I prefer this because it avoids all the speed of light lag, photon intensity and other problems.
Instead of beaming a laser at the moon, simply set up an array of lights, many, many miles long. Assign each light a clock. We will need the clocks to be and remain synchronised within our selected reference frame (which is the frame we originally used to fire our laser from).
We set up our system so that the lights switch on at arbitrary times according to their atomic clock. We can send the entire array with a program to set up the times we want them to flash. The program is sent at some arbitrary time before we want the flashes to occur.
Now we can have apparent signals travel around our array at any speed we want. FTL, simultaneous, whatever. That's because there is no real signal; the information isn't travelling through the array but instead stored at each individual point. The lights don't signal each other but instead carry out pre-arranged actions. For all intents and purposes each of the lights is an independent system with absolutely no dependence on any other. At this point we can treat each light flash as an independent event and the relationship of one flash to another can be treated in the same way as the relationship between any two events in special relativity.
The laser is effectively the same setup, only with the condition that a single light (a single point on the moon) can be triggered to flash,
relative to the laser origin (eg, our reference frame). We might be able to cheat and illuminate multiple points using speed of light delays, but that's insignificant here. The program, in this case, is the pattern the laser follows, and we could entirely simulate this with a program that only triggered a single light at any given time (by their synchronised clocks).
The important thing to see here is that the different points don't transmit the program to
Using the lights as our starting point and treating the laser as a special case should help with working with these problems. It's also a fun way to start exploring the concept of non-local variables (although it does still need some modification to get all the way), which crop up in some Hail-Mary end-runs around
Bell's theorem while retaining some manner of realism.