skip to content






The PI Blogs

End-to-End Social Networks

Andrew Lippman

January 1st, 2010

Networks are primarily about connections and secondarily about endpoints. Sometimes the idea is to get to a specific place, for a specific service, but equally often, it’s about the raw ability to rapidly and flexibly to get anywhere. The early mistake about telephony is revealed in the statement that “no one wanted to talk to Ohio” (or whatever the real statement was…) It’s not about the ability to reach Ohio, it’s about the potential get anywhere, under your control. What happens when we design the network and its attachments with that goal in mind rather than think about it from the perspective of the sites?

Let’s look at one example, emergency services. US telephony’s 911 service was invented 42 years ago this year. It was premiered in Haleyville, Alabama, by a small operator who got wind of the fact that AT&T was working on such a service, and he wanted to scoop them. He rewired the exchange to make the shortened number ring directly in the police station, making it simple to call for help. Ultimately, the idea of a single nationwide police number was coded into the regulations by which telephony operates: 911 is now considered a telephony service, mandated for those that provide it. In 1968, that was a great step forward but now it is a misdirected burden.

Consider what society needs for emergency services, and how modern networks ought to be able to provide them.

Emergencies are no longer solely about getting help for a fire or heart attack. Nor are they purely personal affairs, directed at or for a single individual. Consider the recent attempted attack on a Detroit-bound airplane where passengers provided the “service” (saving the plane). Early reports portrayed this as a fine solution. Indeed, there is discussion that the best result of increased airline security is that it has made people aware of the fact that they all have to pitch in to help when it is needed; they can no longer just rely on a remote entity—a site—to solve the problem for them.

David Brin noted a similar example during a talk he gave at the Media Lab. In the case of the World Trade Center attack, people with cell phones saved the most lives; the authorities were not organized to evacuate the building, fight the fire, or even to know who was inside—but the friends and relatives did so effectively.

Even in the case of personal need, a specific or single site may not be the whole answer. When you are sick, it is a matter for your family and often the community as well as the doctor. Some medical matters are a spectrum, not a point event—a coordinated and well-publicized activity can be both advance warning and a diagnostic aid. Jon Moore’s HealthMap allows people to post outbreaks and provides a simple interface to scan for emerging epidemics. While open to noise and false positives (much as other medical screening is…) it provided warning of the H1N1 epidemic far earlier than the official test results, and helped identify the source. Imagine the emergency is about food poisoning rather than heart attacks and the benefits of early warning and identification become crystal clear.

What is needed in a network for this kind of fluid, flexible response? First, our personal networks need to be able to operate “radially” rather than centrally. Until the advent of smart phones, in the last few years, a portable phone was technologically restricted from talking to another one; it could only communicate with a tower. The first phone-to-phone bluetooth interactions were a bar joke that relied on the other users leaving their phones in discovery mode, and some people just beamed jokes and messages across the room. Discovery mode was soon feared as a security flaw and is now routinely disabled.

But we can add some intelligence to this, as Nadav Aharony and Polychronis Ypodimatopoulos of the Media Lab have done, with the codification of a “social aura” and a control structure for determining what information you distribute and what you are willing to listen to. This transforms bluejacking from a technical bug to a programmable feature, based not on radio operation but on user intent.

We need an interface to enable individuals to set up the suite of connections and policies in which they are willing to participate. I think this is as much an interface issue as a policy point—an extensible framework can let people keep control and keep it simple enough to manage.

Finally, we need an open social interface that readily admits to extension. That’s where the end-to-end principle applies to social networks. Sure, communication is about the specific places that one can reach; that is important and will never go away. But just as much, it’s also about presenting the option to reach anyone, anytime, and about learning and discovery. That’s what engagement with society is all about, and that’s what works both for emergencies and for everyday life.

Leave a Reply

You must be logged in to post a comment.