What's the best way to wiretap a network?

I believe you're correct. In fact, I first learned of these devices
from government documents during the Carnivore discussions a few years

    --Steve Bellovin, http://www.research.att.com/~smb

Lots of people seem to be making the assumption that all networks work
the same way or everyone wants the same data. Tapping an OC192 SONET
circuit is expensive, but relatively straightforward. Tapping a V.92
analog modem is expensive and not straightforward. Tapping WiFi-to-WiFi
traffic is cheap, but only if you are local. A sniffer on an upstream
switch won't see the traffic below a network access point.

But a Title III warrant for "full content" is relatively difficult to
obtain in the US. The public reports filed with the courts show a small
percentage of wiretaps require full content. What's also interesting is
if you read the various public submissions to many different working
groups since the Carnivore discussions a few years a go, you'll notice a
dramatic re-definition of more and more data as "call identification
information" instead of "content."

The public proposals also seems to be somewhat arbitrary which provider
gets "tasked" with collecting the wiretap data. Should the first mile or
last mile or middle mile provider be tasked with isolating call
identification information and decoding it?

So what is the best way to wiretap a target using public WiFi hotspots
connected through multiple wholesale providers and service providers
to collect call identificaiton information to call identification
information about who the target is communicating with through multiple
application protocols including Webmail, IM and massively multi-player
role playing games.