MicroscopeONE studies what probabilistic systems can infer from public organizational surfaces, and where that inference drifts from what organizations believe they communicate.
Inference stability depends on the structure of observed context, not on content quantity.
The laboratory asks what can actually be inferred, not how the organization wants to appear.
Human recognition does not guarantee agentic clarity.
We observe a structural shift in how probabilistic systems reconstruct organizations. A new mediating variable has become visible in the reconstruction chain — and it may have its own agenda.
A fictional company, three versions of its website, identical content. The inference quality delta: 27 to 50 points.
The calibration dataset exposed a brutal distinction: human brand recognition does not transfer to agentic clarity.
Foundational documents from Phase 1, adapted for public reading.
A controlled experiment testing whether semantic architecture changes agent inference under zero prior knowledge.
When we publish a finding, hypothesis, or experiment.