SWM ISO empirical operational data...

  There just isn't useful public data out there that is
  extensive enough, relevant, or complete. There are
  lists of "top 50 web sites in the UK", "top 10 most
  popular sites", and "15 top financial sites", and
  such, but I couldn't find anything useful that was
  based on true traffic analysis, or wasn't skewed by
  incomplete views.

Of course, many commercial providers consider this highly
confidential data and don't share it, assuming they reliably
collect it at all. The last commercial provider I saw with
any public traffic analysis was ANS.

The academic folks have bits and pieces from various
networks, but as I suspect you know, it comes and goes.

  Netflow data from various sources is proving to be the
  most useful, but I need more views. If anyone has any
  [Netflow or Netflow-like] top-speaker information that
  they would be willing to share, I'd very much appreciate
  it-- please drop me a note offline.

A number of the intrusdion detection/anti-DDOS vendors
may have probes located where they can collect some of this
information. And depending on the permision of the customers'
the exchange point operators could assist.