6  Discussion

The central finding — that the binding axis of governance is the trigger signal, not the instrument — repositions a question that the polycentric governance literature has mainly framed in terms of institutional design. Carlisle and Gruby (2019) and Heikkila et al. (2018) characterise polycentric systems by the rules, instruments and coordination mechanisms they deploy and the institutional conditions under which those instruments succeed. Morrison et al. (2019) add that power asymmetries between decision centres can undermine even well-designed instruments. Our results suggest that a prior question must be answered first: what is the regulator looking at? A perfectly designed instrument armed on the wrong signal — the lagging symptom rather than the leading cause — is structurally too late, regardless of institutional arrangements.

The failure mode at the dt boundary reinforces this asymmetry. When the governance cadence falls below the emitter’s collapse rate, the emitter is sacrificed even as the downwind arenas stay protected: the timescale separation between fast capital collapse and slow pollution import is the leading-on-capital regulator’s critical resource, and the first thing it loses (mechanism derived in Section 4.1). Read in the vocabulary of rate-dependent tipping, this makes the governance sampling rate a design dimension in its own right — to be set alongside the forcing rate that the cascade-tipping literature already tracks. The present model’s simplicity (a deterministic ODE substrate with a single emitter geometry) limits how far this can be pressed; the analogue mechanism in stochastic, heterogeneous, or higher-dimensional polycentric settings remains an open question.

This complements, rather than contradicts, the instrument-centred view. Building on Dorsch and Flachsland (2017)’s argument for site-specific experimentation in polycentric climate governance, we add a temporal axis to the assessment: site-specific monitoring must be evaluated not only for coverage (are the right variables observed?) but for temporal position (is the observed variable leading or lagging the process that must be governed?). The leading/lagging distinction is our contribution, not theirs; we read their site-specificity as the spatial complement to it. The Bardi/Seneca substrate makes this distinction unusually sharp because pollution P is multiplicatively downstream of capital C in the ODE, but the same structural question arises whenever a regulator must choose between a direct and a derived signal. In Ostrom’s design-principle vocabulary (Ostrom 1990), the finding maps onto Principle 4 (monitoring): effective governance requires not just that someone watches, but that they watch the right variable at the right point in the causal chain.

A real-world analogue, beyond the present model’s scope, is the cross-boundary cascading of nutrient pollution discussed by Ahlström and Cornell (2017) on the global nitrogen and phosphorus cycles. Their empirical mapping of how fragmented governance leaves room for N/P cascades to propagate across jurisdictions echoes our A\(\to\)B/C topology: the emitter’s production (analogous to our C) generates the nutrient load (analogous to P) that propagates downstream. Our result predicts that monitoring production intensity rather than ambient nutrient concentration would be the structurally earlier trigger — a hypothesis falsifiable with field data.

The limitations discussed in Chapter 5 constrain how far these conclusions can be pressed; two directions for future work follow directly. First, testing whether a tighter clamp (a hypothetical throttleK1 = 0.005 ceiling, roughly a quarter of the fixed 0.02 cap applied in every leading cell — the intervention setting, not an OFAT-swept factor) or sub-turn governance ticks restore the full win-win at dt > 3 would map the boundary between rate-induced failure and instrument-intensity failure — in Lohmann et al.’s vocabulary, bringing the governance sampling rate back inside the critical timescale. Neither the tighter clamp nor the sub-turn cadence is in the OFAT sweep; both are future-work hypotheses. Second, coupling the Seneca substrate to a stochastic or LLM-driven agent layer — as in the companion papers on single-arena governance (Brunelli 2026c) and cheap-talk dynamics (Brunelli 2026b) — would test whether the leading-signal advantage survives when the regulator itself is a boundedly rational agent rather than an omniscient meta-agent. A deterministic regulator armed on the right signal already outperforms any reactive baseline; the open question for the series is whether an LLM regulator would combine the adaptivity documented in BDPD\(^2\) and BDPD\(^1\) with the temporal discipline of leading-indicator governance — or whether the noise and hallucination tendencies of LLM agents would erode the timing advantage that makes cap-leading effective.

Methodological Coda: LLMs as Synthetic Collaborators

Unlike the companion papers BDPD\(^2\) and BDPD\(^1\), this paper uses no LLM agents experimentally — all players are deterministic RCPAgent instances and the governance meta-agent is algorithmic. LLMs contributed exclusively in Role A (epistemic/generative): conceptual development, implementation, formal analysis, and manuscript preparation, following the human-AI collaborative methodology described in BDPD\(^0\) (Brunelli 2026a, §Methodological Coda). Multiple frontier LLMs with different training distributions (Anthropic’s Claude, DeepSeek, Alibaba’s Qwen, Z.ai’s GLM-5.1) were employed for cross-validation of analysis and framing, under the author’s direction.

Data and Code Availability

The BDPD simulation platform, the Seneca engine, experiment definitions, and analysis scripts are available at gitlab.com/bdpd/bdpd. The canonical OFAT sweep artefacts are in data/pilot/ of the repository.

Ahlström, Hanna, and Sarah E. Cornell. 2017. “Governance, polycentricity and the global nitrogen and phosphorus cycles.” Environmental Science & Policy 79: 54–65. https://doi.org/10.1016/j.envsci.2017.10.005.

Brunelli, Roberto. 2026a. BDPD⁰ — a Computational Laboratory for Generative Agents, Commons Dilemmas and More. Zenodo. https://doi.org/10.5281/zenodo.20678100.

Brunelli, Roberto. 2026b. BDPD¹ — Governance: Cheap Talk and Sanctioning on a Fragile Commons. Zenodo. https://doi.org/10.5281/zenodo.20679157.

Brunelli, Roberto. 2026c. BDPD² — Governance of Nested Commons: Four Vignettes. Zenodo. https://doi.org/10.5281/zenodo.20679946.

Carlisle, Keith, and Rebecca L. Gruby. 2019. “Polycentric Systems of Governance: A Theoretical Model for the Commons.” Policy Studies Journal 47 (4): 927–52. https://doi.org/10.1111/psj.12212.

Dorsch, Marcel J., and Christian Flachsland. 2017. “A Polycentric Approach to Global Climate Governance.” Global Environmental Politics 17 (2): 45–64. https://doi.org/10.1162/glep_a_00400.

Heikkila, Tanya, Sergio Villamayor-Tomas, and Dustin E. Garrick. 2018. “Bringing polycentric systems into focus for environmental governance.” Environmental Policy and Governance 28 (4): 207–11. https://doi.org/10.1002/eet.1809.

Morrison, Tiffany H., W. Neil Adger, Katrina Brown, et al. 2019. “The black box of power in polycentric environmental governance.” Global Environmental Change 57: 101934. https://doi.org/10.1016/j.gloenvcha.2019.101934.

Ostrom, Elinor. 1990. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press. https://doi.org/10.1017/CBO9780511807763.