Publishing transaction data is only the first step. The execution engine still needs to know which inputs are valid and what order to process them in.
Matrix-node reads the public input history, validates it, and prepares execution batches.
The engine works from that ordered record, not an operator's private queue.