A new view of network externalities

For those of us who have been involved in running or regulating network infrastructure, there's been a sea change in the framework for deciding on appropriate policy concerns.  In the old days, all we had to "worry" about were what the economists call network externalities. These externalities could either be positive or negative in nature.

A positive externality occurred every time someone would join up to a network, say, by subscribing to the early telephone system.  While each person received a certain value in subscribing, all other uses also received an enhanced value from that person joining the system.  Why? Well, simply put, everyone could now reach an additional subscriber at minimal extra cost.

A negative externality would occur when the network would become congested.  In such a case, each additional subscriber would slow or degrade the service quality for all the incumbents, causing a need for capital investment to restore service quality, or a time-of-use pricing regime to ration service during congested periods.

In the days of regulated monopolies, a government body would intervene in the design and operational aspects of the network service to decide what level of service quality was appropriate, what level of investment was needed, and what the pricing design should be.

Now, though, we've arrived at the Wild West of network service, the Internet. A 2012 article by J.M. Glachant in the Review of Economics and Institutions summarizes:

How does the inability of public policy makers to keep up with things become evident?  Let me provide an example from a talk I heard by Tom Leighton, the CEO of Akamai.  Akamai's role in the Internet world is to optimize traffic flow for its customers. If you are Facebook or Google or Amazon, you want your users to have an instantaneous response when they click on a link: People are much more likely to buy something if there is no perceived delay when they do so. Akamai, therefore, has set up a worldwide network of thousands of servers to help route traffic and speed up information flows between its large commercial customers and the millions of computer and device users.

But even Akamai can suffer from negative network externalities.  The number of users in the world is growing, the number of apps and sites is growing, and the number of objects on each web page is also growing.  All this suggests an exponential surge in traffic over the coming years.  For Akamai, the cost associated with installing thousands more servers would be large, and even if possible, would not result in the end-to-end quality service that is desired.

So the answer has been to reach out further along the branches of the tree and to enlist individuals' computers to be miniature servers in the Akamai optimization network.  Here's how it works, emphasis added.:

The Akamai NetSession Interface is a secure application that may be installed on your computer to improve the speed, reliability, and efficiency for application, data and media downloads and video streams from the Internet. It is used by many software and media publishers to deliver files or streams to you. 

If the software or media publisher uses the feature and if you enable it, NetSession can also use a small amount of your upload bandwidth to enable other users of the NetSession Interface to download pieces of the publisher's content from your computer. The NetSession Interface runs in the background and uses a negligible amount of your computer resources or upload bandwidth when you are not actively downloading content. 

How is it installed and how do you give permission? It is enabled when you "agree" to the terms and conditions of one of the applications to which you subscribe.

Akamai NetSession Interface downloads or streams content to you only after you have requested it from your software or media publishers. 

And, as the site explains, you can opt out at any time.

But who actually reads those terms and conditions? Well, no one really, so this was news to me and many others when we heard Tom describe it at a recent seminar.  I asked, "Why should I feel good about having this software installed on my computer?" His answer, "It will make your own service faster, and you will have contributed to the Internet ecosystem in lowering costs to everybody."

Putting aside whether I would actually ever be able to detect such cost savings, is there something immoral or manipulative in this process? That question, too, was asked by an audience member at the seminar. Tom's response was, in essence, amoral. He said that this was a required technical fix to the massive growth in traffic on the Internet, and he repeated what's on the company's website:

The NetSession Interface is safe and secure and does not contain spyware, adware, or a virus. It does not gather and transmit your personal information, nor does it harm your computer. Its purpose is to be a tool to improve the speed, reliability, and efficiency for downloads and streams.

I think a number of us were concerned about what we heard. Was Big Brother going to take over our computer for some nefarious purpose? Tom is certainly trustworthy, but what if some nasty person becomes head of Akamai? Or what if some circuit designer deep in the company attaches a nasty bug to our computer?

Well, the truth of the matter is that if you are connected to the Internet, you have bigger things to worry about.  The chance of your computer being corrupted by a bad actor somewhere in the world--a criminal organization or an insidious domestic agency or foreign government--is already remarkably high.  If you are concerned about privacy or malware, you truthfully should be off the Web totally.  So the incremental risk of being part of the Akamai distributed network is small compared to what you are already experiencing.

So, oddly, Tom's amoral reply is actually the right one.  His job is to optimize web traffic for his customers and to design technical fixes to do so. If one of those technical fixes is to enlist your computer in the distributed network, that's what he needs to do. (That kind of fix, by the way, is a nice way to balance the positive and negative externalities associated with network expansion.) The only appropriate public policy response to the plan that we could devise is the one already employed by Akamai, to disclose the existence of this fix and to give you the right to opt out.  Could the company do that in a more outgoing way, so we might be more knowledgeable and make the choice not to play? Perhaps, but truly, how many of us would do so? A negligible number, I'd guess.

If this story is typical of what to expect in the Wild West of Internet externalities, in contrast to earlier network services, public policy makers will find themselves ever more irrelevant.