking: Coarsened Exact Matching', Political Analysis (2012)", "doi_or_url": "https://doi.org/10.1093/pan/mpr013", "grade": "A"}, {"source": "Imbens & Wooldridge, 'Recent Developments in the Econometrics of Program Evaluation', J. Economic Literature (2009) (candidate ref [105])", "doi_or_url": "https://doi.org/10.1257/jel.47.1.5", "grade": "A"}], "facet": "empirics", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "Using a single sensor lineage as its own counterfactual (within a lineage where the geophysical retrieval and validation reference are held nearly constant) is the synthetic-control device for netting out the secular trend, and is the strongest available cure for between-lineage vintage heterogeneity; the candidate's within-family re-estimation is the right instrument for it, but the per-generation cost-and-accuracy trajectories for MODIS-to-VIIRS, the AMSR line, or a sounder series cannot be exhibited because the instrument-level cost-per-generation series is the unresolved data-access dependency and the table is unassembled.", "evidence": [{"source": "Abadie, 'Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects', J. Economic Literature (2021) (candidate ref [94])", "doi_or_url": "https://doi.org/10.1257/jel.20191450", "grade": "A"}], "facet": "rival", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "The dependent-variable construct (requirement-normalized accuracy) is admittedly forced rather than emergent: grounded theory requires the central construct be shown to fit and work in the substantive area by an incident-to-category audit trail, but the candidate concedes requirement-normalization is a theoretical commitment made before looking at the data and that it 'does not make the underlying references equally precise across families' (a kelvin of SST error against a drifting buoy and a hundredth of soil moisture against a core site forced onto one scale despite different fractional reference uncertainties).", "evidence": [{"source": "Glaser & Strauss, The Discovery of Grounded Theory (1967/2017)", "doi_or_url": "https://doi.org/10.4324/9780203793206", "grade": "A"}], "facet": "measurement", "chapter": "ch4_data_and_measurement", "subclaim": "residual_risk"}
{"claim": "By the constant-comparative test, requirement-normalized accuracy is an imposed equivalence rather than a saturated category: an AERONET expected-error compliance fraction, an SST bias-and-standard-deviation matchup, an SMAP unbiased-RMSE against sparse core sites, and a precipitation gauge comparison are not incidents of one underlying analytic property, and dividing each by its own requirement (a denominator defined by a different community, reference standard, and error model per family) does not make them comparisons of like incidents.", "evidence": [{"source": "Glaser, 'The Constant Comparative Method of Qualitative Analysis', Social Problems 12(4):436-445 (1965)", "doi_or_url": "https://doi.org/10.2307/798843", "grade": "A"}, {"source": "Glaser & Strauss, The Discovery of Grounded Theory (1967/2017)", "doi_or_url": "https://doi.org/10.4324/9780203793206", "grade": "A"}], "facet": "measurement", "chapter": "ch4_data_and_measurement", "subclaim": "residual_risk"}
{"claim": "Saturation is a falsifiable predictive statement about the unobserved and has four distinct conceptualizations in circulation, so it must be defended as earned rather than assumed; the candidate freezes the matched table at Step 1 and fixes the population a priori by a three-condition availability intersection (NICM cost record AND validation record AND passive radiometry), which is convenience sampling, the structural opposite of theoretical sampling, so no documented row exists at which new instruments ceased to yield new properties and the over-specification edge cannot be distinguished from the edge of the assembled census.", "evidence": [{"source": "Saunders et al., 'Saturation in qualitative research', Quality & Quantity 52(4):1893-1907 (2018)", "doi_or_url": "https://doi.org/10.1007/s11135-017-0574-8", "grade": "A"}, {"source": "Glaser & Strauss, The Discovery of Grounded Theory (1967/2017) (theoretical sampling)", "doi_or_url": "https://doi.org/10.4324/9780203793206", "grade": "A"}], "facet": "identification", "chapter": "ch4_data_and_measurement", "subclaim": "residual_risk"}
