Neither Consent nor Property: A Policy Lab for Data Law
Pith reviewed 2026-05-18 03:28 UTC · model grok-4.3
The pith
Shifting liability for data harms to downstream buyers maximizes social welfare more than informed consent rules.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The paper establishes that property-rule mechanisms such as informed consent fail to maximize welfare in the data market. Social welfare peaks when liability for substantive harm is shifted to the downstream buyer, consistent with the least cost avoider principle because downstream users control post-acquisition safeguards and are best positioned to mitigate risk efficiently. This provides an economic justification for de-romanticizing seller-centric frameworks and supporting doctrines of downstream reachability.
What carries the argument
A spatially explicit Agent-Based Model of the data market, calibrated through a two-stage pipeline that converts multi-year fieldwork into agent constraints and employs LLMs as subjects in a Discrete Choice Experiment to recover preference primitives.
If this is right
- Regulators should consider shifting liability rules toward downstream buyers to improve overall welfare in AI data economies.
- Consent-based property rules are shown to underperform compared to liability allocations focused on efficient risk mitigation.
- Policy design should prioritize the position of the party best able to avoid harm rather than romanticizing seller consent.
- Emerging legal doctrines emphasizing downstream accountability receive support from the welfare simulations.
Where Pith is reading between the lines
- Similar modeling approaches could test liability allocations in adjacent domains such as personal health data or creative content markets.
- The findings suggest that uniform consent rules may need replacement by context-specific assignments based on control over safeguards.
- Extending the model to include dynamic learning by agents could reveal long-term effects on data market evolution.
- Real-world validation might involve comparing jurisdictions with differing liability rules for data harms.
Load-bearing premise
The two-stage methodological pipeline that translates fieldwork decision rules into agent constraints and uses LLMs in discrete choice experiments accurately recovers actual bargaining frictions and unobservable preference primitives.
What would settle it
Empirical data from actual data transactions showing that downstream buyers do not reduce harm more effectively than consent mechanisms, or welfare measurements in markets with different liability rules contradicting the simulation results.
read the original abstract
Regulators currently govern the AI data economy based on intuition rather than evidence, struggling to choose between inconsistent regimes of informed consent, immunity, and liability. To fill this policy vacuum, this paper develops a novel computational policy laboratory: a spatially explicit Agent-Based Model (ABM) of the data market. To solve the problem of missing data, we introduce a two-stage methodological pipeline. First, we translate decision rules from multi-year fieldwork (2022-2025) into agent constraints. This ensures the model reflects actual bargaining frictions rather than theoretical abstractions. Second, we deploy Large Language Models (LLMs) as "subjects" in a Discrete Choice Experiment (DCE). This novel approach recovers precise preference primitives, such as willingness-to-pay elasticities, which are empirically unobservable in the wild. Calibrated by these inputs, our model places rival legal institutions side-by-side to simulate their welfare effects. The results challenge the dominant regulatory paradigm. We find that property-rule mechanisms, such as informed consent, fail to maximize welfare. Counterintuitively, social welfare peaks when liability for substantive harm is shifted to the downstream buyer. This aligns with the "least cost avoider" principle, because downstream users control post-acquisition safeguards, they are best positioned to mitigate risk efficiently. By "de-romanticizing" seller-centric frameworks, this paper provides an economic justification for emerging doctrines of downstream reachability.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper develops a spatially explicit Agent-Based Model (ABM) as a computational policy laboratory for the AI data economy. A two-stage pipeline first translates multi-year fieldwork (2022-2025) decision rules into agent constraints and then deploys LLMs as subjects in a Discrete Choice Experiment to recover unobservable preference primitives such as willingness-to-pay elasticities. Calibrated on these inputs, the model compares welfare outcomes across informed consent, property-rule, and downstream-liability regimes, concluding that social welfare is maximized when liability for substantive harm is assigned to downstream buyers because they control post-acquisition safeguards and are the least-cost avoiders.
Significance. If the calibration pipeline is shown to be robust, the work supplies a simulation-based, quantitative comparison of rival legal institutions in data markets and supplies an economic rationale for downstream reachability doctrines. It demonstrates how ABMs can generate policy-relevant rankings when direct empirical data on bargaining frictions and risk valuations are unavailable.
major comments (3)
- [Abstract and methodological pipeline] The two-stage pipeline (Abstract) recovers willingness-to-pay elasticities and bargaining frictions from LLM subjects in the DCE; because these fitted primitives directly determine the welfare ordering across regimes, the manuscript must supply validation against human subjects, robustness checks for LLM training-corpus biases, and sensitivity analyses on the recovered elasticities.
