Well, there's also the subproblem approach: (Note, I'm completely speaking out of my ass) - Launch a string of relays between orbit & the moon. They may only

Well, there's also the subproblem approach: (Note, I'm completely speaking out of my ass) - Launch a string of relays between orbit & the moon. They may only work once a month, but that just means you have to have the judges there that day :-) Each one can be a control board, radio, and solar panel. Swarm satellites, I think they were called, only in a straightish-line. - Make the actual lander dead, dead, dead simple. Like forward & turn-in-reverse RC car simple. Make it small & rugged, but put little/no control electronics onboard. Power (unless you can beam it in from a near-moon satellite into a battery pack -- it's fine to let the rover sit around 29 days, 23 hr, 45 min of the month, to let it charge), and receiver(s) to control the motor(s). - Data transmission could be a controllable mirror, and a relay with a good camera! But I guess this all really depends on another question: what's the cost to weight function? Purely linear? Is it cheaper to do two launches of weight W instead of one launch of weight 2W, over a minimal value for W? If a launcher for a standard size object can be the primary cost of the system, we could launch relays and landers in swarms and leave the big money for the launcher. 10% success rate? Launch 15 to be sure. Also, one situated between three high-altitude balloons would look pretty sweet.