01
Use the initial snapshot to compute the first update
At step zero the active set is A alone. Its two outgoing connections provide input to B and C, so those two nodes are active at step one. A has no incoming connection and becomes inactive. D receives no qualifying input from the step-zero snapshot because B and C were both inactive then. The step-one active set is therefore B and C. Do not use their newly computed activity to activate D within that same update; doing so would mix the old and new snapshots and violate the stated simultaneous rule.
02
Combine incoming signals without duplicating the node
At step one, B and C are active. Each has a directed connection to D, so D receives two incoming signals for the next update. The rule only asks whether at least one exists. D is therefore one active node at step two, not two copies of a node. B and C become inactive because A was inactive in the step-one snapshot. At step three, no node has an active incoming neighbour: D has no outgoing connections, so the active set becomes empty. Count incoming signals and active destinations separately to avoid confusing two different quantities.
03
Save the model and its complete short trace
The worksheet records node identifiers, directed edges, the initial active set, the activation condition, the timing convention and whether activity persists. Its trace is A, then B and C, then D, then the empty set. Under these rules, the empty state remains empty on later steps. If activity were retained automatically, or if D had a connection back to A, the trace could change. Record either modification as a new model rather than describing it as a correction to this one. Keeping the full specification beside the trace allows another reader to reconstruct every update.