Gorde:
| Egile nagusia: | |
|---|---|
| Formatua: | Recurso digital |
| Hizkuntza: | ingelesa |
| Argitaratua: |
Zenodo
2026
|
| Gaiak: | |
| Sarrera elektronikoa: | https://doi.org/10.5281/zenodo.19871131 |
| Etiketak: |
Etiketa erantsi
Etiketarik gabe, Izan zaitez lehena erregistro honi etiketa jartzen!
|
Aurkibidea:
- <p>Heterogeneous autonomous agents from different manufacturers, organizations, or jurisdictions must coordinate in shared physical space without subordinating to a central authority and without revealing proprietary internal state.</p> <p>Mechanism design has historically proceeded under two assumptions that this setting violates: (i) agents report their types, and (ii) the mechanism evaluates outcomes through declared types alone. The Myerson-Satterthwaite (1983) impossibility result follows directly from those assumptions; under bilateral trade with quasi-linear utilities, no mechanism is simultaneously efficient, individually rational, budget-balanced, and strategy-proof.</p> <p>We study a structurally distinct setting in which (a) each agent emits only a lossy, bounded projection; a Minimum Viable Reality (MVR) into a shared state, (b) physical trajectories are observable ex-post and comparable against the MVR fields the agent declared, and (c) trust is a degradable scalar in [0,1] that weights an agent’s contribution to the convergence operator and is decayed by detected inconsistency.</p> <p>We make four formal contributions.</p> <ol> <li>First (Theorem 1), we define a Kinetically Blind Referee (KBR); an arbiter whose actions depend only on observable physical derivatives, never on declared intent and prove it detects spatial-envelope fraud in O(1) cycle.</li> <li>Second (Theorem 2), we prove that under MVR projection, physical verification, and degradable trust, truthful envelope projection is the unique strict best response against any one-shot deviation under any positive discount rate; the Myerson-Satterthwaite impossibility does not apply to this setting.</li> <li>Third (Theorem 3), we establish antifragility: the attacker’s cumulative payoff under iterated adversarial play decreases monotonically; we report empirical evidence under both a discrete-event cost function and a continuous-cost formulation, and discuss honestly why only the latter recovers the temporal signature the theorem predicts.</li> <li>Fourth (Theorem 4), we recast what prior work called “emergent axiom convergence” as architectural convergence: under the protocol architecture ( +KBR+ extended Integrity), Syncference agents are Axiom-compliant by construction, and deviations from compliance correlate strictly with deviations from protocol-respecting behavior.</li> </ol> <p>Empirically, 7,500 runs in MAZ3 v1.1.0 (commit fef3fc6) confirm the predicted signatures; raw data and plots are public. We position the work in the Hurwicz (1972) tradition of mechanism design, applied to a domain (real-time physical coordination with adversarial agents) that was outside the scope of his original framework.</p> <p><strong> </strong></p> <p>This work contributes foundational results to <em><strong>Perspectival Theory</strong></em> (alternatively, <strong><em>Sovereign Coordination Theory</em></strong>): the formal study of how sovereign agents coordinate through minimal revelation without surrendering internal state.</p> <p> </p>