Nothing extra is being measured here. The docs describe confidence as “a statistic computed from the probability distribution the answer already gives you”, collapsing the shape of that distribution into one number. A Choice that puts 1.0 on returns and 0.0 everywhere else comes back with confidence 1.0. A Score split 0.7 and 0.3 across two adjacent levels comes back with confidence 0.54.
TypeSafe’s guidance runs in three bands. High confidence means act automatically. Medium means proceed with caution, asking the user to confirm or flagging the case for review. Low means do not act: route it to a person or fall back to a different system.
The worked example in the docs puts real numbers on those bands. Below 0.5 it treats the answer as uncertain and routes to a human. Above 0.9 is the bar it sets before a destructive operation like a funds transfer runs without confirmation, while a read-only action like a balance check can go ahead lower down.
Where the lines land is your decision, in the docs’ words: “Your code encodes the risk tolerance.” How to build with System One models works through the pattern.