{"claim": "H1 is structurally a formal claim (one concave frontier and one over-specification channel count across radiometers), but grounded theory generates a formal claim only by comparison across substantive areas, and the candidate's own scope section withdraws universality across geophysical variables while Chapter 6 names family-level rivals (validation-reference ceiling, difficulty assignment); the family-wise comparison that would license a formal claim or force a retreat to a substantive claim with a stated boundary is named but only planned, so the pooled concavity is on present evidence an imposed aggregate.", "evidence": [{"source": "Glaser & Strauss, The Discovery of Grounded Theory (1967/2017) (substantive vs formal theory)", "doi_or_url": "https://doi.org/10.4324/9780203793206", "grade": "A"}, {"source": "'Space Economy and Sustainability: A Systematic Review', Sustainable Development (2025)", "doi_or_url": "https://doi.org/10.1002/sd.3383", "grade": "A"}], "facet": "rival", "chapter": "ch4_data_and_measurement", "subclaim": "alternatives"}
{"claim": "The over-specification edge is an independently computable forward-model quantity, not only a fitted second derivative: for an optimal-estimation retrieval the marginal information an added channel contributes is the degrees-of-freedom-for-signal read off the averaging kernels, derived from the forward-model Jacobian and measurement-error covariance independent of any outcome fit. Operational channel-selection methods already locate the DFS-saturation point for hyperspectral IR sounders, and the published MODIS aerosol forward model with AERONET-validated error envelopes makes a head-to-head feasible; on evidentiary grounds the forward-model DFS number is the trustworthy verifier and a noisy fitted second derivative over dozens of instruments should yield to it where they conflict.", "evidence": [{"source": "Maddy & Barnet (review), 'Optimal Estimation Retrievals and Their Uncertainties', BAMS (2020)", "doi_or_url": "https://doi.org/10.1175/bams-d-19-0027.1", "grade": "A"}, {"source": "Xu et al., 'A channel selection method for hyperspectral atmospheric infrared sounders based on layering', Atmospheric Measurement Techniques (2020)", "doi_or_url": "https://doi.org/10.5194/amt-13-629-2020", "grade": "A"}, {"source": "Worden et al., 'Predicted errors of TES nadir retrievals from spectral window selection', JGR Atmospheres (2004)", "doi_or_url": "https://doi.org/10.1029/2004jd004522", "grade": "A"}, {"source": "Remer et al., 'The MODIS Aerosol Algorithm, Products, and Validation', J. Atmos. Sci. (2005)", "doi_or_url": "https://doi.org/10.1175/jas3385.1", "grade": "A"}, {"source": "Hubert et al., 'Olympiad-level formal mathematical reasoning with reinforcement learning' (AlphaProof), Nature (2025); verifier-grounding standard", "doi_or_url": "https://doi.org/10.1038/s41586-025-09833-y", "grade": "A"}], "facet": "mechanism", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "A pre-registered negative case is groundable: for thermal-infrared SST and LST window retrievals the documented binding accuracy limit is radiometric calibration stability/drift and geolocation, not spectral-channel redundancy (SST is retrieved from few thermal-IR window channels and its multi-decade error budget is dominated by calibration stability). A forward-model DFS analysis of such a few-channel calibration-limited retrieval therefore predicts no channel-redundancy concave edge, so the candidate's fitted g(cost) must reproduce this exception rather than smear one concavity across heterogeneous physics.", "evidence": [{"source": "Minnett et al., 'Half a century of satellite remote sensing of sea-surface temperature', Remote Sensing of Environment (2019)", "doi_or_url": "https://doi.org/10.1016/j.rse.2019.111366", "grade": "A"}, {"source": "Li et al., 'Satellite Remote Sensing of Global Land Surface Temperature', Reviews of Geophysics (2022)", "doi_or_url": "https://doi.org/10.1029/2022rg000777", "grade": "A"}], "facet": "empirics", "chapter": "ch4_data_and_measurement", "subclaim": "alternatives"}
{"claim": "The technology-vs-redundancy discriminator is constructible and the literature supplies both arms: validated SST accuracy improves substantially across reprocessing/algorithm-version generations with the physical channel set held fixed (the disembodied-vintage arm, grounded in published detector-noise and radiometric-stability trends such as CrIS noise characterization and AIRS L1B radiance uncertainty), while at fixed vintage marginal accuracy from adding channels flattens as retrieval information content saturates. The disembodied delta is large and non-trivial, so it is a genuine confounder that must be regressed out, not assumed away, before concavity can be attributed to channel redundancy.", "evidence": [{"source": "Minnett et al., 'Half a century of satellite remote sensing of sea-surface temperature', Remote Sensing of Environment (2019)", "doi_or_url": "https://doi.org/10.1016/j.rse.2019.111366", "grade": "A"}, {"source": "Merchant