Most of these concerns can be overcome by only polling sites that request
or consent to being polled. I'd assume that it wouldn't be done at all
unless a way of polling can be found that introduces only trivial overhead,
but if an individual site doesn't want to put up with the amount of added
overhead, they surely shouldn't have to. Likewise, a reflector which was
used for real work would ask not to be polled, or wouldn't ask to be
polled. As to people who object to having their activites widely
advertised, one solution would be not to have names in the report that's
produced, only have number of people present. There are other solutions
too, including the non-solution free market approach, which says people who
don't want their activity advertised will only connect to server's they
know aren't polled.
I agree with you that there are a number of different trends developing in
CU-SeeMe use. One of the most prevalent at present is the social "chat"
function, and I imagine that this won't decrease much in the future either.
People like talking socially to strangers, just look at IRC, or some Usenet
groups, or BBSs. The polling device seems to be designed for that type of
use. I don't see any insurmountable problems it would cause for any other
type of use, however.