Correctness asks whether the functions produce the right result with the precision required. A measure compares correct results with the results the product produced. This is the family where the expected result is the evidence. Without a recorded expectation for each case, a pass is an opinion. The draft puts precision into the definition of the subcharacteristic, so a result that is right to two decimal places when the task needs four is not fully correct. The measure should carry the case population and the definition of correct that was used, and it should distinguish a wrong result from a result that could not be produced at all.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.