It's worth reading Ballard's memoir. During the 1960s in oceanography he went through many of the same fights as were happening in space exploration in terms of the utility of man-on-the-scene exploration, of visual imaging vs more abstract remote sensing, etc. Ballard of course was a big early advocate of manned exploration of the ocean bottom via highly mobile submersibles (e.g., Alvin).

Since then, he's turned around 180 degrees. He's still very big on the same methods of exploration -- on-scene, very agile, using high detail visual inspection and sample collection -- but now thinks that for oceans telepresence totally beats the expense and disadvantages of manned deepwater submersibles. It's still "manned" in the sense that his approach is to use small, nimble vehicles with very high real-time bandwidth connected to human operators. I suppose that's "remote sensing", but in a very different way than towing magnetometers and sonars on the surface.

The problem of course is that doesn't work for planetary science because of the lightspeed delay, and even w/o that there are coverage and bandwidth problems (e.g., MERs relying on intermitent relays, even then bandwidth is too small for real time continuous video, etd). The Moon is an exception, of course (see Lunakhod).

Thought experiment: if Spirit and Opportunity had exactly the same capabilities, except that there was continuous communication with no lightspeed delay, how much more quickly would we have accomplished what we've done in the past 5.5 years?