Eric Cosky

In the event of a large scale disaster, our communication infrastructure may very well become unavailable. Most of us, myself included, depend upon our service providers to keep such services running smoothly. However, if something terrible were to happen and if one of the effects of that were the disruption of modern communications services, many of us would feel stranded. For good or bad, we have come to depend in many ways on the easy access to information and the ability to communicate through the Internet. So what do we do to protect ourselves from this? How do we guarantee that our society does not regress to a pre-information age state?

One of the things we often hear is that the Internet is a fault tolerant system capable of reacting to, and recovering from, catastrophic failure. What isn’t discussed much is what would happen if our local providers were simply unavailable? The fault tolerance of the Internet doesn’t do us much good if we cannot even connect to it. This could happen for any number of reasons unrelated to the internet’s fault tolerant design. For example, something as simple as an extended power failure or un-repaired physical damage to equipment somewhere between your home and your provider could result in the lack of a connection for who knows how long, especially in the event of some serious disaster. Is this the last hop from the Internet to the individual that is most at risk because it is the least fault tolerant; there is often only one connection from the home to the net. Should it fail, the affected individuals will find themselves thrown back in time getting news from the radio or television, which in the event of a national disaster may not be available either. While there are definitely certain individuals, such as ham radio operators, who would not be affected quite as much, most of us would have no choice but to depend upon other people for communication.Fortunately, there it is an alternative means of communication that can help the rest of us get by during such a scenario. I call it Sneaker Net 2.0.

Sneaker Net is the term use to describe the physical transport of information from one location to another. Literally, putting on your sneakers and carrying some kind of physical media to another location. I did not invent this term; it’s been around a long time. What I propose here, for Sneaker Net 2.0, is to use a standardized means for physically transporting information from point to point in a convenient manner so that the data “package” will eventually reach its destination beyond the first drop point. In some ways, to define a postal service for electronic data that does not rely upon the availability of the Internet for data to reach its destination. This will take advantage of the 6° of separation rule of society which suggests that with sufficient participation, information can be sent to anyone on the planet by being sent (“routed”) through only a handful of intermediaries.

This is not a novel idea either, in fact prior to the Internet’s rise, e-mail used a similar scheme to route mail between computers that were rarely connected to a common network. Sending an e-mail required the sender to know how to route the e-mail to the destination using what is called UUCP routing as defined by RFC 976 (http://www.ietf.org/rfc/rfc976.txt). For example, to send mail to Joe@destination, you might have an address that looked like myconnection!randomintermdiary!destination!joe. This is really just defining which computers to send the data to on its way into the real destination. Each computer along the way would forward the package to the next computer in the list when it happened to be available. Before the always-connected nature of servers on the Internet, computers only connected to each other periodically to transfer queued files and e-mail. It took a lot longer back to then for e-mail to reach its destination, and this form of transmission has become effectively obsolete because e-mail today takes advantage of the always-connected nature of the net to (for all practical purposes) send e-mail directly to the destinations mail server. Unfortunately, if the network is unavailable for whatever reason, even this most basic of communication tools will be unavailable. In the event of a long-term failure of the network or merely a long-term lack of access to the network for certain individuals, we can and should take advantage of these proven technologies to provide the basic communication tools to be affected people.

One of the great things about technology is that things just keep getting better. High capacity portable storage devices have made it possible for Sneaker Net 2.0 to exist as a functional communications system because this increased capacity provides the space required to store the connectivity for a large number of people. This connectivity information is required to have some reasonable expectation of delivery. Let us see how the numbers work out by starting with the worst case scenario: the entire population of the planet, which is estimated to be approximately 6.5 billion which we will round down to 6 billion for the sake of discussion (besides, it is unlikely that every single person will participate in this network). If each participant were to track all of their connections, and periodically broadcast some limited number of connections, perhaps 5, that they expected to continue to exchange data with, this would mean that the connectivity data for the entire planet could be maintained with 30 billion entries. If each entry were a 128-bit value, this would be 210 gigabytes of connection data in addition to 66 gigabytes of user ID and connectivity index information for a total of 276 gigabytes of data.

While this is a lot of data, it is not beyond the realm of reason to expect portable media to be able to carry this and for computers to be able to process this in a reasonable amount of time in the foreseeable future. For now however, portable media is insufficient. Fortunately it is not necessary to maintain this much data yet, because the odds of this many people being involved is quite slim at least for now. That is not a solution, however there a reasonable compromise we can use to prevent the system from breaking down if it happens to be used more than the storage devices have capacity for. The solution is to essentially have individual computers track as much data as they can afford, while allowing anyone to broadcast their connection information when it changes. This connection information may be sent by anyone and will be propagated by everyone until it permeates the network. Individual nodes that are interested in tracking connection information may retain this information as they see fit, applied geographical and connectivity rules to help guide what information is saved. This request would propagate through nodes, and each node would slowly propagate their request to its connections. This system will be protected from flooding through a simple mechanism that tracks the number of messages from any given route and reduces the frequency of propagation based on how many requests have come from a given route, thus isolating spam related damage to the first node they interact with.

So, we have reached the point at which sufficient routing information to send data to anyone on the planet may be retained by anyone willing to purchase a 300 gig hard drive for storage. So long as Moore’s Law increases capacity faster than our population rate, this will continue to be the case. Since this is the worst case scenario, and in all likelihood there will be far fewer people interested in using this technology than technology has the capacity to support, it is unlikely that we will hit any storage related problems. In other words, technology is not limiting our ability to solve the problem of knowing how to route messages to anyone on the planet. We have the room for all the data we need, we just need to use it effectively. The algorithm required to find the route a message would need is a well understood problem that has been solved. Choosing an identifier for each individual is another important aspect of this system, and this too is a problem that has a known solution. From what I can tell, none of the issues that might prevent this idea from being possible are insurmountable. It is really just a matter of working through the implementation details, which is no small thing, but is clearly an achievable task.

One of the basic problems is moving messages from node to node using the predefined routes embedded within the message. Routes can take many forms. A single route may include physical transport to an off-grid computer (ie disconnected from the Internet) which will periodically have data moved on to another physical transport, eventually reaching a computer that occasionally connects to another computer through a packet radio, WiFi mesh, dial up modem, or Internet connection to transfer routed data. The data may continue its journey through any number and type of connections on its way to its final destination.

The actual transfer of data between nodes can occur in any number of ways. Simplest of these will probably be the use of high capacity USB flash memory drives that people simply connect when they happen to visit. A properly configured computer will automatically detect the USB drive and will copy data that has been marked for routing through it, and will copy data to a USB drive that has been marked for routing through the owner’s computer. A more advanced version might someday perform an ad-hoc WiFi connection automatically transferring data when in the presence of a participating node, automatically extending the reach of the network. These might even take the form of a solar powered system residing in a fortified structure designed to withstand catastrophic events, serving as a last means of defense for our communications infrastructure.

There are other uses for this idea beyond communication during times when of the Internet is not available to people who normally have access. There are many people who simply do not have routine Internet access and could use this to at least get e-mail. People living under oppressive regimes may not want to have their communications on a public network where they may be intercepted and this could be used as a form of courier message system that would have a broader reach than system limited to a small group.

Clearly, this form of communication with its extremely high latency is not something people would choose to use when the Internet is available. However, if the Internet is not available, there is no reason for its absence to prevent us from communicating at least as effectively as physical mail. This system could be implemented using today’s technology and has the potential to provide long range communications in the face of a disaster which might otherwise leave us in the dark.