Firstly, I applaud Mike for taking on this fight.
While there has to be a balance between delivery costs and service quality, traffic management activities need to be more clearly defined, as they have the potential to, and infact do (as is clearly evident on this forum) impact the whole user base, not just those deemed to be "excessive downloaders". VM's line about traffic management being beneficial to 95% of their customers is therefore up for debate. That is unless we represent only those pesky 5%, which i highly doubt.

(Why don't VM penalise
consistently heavy transfer users by implementing throttling only after a period of excessive usage on successive days?)
I'm unable to host a 12 person game on Call of Duty: World at War (on the PS3 console) without it kicking everyone out half way through (this is with no other load on the WAN link). I know various people who are able to host these games flawlessly on ADSL connections that provide (in some cases) less than half the bandwidth of my 20Mbit cable connection. So bandwidth isn't the problem. Nor is my connection according to VM Tech Support who assure me my connection is fine. My ping response times which range wildy between 15ms and 50ms (with the odd spike to 70ms) to bbc.co.uk are apparently "within normal limits", dispite the fact that a healthy connection should show much stabler latency statistics. Response times to my default gateway vary wildly between 6ms and 30ms (..it's only one hop!).
Is this gaming problem down to STM, and the fact that the game uses P2P UDP traffic? Is it down to over subscription on my UBR? I'll never know because they can't tell me what's going on behind the scenes in their network at the time of failure.
Providers should be forced to provide realtime information on traffic management (or anything that might affect ones ability to achieve their stated service level), on a per customer basis, so that we could easily differentiate between a poor performing connection, and a managed connection.