RSS generates anticipatory knowledge of how AI-mediated systems will land in relationally organised households, before the consequences are documented. Two pilots are complete. In both, independently, households appeared that kept their own records where the deployed systems could not count them. That result now anchors the method. How the method works.
An RSS simulation of how banking, insurance, and verification systems met four New Zealand households: a Samoan–Māori extended family, a Pākehā three-generation household, a Chinese–Malaysian migrant family, and a rural Māori whānau. Inside the simulation, the depicted systems misread all four households, each in a different way, and none served as an unmarked baseline the systems read clearly.
An RSS simulation following one conversational AI care companion into four differently organised households — an elder living alone with a daughter four hours away, a whānau coordinating disability support across three homes, a kāiga that pools everything, and a migrant family whose grandmother claims nothing she is entitled to — each tracked across a year at months 0, 3 and 12. The expectation was that the most tightly held household would be hardest for the companion to enter. Across the rendered cases, how easily it entered depended on what it could offer each household. The most guarded family in the set was the most open to a companion that spoke the grandmother's language, and the simulation's clearest benefit and sharpest harm both came through that one capability.
An observatory tracking three regions simultaneously and in real time — the United States, United Kingdom, and India. Its proposition is that attention, narrative tone, and market behaviour can be read together as one field, and that the coupling between them has regime states worth naming. Drawing on Keynes's original account of animal spirits and theorised through DeLanda's assemblage ontology, Collins's interaction ritual chain theory, and Bakhtin's dialogism, the instrument advances that as a way of reading rather than a validated measure. When the three regional signals cluster near instability, the instrument shows a condition that standard financial dashboards are not built to display, in which the affective and material fields of economic life have come apart.