What I read from his post was to directly launch the test article to Mars, which probably wouldn't work because it wasn't built to flight standards and is too heavy. I agree that building more copies of the existing MER design would be a good idea because all of the R&D costs have already been spent, so the per unit cost would be much lower than doing R&D on a new design. Counting launch costs, I suspect the price per MER would be closer to $150-200 million a copy but that's certainly better than the $420 million per copy today. We'd probably want to spend some additional R&D to incorporate improvements based on lessons learned, but we'd have to resist the engineer's temptation to "hobby shop" the project. As a group, many engineers have a hard time holding to the "If it ain't broke, don't fix it" philosophy.

It'd be difficult to put more than a few MERs on Mars at any one time simply from a command and control perspective. There are only so many antennas available on the DSN, and there are only so many people to operate them in the control centers. The JPL controllers are being stretched pretty thin just operating the two MERs operating today. Hiring a bunch more people would help, but that drives up the costs and does nothing about the DSN antenna limitations. Putting Mars Network style repeater satellites around Mars could address the DSN antenna issue, but that costs, too. However, Mars Network repeaters could reduce the need for high gain antennas on each rover, potentially reducing the cost of each rover. You could still use the existing low gain antenna for contingency operations and it might have enough EIRP to still have good data rates to the repeater satellites in orbit.