But is it reasonable to expect that someone using the internet to mix music and going through 5-20GB per month should pay the same for access as someone that is lucky to use 1GB per month?
I think this is called 'tiered access' in the US. While bits themselves are free, more use results in more congestion and providers need to build out more network to cope. I think this is why I get a chunk of data to use at night when there is less demand (just like off-peak power pricing).
If use more electricity or water then I pay for it. Why not the same for the internet? At least the internet is flexible and slows down rather than just stopping (as an overloaded power grid does).
A dirty cheap ISP that has too much congestion and squeezes too many people onto its uplink will loose some customers, but not all. People that are reading email and news online might not notice, but gamers and online video viewers will, and are likely to go elsewhere. Of course this depends on competition which from the sound of it is sorely laking with US ISPs. Here in Australia the copper network is mostly owned by Telstra who lease the ADSL spectrum to ISPs to communicate with the ISPs' equipment at the exchange. The ISPs are then responsible for getting data to the internet. This means that two neighbours can be sharing the same multicore cable back to the exchange, but have dramatically different performance. The same applies with resellers of 3G/UMTS internet. It is common spectrum to the carrier's switch and then the ISPs' responsibility to get it to the internet. I think the UK and NZ (and probably the rest of the world except the US) operates similarly.