et al., 'Satellite-based time-series of sea-surface temperature since 1981 for climate applications', Scientific Data (2019)", "doi_or_url": "https://doi.org/10.1038/s41597-019-0236-x", "grade": "A"}, {"source": "Han et al., 'Noise performance of the CrIS instrument', JGR Atmospheres (2013)", "doi_or_url": "https://doi.org/10.1002/2013jd020457", "grade": "A"}, {"source": "Pagano et al., 'SI-Traceability and Measurement Uncertainty of the AIRS Version 5 L1B Radiances', Remote Sensing (2020)", "doi_or_url": "https://doi.org/10.3390/rs12081338", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "An external physics-based verifier exists and should be pre-registered as the Hassabis discipline requires (a learned output is trustworthy only when grounded in an independent correctness oracle): the standard retrieval-theory channel-selection calculation computes degrees-of-freedom-for-signal / Shannon information content directly from channel bandpasses and an assumed forward model and covariance, and shows information content saturating as channels are added, with information content depending on spectrometer/bandpass design. Confirmation requires the regression's flattening channel count to coincide with the DFS-flat count; on divergence the physics verifier is trusted and the divergence diagnoses the regression as reading the technology/vintage curve.", "evidence": [{"source": "Xu et al., 'A channel selection method for hyperspectral atmospheric infrared sounders based on layering', Atmospheric Measurement Techniques (2020)", "doi_or_url": "https://doi.org/10.5194/amt-13-629-2020", "grade": "A"}, {"source": "'Hyperspectral Satellite Radiance Atmospheric Profile Information Content and Its Dependence on Spectrometer Design', IEEE JSTARS (2021)", "doi_or_url": "https://doi.org/10.1109/jstars.2021.3073482", "grade": "A"}, {"source": "Hubert et al., 'Olympiad-level formal mathematical reasoning with reinforcement learning' (AlphaProof), Nature (2025)", "doi_or_url": "https://doi.org/10.1038/s41586-025-09833-y", "grade": "A"}], "facet": "mechanism", "chapter": "ch5_research_design", "subclaim": "mechanism"}
{"claim": "The single population-average over-specification threshold is a Hayekian aggregate that can be wrong for every individual mission: the candidate concedes the retrieval-difficulty control is a coarse ordinal index that absorbs gross across-family difficulty differences but not the fine within-family gradients (the dispersed, condition-specific, partly-tacit knowledge of which retrieval question matters and how hard it is), and the within-family re-estimation is only a pre-specified sensitivity check conditioned on family size, so the coincidence of within-family thresholds is the candidate's burden to demonstrate on executed data, not a refuted claim but an unvalidated aggregate.", "evidence": [{"source": "Boettke & Candela, 'Retrospectives: Friedrich Hayek and the Market Algorithm', Journal of Economic Perspectives 31(3):215-30 (knowledge problem / pretense of knowledge)", "doi_or_url": "https://doi.org/10.1257/jep.31.3.215", "grade": "A"}], "facet": "measurement", "chapter": "ch7_discussion", "subclaim": "residual_risk"}
{"claim": "The published channel-selection literature establishes that the optimal channel subset is scenario-, regime-, and instrument-dependent, not regime-invariant: the standard information-entropy iterative ranking is not optimal for a geostationary sounder observing fast-changing weather, and adding a Jacobian-variance-over-time index changes which channels are selected. This is direct evidence that per-channel marginal value shifts with the intended retrieval and temporal regime, so the dispersion Hayek demands be measured is real and material and cuts against a stable pooled edge; the cross-mission dispersion statistic itself remains the candidate's to report.", "evidence": [{"source": "X. Wang et al., 'Geostationary Hyperspectral Infrared Sounder Channel Selection for Capturing Fast-Changing Atmospheric Information', IEEE Trans. Geoscience and Remote Sensing (2021)", "doi_or_url": "https://doi.org/10.1109/tgrs.2021.3078829", "grade": "A"}, {"source": "Maddy & Barnet (review), 'Optimal Estimation Retrievals and Their Uncertainties', BAMS (2020)", "doi_or_url": "https://doi.org/10.1175/bams-d-19-0027.1", "grade": "A"}], "facet": "identification", "chapter": "ch7_discussion", "subclaim": "alternatives"}
