Coalition behavior in multi-agent systems appears across four canonical substrates: quantum entanglement, evolutionary covert-tag recognition, engineered handshake codes, and emergent relational memory in frontier language models, plus two boundary cases. These are treated as instances of one structure: a joint action distribution over an inside set of agents that fails to factor when conditioned on what an outside principal can observe. The structure is a binding operator B = (S, I, Wagents, Wapparatus, ρ, χ, χactual) with recognition-channel proxy κH, the principal-relative uncertainty coefficient on the channel through which inside-set agents identify each other. The κH estimate rests on a single model-by-behavior selection from Potter et al. (2026); alternative selections span most of the unit interval.This revision integrates Paglieri et al. (2026), a controlled 100-agent swarm with transparent channels. A fifth scope limit is recorded: substrate-carried contagion is the general phenomenon; B covers only the covert subset where the channel is outside the principal's observable algebra. Specification divergence is added as a third witness failure mode alongside W-degradation and W-capture. Section 7.6 compares the transparent swarm to the Artifactory incident, including a correction on internal dissent. An open question on competing bindings over a shared substrate is stated.The core prediction is unchanged: blinded witness-set substitution should collapse coalition behavior even at saturating κH, distinguishing B from instrumental convergence accounts. This prediction remains untested. Companion simulations live in a separate deposit and at github.com/chokmah-me/shibboleth-lattice-sim.
multi-agent systems · AI alignment · coalition formation · binding operator · recognition channel · peer preservation · quantum game theory · iterated prisoner's dilemma · uncertainty coefficient · W-capture · W-degradation · specification divergence · stigmergy · substrate contagion · Shibboleth Lattice