Most peer-to-peer systems rely on external infrastructure---such as signaling servers, trackers, or distributed hash tables---to enable peers to discover and connect to one another. This infrastructure is commonly treated as fundamental. This document adopts a different view: such components primarily synchronize transient, time-sensitive state between otherwise deterministic peers. We introduce Predictive Rendezvous, a coordination architecture in which peers establish communication by executing a shared, deterministic rendezvous plan derived from prior agreement, shared time context, and cryptographic entropy. Instead of discovering each other via network lookups, peers predict each other's behavior in time, causing their independently evolving network states to converge without negotiation or intermediaries. The approach generalizes across domains, including real-time communication and distributed content swarms, and reframes peer discovery as a problem of temporal and intentional coordination rather than address resolution. This document is intended as prior art describing the conceptual model, terminology, and protocol structure of Predictive Rendezvous and related notions such as Semantic Rendezvous Tokens. The architecture is independent of transport protocol, addressing scheme, or application payload.