{"claim": "The discovery-suppression objection is supported in its premise: the sounder record documents validated geophysical products and applications developed after launch that were not the design driver (three-dimensional wind fields, atmospheric-instability trending for high-impact-weather nowcasting), so radiance channels function as a seedbed for later unanticipated retrievals and a cap fitted on the flown population would prune exactly that substrate, meaning the welfare claim must net foregone-discovery cost; the specific over-specified-vs-later-exploited channel overlap statistic remains to be produced.", "evidence": [{"source": "J. Li et al., 'Applications of Geostationary Hyperspectral Infrared Sounder Observations', BAMS (2022)", "doi_or_url": "https://doi.org/10.1175/bams-d-21-0328.1", "grade": "A"}, {"source": "Goldberg / Menzel et al., 'Satellite-Based Atmospheric Infrared Sounder Development and Applications', BAMS (2017)", "doi_or_url": "https://doi.org/10.1175/bams-d-16-0293.1", "grade": "A"}], "facet": "rival", "chapter": "ch7_discussion", "subclaim": "alternatives"}
{"claim": "Hayek's signal-not-command alternative is well-founded in existing retrieval theory: optimal-estimation / information-content theory already produces a local, per-state marginal-information quantity (degrees of freedom for signal, averaging kernels, marginal reduction in posterior uncertainty per measurement), and operational channel selection is an iterative marginal procedure ranking each channel by incremental information gain conditional on those already chosen, which is structurally the per-channel shadow-price object Hayek asks for; recasting the estimator's output as a decentralized per-channel marginal-accuracy-per-dollar curve is therefore consistent with the field's own tooling, though the comparative out-of-sample-robustness claim remains to be demonstrated.", "evidence": [{"source": "Maddy & Barnet (review), 'Optimal Estimation Retrievals and Their Uncertainties', BAMS (2020)", "doi_or_url": "https://doi.org/10.1175/bams-d-19-0027.1", "grade": "A"}, {"source": "X. Wang et al., 'Geostationary Hyperspectral Infrared Sounder Channel Selection', IEEE TGRS (2021)", "doi_or_url": "https://doi.org/10.1109/tgrs.2021.3078829", "grade": "A"}], "facet": "mechanism", "chapter": "ch7_discussion", "subclaim": "alternatives"}
{"claim": "The procurement-regime confound is real and grounded in the source cost-model literature: NASA maintains a SEPARATE cost-estimating relationship, NICM-E, for Explorer-like instruments defined by mission class and institutional regime (Class C, university/research-foundation-led, significant inheritance), and the existence of a distinct CER for that population confirms the regime moves recorded instrument cost orthogonally to the standard NICM CERs WITHIN the modern epoch, so a single epoch dummy cannot absorb it and a contracting-regime plus building-center fixed effect is the owed test.", "evidence": [{"source": "Mrozinski et al., 'NASA Instrument Cost Model for Explorer-Like Mission Instruments (NICM-E)', NASA NTRS 20150007881", "doi_or_url": "https://ntrs.nasa.gov/citations/20150007881", "grade": "C"}, {"source": "'Latest NASA Instrument Cost Model (NICM): Version VI', NASA NTRS 20160008251 (introduces NICM-E CER)", "doi_or_url": "https://ntrs.nasa.gov/citations/20160008251", "grade": "C"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "Heritage reuse is a recognized, named, separately-modeled cost driver, not an assumable-away nuisance: NICM-E explicitly conditions on 'significant level of inheritance' as one of three defining characteristics of its instrument population, and the NTRS convergence study documents that recent instrument-electronics costs diverged from their heritage cost-model predictions, corroborating that heritage decouples recorded cost from embodied effort and that cheap heritage instruments and expensive clean-sheet instruments occupy different cost regions; whether the candidate's concavity is a heritage-clustering artifact is a falsifiable test the candidate must run.", "evidence": [{"source": "Mrozinski et al., 'NASA Instrument Cost Model for Explorer-Like Mission Instruments (NICM-E)', NASA NTRS 20150007881 (inheritance as defining characteristic)", "doi_or_url": "https://ntrs.nasa.gov/citations/20150007881", "grade": "C"}, {"source": "'On Convergence of Development Costs and Cost Models for Complex Spaceflight Instrument Electronics', NASA NTRS 20080030195", "doi_or_url": "https://ntrs.nasa.gov/citations/20080030195", "grade": "C"}], "facet": "measurement", "chapter": "ch4_data_and_measurement", "subclaim": "alternatives"}
