In other words, bytes are free, cable is not.
This would be true in a closed system such as a home network. The cost to transfer a file from my server to my laptop is practically free. The costs there are my equipment costs and the electricity required to run the devices.
Broadband providers are not on a closed system--they have to connect to the internet backbone and other providers. Unfortunately, the broadband providers are pretty cagey about these costs. In a moment you'll see why.
Over at the BITS blog at WSJ there is an article that tries to determine the costs of data. The article peters out near the end without an answer, but Dave Burstein offers some insight in the comments.
The data costs work out to about $10k/Gb. According to Burstein, this comes down to somewhere between $.05 and $.15 per gigabyte per customer. If TWC is charging $1/GB then that's an astonishing 700%-1500% markup! If this is true, it makes sense they don't want their customers to know how much the cost to connect is.
But that's not the end of the story, because that only tells us where they are right now.
Using these assumptions we can figure out the break-even point for HSD customers. If the average customer is spending $50 per month for their broadband service then the maximum break-even point is 500GB/mo. (with an average of $.10/GB/customer).
When I first figured this out I was a bit shocked. My expectation was that the number was going to be ridiculously high--say 5TB/mo. But 500GB/mo. is absolutely within the realm of possibility.
I'd like to find some hard numbers on data transfered per month, but I've personally transfered more than 500GB in one month. It doesn't happen often, but it does happen. There are plenty of stories out there of individuals hitting Comcast's 250GB/mo. limit.
These numbers could certainly be pushed around a bit. This rough estimate reads a bit high because no other costs are factored in--it assumes that all the money customers pay for HSD is pure gravy. If we assume that the cable companies set the price point at their standard 70% markup then the break-even point drops to 205.9GB/mo.
But 500GB/mo. is a probably a good compromise figure. Costs will probably decrease, not all of the traffic has to go extra-network, etc.
Although few users are hitting 500GB/mo. now it is certainly plausible to envision an increasing percentage of users hitting that within the next 2 years. Within 10 years that could easily be the norm.
Both sides of the argument get knocked around by this. Charging $1/GB is absolutely usury. But arguing that the costs are negligible or that the exaflood is pure fantasy doesn't work either. There should be a comfortable middle ground and both sides covering their ears and stamping at each other won't find it.