\section{Conclusion}
\label{sec:conclusion}

We set out to state a universal theory of human coordination in a form that could be tested, and to test it. The statement survived: institutions as shared codes, records and routing structures that lower the communication cost per coordinated task are a coherent object, with five provable properties and a simulable cost structure, and the simulations show both the gains and the three ways institutions can hurt---unpaid set-up costs, saturated hierarchies, and task environments with too little regularity to compress. The test did not confirm the theory's empirical corollary. In the Seshat databank and in a two-century country-year panel, the mismatch between coordination demand and coordination capacity is associated with instability in the predicted direction, robustly across regions and eras, and yet it adds nothing identifiable to what scale alone (in history) or capacity alone (in modern states) already explain; a placebo does as well, past instability loads on the index as strongly as future instability, and the coefficient does not survive a standard adjustment for unobservables. Under the rule we committed to before looking, the universal framing is not supported, and we have written the narrower paper instead.

The result should be read for what it says about measurement as much as about the hypothesis. Scale and capacity are so tightly coupled across the historical record that their difference is a residual carrying one sixth of their joint variance; a mismatch index built from presence codes cannot track utilization in the sense the theory requires, and the data that would---message volumes, administrative workloads, backlogs, the reach rather than the existence of institutions---are not in any databank we know of. The one historical case in which such quantities have been reconstructed, \citeauthor{sng2014}'s late-imperial China, found the effect of distance on fiscal reach that the framework predicts. A credible next test would need either more cases of that kind, or a source of variation in demand or capacity unrelated to local instability, or a historical databank that codes institutional reach and administrative load rather than presence. Until then, the honest summary is that capacity matters, scale matters, and that whether their mismatch matters is a question the available evidence cannot answer.

Three things are offered for reuse: a pre-registration and measurement model for Seshat that treats missing as missing and carries ranges, dates and latent-score uncertainty through to inference; a validated bibliography and novelty matrix for the strands that converge on this hypothesis; and a factorial agent-based testbed in which the mechanisms are measurable. The replication package reproduces every number in this paper from the raw downloads with one command.