{"claim": "The 'cost as near-sufficient hedonic index of embodied quality' premise is empirically vulnerable: NICM is explicitly a probabilistic estimator (system/subsystem CERs with joint-confidence-level analysis), so design attributes leave residual cost dispersion by construction, and the convergence study shows instruments with otherwise-modeled attributes carrying costs that diverged from prediction, which makes the worry that residual cost indexes institutions and budget environment rather than quality a live concern; whether the orthogonalized cost residual carries the accuracy signal is the direct validating regression the candidate must run.", "evidence": [{"source": "'Latest NASA Instrument Cost Model (NICM): Version VI', NASA NTRS 20160008251 (probabilistic system/subsystem CERs with JCL)", "doi_or_url": "https://ntrs.nasa.gov/citations/20160008251", "grade": "C"}, {"source": "'On Convergence of Development Costs and Cost Models for Complex Spaceflight Instrument Electronics', NASA NTRS 20080030195", "doi_or_url": "https://ntrs.nasa.gov/citations/20080030195", "grade": "C"}], "facet": "mechanism", "chapter": "ch4_data_and_measurement", "subclaim": "residual_risk"}
{"claim": "Separating a secular trend from a cost effect is the embodied-vs-disembodied (investment-specific technical change) identification problem: a tercile epoch dummy cannot absorb a continuous cost-down/accuracy-up trend collinear with cost, so re-entering launch-year as a flexible continuous trend is the correct test and the deflated-cost-vs-launch-year rank correlation within common support is the diagnostic that reveals whether cheap-accurate rows are merely late rows.", "evidence": [{"source": "Greenwood, Hercowitz & Krusell, 'The role of investment-specific technological change in the business cycle', European Economic Review (2000)", "doi_or_url": "https://doi.org/10.1016/s0014-2921(98)00058-0", "grade": "A"}, {"source": "'New Estimates of Embodied and Disembodied Technical Progress' (1978)", "doi_or_url": "https://doi.org/10.2307/20075308", "grade": "A"}], "facet": "rival", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "The accuracy dependent variable carries disembodied (reprocessing-era) vintage that NICM does not date: algorithm reprocessing campaigns demonstrably raise the validated accuracy of an already-built instrument without changing a build dollar (MODIS Collection 6 MAIAC, second-generation Deep Blue, AERONET Version 3 are documented cases), and the record that dates accuracy independently of the build is the DAAC collection/version history plus the ATBD algorithm-baseline date, so the flat high-cost region may be algorithm-era lift mislabeled as instrument economics.", "evidence": [{"source": "Lyapustin et al., 'MODIS Collection 6 MAIAC algorithm', Atmospheric Measurement Techniques (2018)", "doi_or_url": "https://doi.org/10.5194/amt-11-5741-2018", "grade": "A"}, {"source": "Hsu et al., 'Enhanced Deep Blue aerosol retrieval algorithm: the second generation', JGR Atmospheres (2013)", "doi_or_url": "https://doi.org/10.1002/jgrd.50712", "grade": "A"}, {"source": "Giles et al., 'Advancements in the AERONET Version 3 database', Atmospheric Measurement Techniques (2019)", "doi_or_url": "https://doi.org/10.5194/amt-12-169-2019", "grade": "A"}], "facet": "measurement", "chapter": "ch4_data_and_measurement", "subclaim": "alternatives"}
{"claim": "If channel count and calendar vintage co-move, an over-specification edge is observationally indistinguishable from obsolescence (a late cheaper instrument beating an early elaborate one for vintage, not redundancy, reasons); this is the same embodied-vs-disembodied separation problem, VIF / partial-R-squared between channel count and continuous launch year inside common support is the correct collinearity diagnostic, and the NTRS instrument-lineage and DAAC reprocessing-version records are what would have to break the collinearity, so the demand to measure the VIF or concede non-identification is legitimate.", "evidence": [{"source": "Greenwood, Hercowitz & Krusell, 'The role of investment-specific technological change in the business cycle', European Economic Review (2000)", "doi_or_url": "https://doi.org/10.1016/s0014-2921(98)00058-0", "grade": "A"}, {"source": "'New Estimates of Embodied and Disembodied Technical Progress' (1978)", "doi_or_url": "https://doi.org/10.2307/20075308", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "alternatives"}
