I would think this would be a breach of contract, if not false advertising. Is there a reason they escape traditional routes of retribution for these issues?
Isn't it also packet loss ? The Mathis equation:
http://en.wikipedia.org/wiki/TCP_tuning
tput <= mss/(rtt*sqrt(plr))
if you have no packet loss then rtt starts to matter less it seems to me.
In the page you linked to, it mentions that Mathis formula does not apply if packet loss is low. In that case, you want the formula above: throughput <= RWIN / RTT. If your window is large enough, then round trip time isn't a big deal (unless you have a really big connection)
RTT is a large factor in throughput at the beginning of the connection, when tcp slow start is restricting the opening of the receive window.
So they could lie to me about that too? I'm lucky to get the bandwidth they promise at peak hours.