- [Calibration and simulation results] Welfare rankings are generated by an ABM whose key behavioral parameters are recovered from the same fieldwork and LLM experiments used to calibrate it (Abstract); this creates a circularity risk in which comparative results depend on the fitted primitives rather than independent variation, requiring explicit out-of-sample tests or alternative calibration scenarios.
- [Welfare comparison and least-cost-avoider discussion] The claim that downstream buyers are the least-cost avoiders (Abstract) is load-bearing for the policy conclusion, yet the model description does not report explicit simulation checks that downstream agents indeed face lower mitigation costs or exercise more effective post-acquisition safeguards under the modeled constraints.
minor comments (2)
- [Abstract] The abstract would be strengthened by reporting at least one key quantitative welfare difference or ranking statistic from the simulations rather than describing the approach only.
- [Model description] Notation for agent constraints and preference parameters should be defined consistently when first introduced to aid readers unfamiliar with the ABM implementation.
Simulated Author's Rebuttal
We thank the referee for these constructive comments, which highlight important aspects of our methodological pipeline and results. We address each major comment below, indicating revisions where we can strengthen the manuscript while being transparent about limitations.
read point-by-point responses
-
Referee: [Abstract and methodological pipeline] The two-stage pipeline (Abstract) recovers willingness-to-pay elasticities and bargaining frictions from LLM subjects in the DCE; because these fitted primitives directly determine the welfare ordering across regimes, the manuscript must supply validation against human subjects, robustness checks for LLM training-corpus biases, and sensitivity analyses on the recovered elasticities.
Authors: We agree that robustness checks and sensitivity analyses are necessary given the central role of the LLM-derived primitives. In the revised manuscript we will add an appendix containing sensitivity analyses that vary the recovered elasticities by ±25% and ±50% around the point estimates from the DCE; these checks confirm that the welfare ranking across regimes is preserved. We will also include comparisons of results obtained from two distinct LLM families to probe training-corpus effects. Direct validation against human subjects, however, lies outside the scope of the present study: the multi-year fieldwork (2022–2025) was designed to elicit decision rules rather than to run a parallel human DCE, and mounting such an experiment would require resources and ethical approvals not available within the current project. We will explicitly note this as a limitation and flag it for future work. revision: partial
-
Referee: [Calibration and simulation results] Welfare rankings are generated by an ABM whose key behavioral parameters are recovered from the same fieldwork and LLM experiments used to calibrate it (Abstract); this creates a circularity risk in which comparative results depend on the fitted primitives rather than independent variation, requiring explicit out-of-sample tests or alternative calibration scenarios.
Authors: We accept the concern about potential circularity. To address it we will add two explicit safeguards in the revised version. First, we will report an out-of-sample exercise in which 30% of the fieldwork observations are withheld from calibration; the held-out data are then used to assess how well the calibrated agents reproduce observed bargaining patterns. Second, we will introduce an alternative calibration scenario that draws behavioral parameters from theoretical distributions drawn from the general literature on information goods and bargaining frictions, independent of our fieldwork and DCE estimates. These additions will demonstrate that the welfare ordering is not an artifact of the specific fitted primitives. revision: yes
-
Referee: [Welfare comparison and least-cost-avoider discussion] The claim that downstream buyers are the least-cost avoiders (Abstract) is load-bearing for the policy conclusion, yet the model description does not report explicit simulation checks that downstream agents indeed face lower mitigation costs or exercise more effective post-acquisition safeguards under the modeled constraints.
Authors: We agree that the least-cost-avoider claim requires direct simulation support. In the revised model-description section we will add new results from 1,000 Monte Carlo runs that tabulate average mitigation costs and safeguard effectiveness for downstream versus upstream agents under each legal regime. These checks will show that downstream agents incur lower expected mitigation costs because they control post-acquisition uses, consistent with the modeled constraints. The new tables and figures will be referenced in the welfare-comparison discussion. revision: yes
Circularity Check
No circularity: welfare simulation applies calibrated primitives to counterfactual regimes
full rationale
The paper calibrates an ABM using a two-stage pipeline (fieldwork decision rules translated to agent constraints, plus LLM DCE to recover WTP elasticities and bargaining frictions) and then runs simulations comparing consent, property-rule, and downstream-liability regimes. The reported welfare peak under downstream liability is an output of applying those independently recovered parameters to policy counterfactuals, not a quantity that equals the fitted inputs by construction. No equations, self-citations, or uniqueness claims are shown that would collapse the result into the calibration data. The derivation is therefore self-contained as standard empirical calibration followed by simulation.
Axiom & Free-Parameter Ledger
free parameters (1)
- willingness-to-pay elasticities
axioms (1)
- domain assumption Decision rules from 2022-2025 fieldwork can be translated into agent constraints that reflect actual bargaining frictions.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.