Tyler Cowen steps into an old issue, of how Internet usage ought to be priced.
In purely economic terms, the idea of charging Google or other “bandwidth hogs” does not sound outrageous.
The casual assumption here is that bandwidth is a scarce resource. But if so, where is it scarce? At Google’s data center? Near my house? Somewhere in between, on the Internet “backbone?”When Hal Varian looked at this issue, he took the view that the main pricing issue regarding the Internet is congestion. In theory, somebody who wants to send bits at a time when everyone else wants to send bits should pay a premium. Otherwise, if you are willing to have your movie download take place at off-peak times, the marginal cost ought to be zero.
So, there is no such thing as a bandwidth hog. In theory, there is such a thing as someone who is willing to pay more to get the fastest response during peak times.
However, the current internet protocols do not facilitate congestion-based pricing. Think of Internet packets as envelopes with very exact formats for the address. The format does not provide for a way to designate the envelope as “high priority.” Even if it did, the cost of reading the “priority bit” on every packet header would almost surely exceed the benefits of congestion pricing.
I think that this design limitation (or feature) would hamper a lot of the efforts of the telephone companies to charge differential prices for different content providers. However, I assume that the telephone companies must think that it is technically cost-effective to charge differential prices, or else they would not be lobbying for it.
Fundamentally, I think that the telephone companies over-estimate the value of what they have. The Internet backbone is critical, but the cost per user is low. The cost per user of the “last mile” is expensive, but the high-bandwidth lines into people’s homes are not critical. I think that we could, and in the not-too-distant future will, have a wireless last mile, which will be embedded in communications hardware, not centrally owned the way that land lines are today.
The land line “Baby Bell” phone companies are not going to design and enforce a new Internet pricing regime. Their destiny is bankruptcy.
READER COMMENTS
Eric
Jan 18 2006 at 10:40am
Dr. Kling,
You write:
There are numerous technologies to allow Quality of Service (QoS) available at the moment, and modern switching hardware and routers, of the sort that the big companies have purchased, comes with this technology built-in.
So they already have the infrastructure to do exactly this sort of pricing. However, these systems only scale so well — while you can easily prioritize google.com traffic over all others, it becomes more difficult to prioritize 10000 different content providers, to differing degrees.
I expect to see this sort of traffic-shaping rolled out as “partnerships” for an “improved user experience” between a given ISP and a handful of web sites / content providers as a result.
Xellos
Jan 18 2006 at 1:12pm
–“In purely economic terms, the idea of charging Google or other “bandwidth hogs” does not sound outrageous.”
Speaking from within the industry, Google and other “bandwidth hogs” already do get charged. They pay, probably by the Mb, to their provider(s). If you think I’m incorrect, try calling up, say, AT&T, and pricing out an OC3 line.
By the same token, users of the services are also already paying, this time to their ISPs.
The whole notion of charging some services more is just the telcos trying to double bill. It’s pure greed, and it’s going to do some serious damage to the first set of morons to implement it.
They may be relying on their last-mile monopolies to carry this through, but alternative technology (especially things like WiMAX) are too viable for this last any more than a matter of months. This could turn out to be one of the ultimate textbook cases of foot-gun disease.
Backbone, btw, doesn’t currently qualify as particularly scarce. The dot-com era caused tons of fibre to be layed out all across the country. Much of that fibre is currently “dark”, as in not being used at all. Between this and advancements in optical technology that allow a lot more information to be passed down the same strand, well, they’d just plain be asking for trouble. Especially given Google’s recent actions towards becoming their own bandwidth provider…
–“There are numerous technologies to allow Quality of Service (QoS) available at the moment”
The problem is that the proposals sound like they’re pretty much the opposite of what QoS is designed for. QoS is about making sure the high priority traffic gets serviced as much as possible, at the expense of the lower priority. What I’ve heard so far sounds more like rate-limiting, where some services (such as everyone-else’s-VoIP) will be restricted in how much bandwidth they can use, regardless of the utilization of the network.
In all, this whole scheme sounds a heck of a lot like extortion. Hope someone at the FCC is listening.
resigned
Jan 18 2006 at 6:10pm
Silly question–but I’ve heard it argued that a nominal tax on email would reduce spam. What are people’s thoughts on this?
R.J. Lehmann
Jan 18 2006 at 11:02pm
“The land line ‘Baby Bell’ phone companies are not going to design and enforce a new Internet pricing regime. Their destiny is bankruptcy.”
Who are these “land line” Baby Bells that you speak of? The companies that once were the local service Baby Bells have all long since been folded into just four companies — Verizon, Bell South, Qwest and AT&T. And all four devote the bulk of their resources to wireless at this point. Verizon (the former Bell Atlantic, which also swallowed Nynex) grew so much in the wireless market that it even replaced its former parent company, AT&T, in the Dow Jones Industrial Index.
Xellos
Jan 19 2006 at 8:59am
Here’s a link to an article that says Google has pretty much refused to even consider such negotiations.
R-Squared
Jan 19 2006 at 2:30pm
Wihle it is difficult for google to be charged, it is doable for google to charge people who want faster respones. I know plenty of people who are willing to pay 10 bucks to increase search response speed from 0.1 second to 0.01 second, although no human being would notice the difference. People pay for Starbuck too.
d.l.
Jan 20 2006 at 2:28pm
Is packet prioritization the least cost solution to the problem of providing real-time applications on the “best efforts” internet? I have no idea. But there doesn’t seem to be any reason to prohibit ISPs from trying it.
Either it is or it isn’t and the market should work to determine if it is. Is there any market failure here (other than monopoly/oligopoly power)? I don’t think so. And if monopoly/oligopoloy power exists, unless there’s price regulation, the entities that have that power will be able to restrict supply and raise prices whether they prioritize packets or not.
Comments are closed.