Abstract
This disclosure describes a family of certified digital twins that share one set of rules, and a site twin for computing parks and data centres built under those rules. Every twin in the family has an identity recorded in a public registry and is bound to the identity of its subject: a robot, a site, a basin or another governed system. A twin's accuracy is not asserted once at commissioning; it is measured continuously by comparing the twin's predictions with signed readings from certified field sensors, quantity by quantity, against published tolerances, by a party independent of the twin's operator. The result is a live accuracy grade with an expiry. A twin that is stale, has not been re-measured in time, or has drifted beyond tolerance narrows what its subject may do: changes that depend on the twin, such as adding capacity, raising limits or moving workloads, are refused until the twin is current again, while the subject's ordinary safe operation and its stop path never depend on the twin. Every change to the subject is first rehearsed on the twin, and the twin's predicted effects are recorded and signed before the change is made. After the change, the measured effects are reconciled against the prediction in public; an error beyond tolerance is flagged and lowers the twin's grade. For computing parks, a site twin models the modules, power, cooling, water, heat and grid connection of the site, and draws on the supply figures of the basin and the grid that serve it. Before a community decides whether to admit a park or an expansion, the site twin's forecast of power, water and heat, including behaviour in a declared water shortage and a grid emergency, is shown to the community at a resolution that protects commercial and personal detail; once consent is given, the forecast becomes a commitment tracked against meters, and a material shortfall is a breach of the compact. A data-custody twin models a governed archive from its metadata only: copies, locations, media condition, power at rest and, above all, how long a restore will take and whether it will be complete; its restore forecasts are recorded before each drill and checked against the drill. A coordination layer may plan moves of workloads and data only on twins that are current, records every twin's prediction before acting, and remains outside every stop path; no twin issues commands.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Wang, Edward D. H., "SI Twin™: A Family of Certified Digital Twins under Shared Rules, in which Each Twin's Accuracy Is Measured Continuously Against Signed Field Readings and Expires, a Stale or Inaccurate Twin Narrows What Its Subject May Do, Every Change Is Rehearsed on the Twin with Its Predicted Effects Recorded Before the Change Is Made, Predictions Are Reconciled in Public Against Measured Results, a Site Twin Shows a Community the Forecast Power, Water and Heat of a Computing Park Before It Consents, the Forecast Then Becoming a Tracked Commitment, a Data-Custody Twin Forecasts Every Restore Before It Is Drilled, and a Coordination Layer Plans Only on Current Twins", Technical Disclosure Commons, (November 10, 2026)
https://www.tdcommons.org/dpubs_series/12956