The cross-reference to 5.3.1 is the useful fact. It tells the reader that reconstruction is not one generic step: it depends on the deployment case. In a two-party run the party that holds the output may be the evaluator, the garbler, or both, and each choice changes the evidence a reviewer needs.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.