“…This is because there is no guarantee on what the set of IDs in the current graph (or distributed system) is, and in a directed graph, we cannot even scan through the graph to find out. On the other hand, the logic in [3] can express, e.g., that the current node has the largest ID, which we cannot do. Of course, with a non-uniform formula allocation function, the circuit-style paradigm can even specify non-computable properties.…”