It is evident that the simulation of digital logic by a cellular automaton cannot proceed on the basis of voltage levels, because changes could only propagate from one place to another at a finite velocity. Something akin to pulses, which can propagate, is required; this supposes, in its own turn, the ability to distinguish between a pulse and its absence.
A traditional solution to the problem has been to maintain two lines, one for the signal and one for its complement. Additional states on a single line can accomplish much the same purpose.
Yet another solution has been to clock the signals; the complement of a signal is simply one which does not appear at the place and time that the signal itself should have done so.
In turn this can create the problem, either of synchronizing several local clocks initially, or distributing clock pulses emanating from a central source throughout the network. The central clock would still have to be started, and allowed to run long enough to reach each region before the circuit as a whole would have to start operation.
Practically speaking, it is better to have local clocks, and to prepare the initial WireWorld configuration meticulously, with attention to the phase and period of each clock.
Another detail which must be watched, is to be sure that any two successive pulses are sufficiently well separated that they do not interfere with one another --- especially when entering or leaving a logic element.