Unpredictable. Unfocused. Just like life.

Techdirt sums up why the "pay by usage" bandwidth model is bad:

However, the bigger problem is the transaction costs it introduces for users. Suddenly, internet surfers really need to see any particular website or service as being worthwhile. Just the act of making them debate whether or not it's worthwhile to pay up to do something represents a mental transaction cost that will slow down adoption of new services. Furthermore, as bandwidth has increased, many of the newer innovative services have come about to make use of that bandwidth -- which only drives further investment in more bandwidth, driving more innovative uses. It's a virtuous circle. Yet, by metering broadband connections, slowing down adoption of these new services, you slow down the innovation and hold people back from trying out or even creating new, innovative and useful services that would require more bandwidth. It's a recipe for slowing innovation online.

Just think of who benefits if a "pay per use" model is introduced. The phone companies make a lot more money/ For now anyway, until people just stop using high-bandwidth applications.

And Hollywood and the record companies would be incredibly happy because people would just find it too costly to download anything. No more digital distribution. But wait - CD sales would still be declining with no other revenue to make up for it.

No more YouTube. No more Joost. No more new innovation. So no need for more bandwidth.And pretty soon, when people stop using all those bandwidth-intensive applications, their "pay per use" cost will probably less than they pay for flat-rate pricing now.

Law of unintended consequences, right?