{"claim": "Channel-count cannot do double duty as both a deconfounding control in X and the do(channel-count) target: on the candidate's own DAG channels is a common driver of cost and accuracy (cost<-channels->accuracy), so it is simultaneously a confounder of the cost->accuracy association and a mediator of the channels->cost->accuracy path; by Pearl's d-separation, holding channels fixed in X blocks every channels-mediated path, so g(cost) identifies at most the residual within-channel cost effect, the two collinear regressions estimate different partial coefficients of one design decision, and the over-specification edge is partition-dependent unless the candidate draws the DAG and demonstrates empirical invariance of the marginal-channel function (the words mediator and collider appear zero times in the text).", "evidence": [{"source": "Pearl, Causality: Models, Reasoning and Inference, 2nd ed. (d-separation, mediator over-control)", "doi_or_url": "https://doi.org/10.1017/cbo9780511803161", "grade": "A"}, {"source": "Pearl, 'The Mediation Formula: A Guide to the Assessment of Causal Pathways in Nonlinear Models'", "doi_or_url": "https://doi.org/10.1002/9781119945710.ch12", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "mechanism"}
{"claim": "The survivorship restriction is collider conditioning that the dissertation mislabels as a bounded residual risk: endogenous selection bias is precisely conditioning on a collider (a common effect of treatment and outcome), and because build-quality and mission-stakes plausibly raise both cost and the probability of reaching a published validation record, selecting on 'has a validation record' can induce a spurious concave association inside the survivor stratum even when none exists in the population; the DAG's testable conditional-independence implications must be stated and checked, and the recovery-and-reweight falsification on flown-but-unvalidated radiometers run, before the headline concavity is distinguishable from a collider artifact.", "evidence": [{"source": "Elwert & Winship, 'Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable', Annual Review of Sociology", "doi_or_url": "https://doi.org/10.1146/annurev-soc-071913-043455", "grade": "A"}, {"source": "Cole et al., 'Illustrating bias due to conditioning on a collider', International Journal of Epidemiology", "doi_or_url": "https://doi.org/10.1093/ije/dyp334", "grade": "A"}, {"source": "Pearl, Causality, 2nd ed.", "doi_or_url": "https://doi.org/10.1017/cbo9780511803161", "grade": "A"}], "facet": "identification", "chapter": "ch5_research_design", "subclaim": "residual_risk"}
{"claim": "There is a rung mismatch the negativity of g'' cannot repair: the stopping rule asserts a rung-2 do(channel-count=k) query while the estimand is the rung-1 quantity E[accuracy|cost,X] the candidate itself labels reduced-form and non-mechanistic; by the back-door criterion these coincide only if X blocks all back-door paths from channels to accuracy AND contains no descendants of channels, but realized cost is a child of channel-count and calibration approach co-varies with the channel-count decision, so conditioning on them is over-control that blocks the very channels->cost->accuracy path the cap is meant to move, leaving the curvature without interventional license unless X is reduced to a valid back-door set, the front-door criterion is invoked through cost, or the deliverable is restated as a non-actionable association.", "evidence": [{"source": "Pearl, Causality: Models, Reasoning and Inference, 2nd ed. (do-operator, back-door / front-door criteria, seeing vs doing)", "doi_or_url": "https://doi.org/10.1017/cbo9780511803161", "grade": "A"}, {"source": "'Drawing Credible Directed Acyclic Graphs for Causal Inference' (2025)", "doi_or_url": "https://doi.org/10.31234/osf.io/u4yta_v4", "grade": "B"}], "facet": "mechanism", "chapter": "ch5_research_design", "subclaim": "mechanism"}
{"claim": "Simon's behavioral mechanism is correctly distinguished from a generic concave technology but is asserted as a concavity prior rather than instrumented: satisficing means search halts once an option meets an aspiration level, and near-decomposability locates collapsing returns in cross-module coupling rather than within-module elaboration, both falsifiable predictions, yet the candidate's partially-linear semiparametric g(cost) does not estimate a requirement-normalized accuracy density (the bunching signature), does not build a coupling/integration-burden proxy, and does not interact the marginal channel contribution with distance-to-requirement, so the Simon mechanism is named but not measured.", "evidence": [{"source": "Barros, 'Herbert A. Simon and the concept of rationality: boundaries and procedures', Brazilian J. Political Economy (2010)", "doi_or_url": "https://doi.org/10.1590/s0101-31572010000300006", "grade": "A"}, {"source": "Collopy & Hollingsworth, 'Value-Driven Design', J. Aircraft / AIAA", "doi_or_url": "https://doi.org/10.2514/1.54033", "grade": "A"}], "facet": "mechanism", "chapter": "ch2_theoretical_framework", "subclaim": "mechanism"}