What perplexes me is why we might wish to use WMWR when ICS USB seems considerably faster? Pinging an arbitrary remote IP address, I get consistent WMWR latencies in the range of 320-450ms with one or two lost packets every 15-25 packets, sometimes more frequently, and an occasional very long latency (over 1 second). With USB, latencies are around 150ms, rock solid, losing few or no packets. So unless portability (phones are pretty portable themselves!) and/or multiple client computers are factors, what's the advantage? Moreover, there is real security on a wired USB connection versus wireless Ad Hoc...
Another question:
On my AT&T Tilt, I can connect to my laptop via ICS BT, and the laptop can see the phone with TCP/IP, but the laptop can't reach the Internet. The laptop doesn't report any available BT services on the phone, and the phone doesn't report any services on the laptop. Is there a Wiki that explains this? Anybody know what I'm doing wrong, or the secret?
Lastly, on the issue of detecting tethering:
ICS is certainly given the MAC address of the tethered client -- it's inherent in connection sharing that the host immediately above a client receives the MAC address of the client. Now, usually that MAC address is discarded by the host (here, the phone) when it, in turn, communicates with it's own host (the provider) -- normally the phone would just report it's own MAC address to the provider. But ICS could certainly be written to pass along client MACs, if any, as well. If WM6.0 ICS doesn't do that (and I assume it doesn't), then it stands to reason that one should stick with the v6.0 version of ICS. My question is, can a provider push a new version of ICS at client phones without the user knowing or pro-acting? Would the WM security paradigm permit that? Satellite TV providers do it all the time, to their Linux-based receivers...