The dashboard is the analogue layer: the interface through which Fellows encounter, explore, and act on the intelligence the machine has surfaced. The heuristic sliders are not decorative. They are the mechanism by which human intention shapes machine inference. A Fellow seeking complementary skills (slider high) will get a different result set from one seeking disciplinary overlap (slider high). The system does not decide what matters. The Fellow does. The system extends the Fellow's reach beyond the Dunbar limit while preserving the Fellow's evaluative authority.
Every response (accept, reject, modify) flows back as a prediction error signal. The system learns not only from what Fellows accept but from what they reject. Rejection is information. Over time, the model's understanding of the network's actual topology, the Fellows' actual interests, and the system's own biases improves. Errors are fuel, not waste.