Should Trust Be a Score?

Should trust be a score?

It is tempting. A single number is quick to read, easy to sort by and simple to set a threshold on. The sources behind Wellzai consistently decline it. The RSM federation architecture says trust emerges from identity, authority, provenance, evidence, relationship history, assertions and observed behaviour, and asks the runtime to preserve that evidence rather than assign one universal trust score. Wellzai may derive confidence from those signals; the signals themselves stay visible.

The same refusal recurs wherever a score would be convenient. Wellzai's formation design treats the viability of a configuration as multidimensional, covering capability and capacity coverage, constraints, timing, evidence confidence, commitment readiness and more, with each dimension meant to stay individually inspectable even when something is aggregated internally. Wellzai's Intent Path model rejects "score = 82.7" with unclear semantics and prefers honest labels instead: evidence strong, one open blocker, coordination medium, timing confidence unknown.

The objection is not to numbers. It is that a score hides which signal is carrying the weight. Two buyers might trust the same supplier, one because of years of deliveries and the other because of a single attestation. They are relying on different things, and a number would make them look identical.

The implication, provisionally, is that "trusted" may not be a property of a participant at all. It may be a relationship between a particular decision and the evidence behind it.

The open question is speed. Labels and dimensions are honest, but they ask more of the person reading them. Someone deciding quickly, in the middle of a season, may not read seven dimensions. What replaces the score for a person who has to decide fast, without bringing the single number back under another name?