• When it comes to communications infrastructure, what you really care about isn’t total bandwidth but peak bandwidth. The pipes have the same capacity at all hours of the day and all days of the week. So during non-peak hours, there’s likely to be plenty of extra bandwidth available. But your heaviest users by bandwidth aren’t necessarily your most expensive users–many of them may use a steady, but low amount of bandwidth around the clock, or they might use a lot of bandwidth during non-peak hours. This is why cell phone companies give you free night and weekend minutes–they’ve got plenty of extra capacity during those times.

• Flat-rate pricing economizes on arguably the most important resource around: human time and attention. When I want to check my email or download the latest video from YouTube, I just do it. I don’t think about whether I’ve exceeded my bandwidth ration, or how much the additional bandwidth will cost. This is particularly important because many consumers have absolutely no idea how much bandwidth various services use, and there’s probably no way to explain it the them.

• It’s not obvious to me that bandwidth is all that expensive anyway. There’s loads of fiber in the ground, and my impression is that the cost to move a bit across the country keeps dropping and dropping. It’s silly to economize on resources that are dirt cheap.

• Metering requires pricing above marginal cost, which leads to inefficiently low bandwidth usage. (that’s especially true during off-peak periods, when marginal cost is zero) That’s bad because there are many low-value uses that taken together may add up to significant value. And (as Templeton points out) often you don’t know if a particular use is going to be useful until a lot of people have started playing with it.

The type of pricing I think would make the most sense would be some kind of priority-based pricing: if you pay a premium rate, then you get more bandwidth during peak hours, otherwise you get throttled during peak hours, but with unlimited usage the rest of the time. That’s roughly what we have now in the sense that the really expensive pipes are priced based on reliability and consistency more than they are on peak throughput.

But doesn’t the idea appeal to your libertarian instincts? If light users are in fact subsidizing heavy users, a switch to usage based pricing should result in vastly lower broadband costs for most users. It would also encourage efficient use of bandwidth.

The argument is that consumers wouldn’t go along, but I’m not convinced of that. It seems like it’s just cover for the fact that the vast majority of us overpay for the bandwidth we use. I don’t think we’re likely to see such a model absent serious competitive pressure.