Dynamic Heresthetics: Agenda Control, Feasible-Set Erosion, and Counterfactual Democratic Auditing

This paper develops a dynamic extension of Riker-style heresthetics, the art of winning by restructuring the choice situation rather than by changing preferences. The new channel is that agenda omission changes future availability: voters have fixed preferences and a social choice rule selects from whatever agenda is offered at each date, but every alternative is supported by a feasibility substrate, meaning the coalition organization, candidate pipeline, legal standing, institutional memory, public salience, professional competence, or interoperability that keeps the alternative practically available. When an omitted alternative's substrate decays below a threshold, the alternative leaves the feasible set. The static heresthetician controls the map from current agenda to current outcome; the dynamic heresthetician controls the composed map from current agenda to current outcome to future feasibility state to future agenda set, converting agenda power from a selection technology into a feasibility technology. The paper positions itself as a framework-plus-identification contribution rather than a full equilibrium theory of lobbying or agenda bargaining. The implementation results are transparent threshold-crossing arguments whose purpose is to isolate the mechanism and mark its exact boundary. The baseline theorem shows that a setter can install a target that would lose under the initial full agenda by repeatedly offering it against a weaker protected default while withholding stronger rivals until their substrates decay, at which point the target becomes the defended winner of the entire surviving feasible set. For Condorcet-consistent rules under linear omission decay the reachability boundary is exact: an initially feasible target is installable if and only if no protected alternative defeats it. A universal-implementation corollary is careful that the dramatic reading, that a dynamic heresthetician can install anything, holds only over the initially feasible domain and only when the protected set imposes no dominance, framed as the feasibility-side counterpart of McKelvey-Schofield richness. Companion results show that fixed feasibility would instead force the Condorcet winner, that omission is robust to sophisticated voting, and that a positive exclusion time is necessary under the linear law. Because the unopposed baseline assumes omitted alternatives have no champion, an opponent-maintenance section adds a maintenance threshold, a one-sided stopping equilibrium under a geometric audit, and a two-sided war-of-attrition benchmark, culminating in an opposed characterization that replaces the protected set with a protected-plus-maintainable set. This version adds three formal contributions beyond the reachability and attrition results. First, it microfounds the linear substrate law as cohort replacement: omission interrupts the routine replenishment of retiring organizational, legal, technical, or attention carriers, inclusion preserves the carrier mass, and selection recruits additional carriers, so the decay, recruitment, and threshold parameters acquire concrete meaning as unreplaced attrition, net recruitment, and minimum viable carrier mass rather than remaining arbitrary reduced-form constants. Second, it embeds the setter and multiple budget-constrained maintainers in a finite-state dynamic game with agenda costs, per-blocker maintenance technologies, and binding budgets, and proves existence of a Markov-perfect equilibrium together with a complete backward-induction characterization: equilibrium play at every date-state pair is a Nash equilibrium of the induced one-period continuation game. This integrated object endogenizes agenda choice and maintenance timing and clarifies the status of the earlier reachability sets, since the protected and protected-plus-maintainable sets are reachability obstructions while the finite game asks the further equilibrium question of which actors actually choose omission, inclusion, maintenance, or abandonment given costs and continuation values; the paper notes explicitly that equilibrium need not be unique and that a technically maintainable blocker can still disappear when its maintainer's continuation value falls below cost. Third, it gives a minimal-cost preservation benchmark for viability rights: in the independent-blocker case the least stock-unit maintenance needed to preserve an audit at a given date equals exactly the threshold shortfall created by omission, subject to prefix capacity constraints. The oversight results remain the paper's main identification contribution. A passive non-identification theorem proves that a finite passive record generated by heresthetic foreclosure, in which the foreclosed blockers never appear, is reproduced by a benign benchmark in which those blockers were simply never available, so no passive audit can distinguish foreclosure from benign non-availability and a data set can pass every revealed-preference test over observed agendas while remaining consistent with the foreclosure of initially feasible alternatives. The paper is careful that this is not an anti-rational-choice claim: revealed preference over realized menus identifies behavior over realized menus but not the history by which those menus became feasible, and the result is positioned as an Austen-Smith-style observational equivalence along a different dimension, the provenance of feasibility rather than the sincere-versus-sophisticated voting strategy. Detection therefore requires stepping outside the realized agenda through counterfactual agenda probes, for which the paper gives a hitting condition, a pairwise-majority special case, and an audit-rank formulation, and it proves an audit-deadline theorem establishing a last feasible date beyond which the decisive counterfactual agenda can no longer be run or credibly reconstructed without restoring the substrate. A viability-right impossibility theorem shows that an institution conditioning only on choices over current menus cannot be simultaneously permissive and foreclosure-complete, so anti-foreclosure protection provably requires non-passive information. Extensions cover stochastic substrate dynamics, endogenous decay rates, multiple setters, and agenda costs, and the paper closes with applications, empirical diagnostics including a worked hearing-denial and bill-reintroduction design and a crossed demand-and-availability shock design, and a checklist for empirical use.