We have always built 2 robots, since 1997. They have ranged from very rudimentary drive bases to almost complete copies. I'd say our average 2nd robot is 65-85% accurate.
Our construction method (lots of flat plates and little welding) lends itself to making/using spares easily. We take advantage of that (and our capable machine shop) to make spares at the same time we are making originals. It makes the parts legal, and it gives us a set that we can modify while the other parts are assembled (and hopefully running) on the robot. Note: for complex parts this is not always practical, and for some parts its not necessary.
While I agree that auto mode is more likely to drive more teams into a 2nd bot, I think the need for practice is more than enough to drive competitive teams there anyway. We get as much or more out of practice than we do out of auto-experimentation-after-ship.
Joe, I don't think auto-mode is an Experiment any more... its a feature
). I belive in the "6 weeks" concept, including the need to make auto mode work during that time. We are all going to get better at running auto mode ... someday.
Ken