reply to dutenhnj
Re: Physics fail We need a solid definition of data before I can agree with your statements. I suppose if data is interpreted as a particular pattern of matter, the pattern can be altered so that the previous pattern can no longer be recognized or interpreted. If we consider the original pattern to then be destroyed, I suppose that works for me but I think we're veering in a direction that has no beneficial argument to the debate at hand.
I believe the debate at hand to be whether or not an ISP has variable infrastructure costs that can be directly correlated to customer usage. To be completely fair, most of today's infrastructure will use less power when idle than when loaded. In this regard, and only this regard, would an ISP have more cost based on usage intensity. If we agree with this assumption, the next issue is are those costs significant to warrant metered billing? I will speculate and say that these costs are NOT overly significant and could even be less significant as EOL devices are replaced. For instance, standard x86 rack servers in our enterprise typically contain two sockets and each socket has two cores. They are now EOL and being replaced by servers with two sockets and each socket has six cores. Since the infrastructure is virtual, one new physical servers enter and three old physical servers leave. If we assume network devices have a similar replacement profile, the variable electricity costs should go down or remain flat if the price of electricity rises to offsets the lower usage.
Therefore variable infrastructure costs don't seem to present any favorable justification for metered billing.
I tend to believe the panic isn't about a looming bandwidth shortage. I believe ISPs are in a panic because of a looming glut. Technology is getting faster and cheaper (Moore's law) and unless we continue to innovate new ways to use more bandwidth, supply will eventually outstrip demand.
When I first had a high speed connection installed in the Summer of 2000, the speed was 512Kbps down and 128Kbps up. Because the cable provider was new to offering data as a service, it was tough to reliably stream a single 128Kbps audio feed. It also took minutes, not seconds to download MP3 songs. Today I would guess every network-connected device in my house (4 laptops, 2 iPhones, 3 iPod Touch, 2 XBox 360s, 2 MP3 streaming radios and one Internet-capable TV) could all play a stream at the same time and barely scratch my 12Mbps down and 1Mbps up connection. Nor would heavy usage (each used for say 6 hours per day) of all these devices rile my ISP (approximately 145GB of downstream per month).
Here's the key question -- in another 10 years, will video streams become just as insignificant? Isn't this what keeps ISP owners up at night wondering how their world will change in the 10 years?