Some of the most powerful results of this current batch of landers have been through data fromclose contact with the soil and rocks. The finding of Olivine in the dust by the Mossbauer spectrometer is a powerful data point, and we see from the amount of dust cleaned off Adirondack that everything is potentially shrouded in a camouflaging layer of dust.

Clearly, if we were talking about dozens or hundreds of robot missions, some would be airborne to give us a different perspective and sampling of the terrain.

I have worries about aircraft, since thay seem to have only a few hours endurance, at best, which requires a local high-speed radio link to an orbiter with a big memory to store everything for later transmission (or an even better direct-to-Earth broad pipeline).

If you can make a balloon work, with a tehered rover or anchor that re-lands repeatedly, then perhaps we can compomoise on a good mission to sample medium-altitude imaging with ground truth.

[I have to say, that the DIMES instrument has been one of the most useful and important instruments this time round - it gives us an immediate context for the lander location, and lets us work on detailed location and identification of horizon features and nearer topo.]

[I also note, sadly, that we STILL haven't had any reports from the radio-tracking people of which small crater Opportunity is in and hence no high-res MOC images.]

But as for the actual design of a mission, you can have rovers with an attached balloon that unreels to several hundred metres height with a "webcam" - OR - zeppelions with a landed component, and anything in between. There is a huge range of parameter space here that we have yet to explore in terms of utility.

What about hoppers, or a set of disposable rocket or mortar launch cameras to get an aerial context image every km or so, in the same way that you can build an Estes mini-rocket and parachute a webcam back to Earth, capturing photos of your neighbourhood from the air?