Mars rover is not real time, doesn't actually do anything, 5 of them failed, there is a team of highly trained scientists and engineers running the project and the whole thing has cost almost $1B.
I was talking about the bandwidth, not the rest. Yes, Mars rovers were stupidly expensive and difficult. It's freaking Mars. I'm pretty sure we don't need to launch a rocket to get stuff to Illinois, that we don't have a latency of 4h20m and that we can have one guy or two on the location to take care of any issue.
The current state of the art does not seem to suggest that it is possible for $1 workers in India to dig a basement in Illinois with a tractor controlled over the internet. It certainly is not possible for less than what it costs to just hire local workers to do it.
Frankly, without numbers I'm not persuaded either way. Consider the advances in farming vehicles it seems plausible to me.
#3 is a big deal. Planetary probes are way, way, way overdesigned compared to what you need on the earth, and all those "ways" cost exponential amounts of money. It probably wouldn't make sense to try to do everything on the job site remotely, but there's really no reason a teleoperated tractor has to cost substantially more than one with a guy in it.