Internet Service Providers in the USA are trying to apply bandwidth caps to their users, with those caps being 2, 4, or 5 gigabytes-per-month for wireless users at various price levels and generally 250 gigabytes-per-month for home users. Most of the press coverage of this issue comes down on the side of consumers but lately the ISP publicity machine has been revved-up and we’re being told that bandwidth caps are necessary, even inevitable. This is, as my 87 year-old Mom would say, BS.

Provisioning is what ISPs call the amount of Internet backbone capacity they buy per subscriber. This number is always less than the amount of bandwidth we think we are buying because most of the time Internet connections aren’t used at all and ISPs count on this to keep costs under control. If you are buying an 8 megabit-per-second connection from your ISP, he in turn provisions you with around 50 kilobits-per-second of backbone. This data arbitrage is part of what makes being a broadband ISP so profitable.

The reason this is an issue, we’re told, is because ISPs fear we are changing our consumption patterns. If we all switch to getting our television and movies over the Internet then there’s no way 50 kilobits will be enough.

So they’ve taken to publishing charts like the one above, which came from a blog here. It shows the massive increase in data consumption at an arbitrary ISP. Arcing into the heavens it looks like ISP costs are exploding and will shortly become infinite unless data caps are applied. If the ISPs can’t make money, we’re told, then we’ll all lose our Internet service.

They’ve become “too big to fail. ”

Remember that one?

Fortunately, at the same time bandwidth consumption is going up, backbone costs are going down and have been doing so for many years. The basic unit of ISP backbone expenditures is called IP Transit. Here’s a chart I found showing IP Transit prices per megabit in several cities over a period of years.

There are a couple interesting points I can make about this chart. You’ll notice for example that backbone costs in Tokyo, where broadband connections typically run at 100 megabits-per-second, are about four times higher than they are in New York or London. Yet broadband connections in Tokyo cost half what they do in New York, and that’s for a connection at least four times a fast!

So Softbank BB in Tokyo pays four times as much per megabit for backbone capacity and offers four times the speed for half the price of Verizon in New York. Yet Softbank BB is profitable.

No matter what your ISP says, their backbone costs are inconsequential and to argue otherwise is probably a lie.

Now let’s try an apples-to-apples comparison of these two charts by adjusting them to cover the same time period like this:

Consumption went up and prices went down. In terms of backbone cost per subscriber, ISP costs have been flat for years.

That 250 gigabytes-per-month works out to about one megabit-per-second, which costs $8 in New York. So your American ISP, who has been spending $0.40 per month to buy the bandwidth they’ve been selling to you for $30, wants to cap their maximum backbone cost per-subscriber at $8.

That doesn’t sound unreasonable on the face of it. Capping consumption at 20-times the provisioning level doesn’t sound so bad, but I think it sets a dangerous precedent.

These data caps are actually a trap being set for us by the ISPs.

Data caps that may make logical sense today make no sense tomorrow, yet once they are in place they’ll tend to stay in place.

IP Transit costs will continue to drop. That $8 price will most likely continue to fall at the historical annual rate of 22 percent. So what’s presented as an ISP insurance policy is really a guaranteed profit increase of 22 percent that will be compounded over time because consumption will continue to rise and customers will be for the first time charged for that increased consumption.

This isn’t about capping ISP losses, but are about increasing ISP profits. The caps are a built-in revenue bump that will kick-in 2-3 years from now, circumventing any existing regulatory structure for setting rates. The regulators just haven’t realized it yet. By the time they do it may be too late.

Most U. S. broadband customers don’t get anywhere near that 250 gigabyte cap. The few who do hit those limits are big gamers or file downloaders for the most part. Maybe they do take unfair advantage of the system, but the question is whether this is the proper way to control their consumption?  I don’t think it is.

In time we will all bump into these caps and our Internet bills will suddenly double as a result, circumventing competition and ending a 15 year downward broadband price trend.

ISPs win, we lose.