Email, Scale-Free Networks, and the Mobile Internet
By Howard Rheingold, Thu Apr 07 08:00:00 GMT 2005

Using e-mail rather than SMS as the messaging medium for mobile phones has made mobile Internet services in Japan more successful than in the West, says an industry expert -- a claim supported by recently discovered mathematical properties of networks.


"E-mail was a great enabler of mobile Internet in Japan, and there is a fundamental mathematical reason for this," claimed Ville Saarikoski, a Finn who lived in Japan when i-mode was launched and former head of mobile R&D for Sonera, in a recent e-mail interview. E-mail networks, he noted, have the unique structure of "scale-free" or "small world" networks, while the potential connectivity between nodes of SMS networks are far more highly constrained -- it takes much longer, with many more hops, to travel across networks that do not have scale-free distribution. If he's right, Saarikoski's notion could be powerfully predictive as well as explanatory: future mobile Internet services that enable the formation of scale-free networks could be far more successful than services that don't allow people to grow networks of that form.

Saarikoski credits two business theories, both available as books, with stimulating his discovery: Christensen's The Innovator's Dilemma and Hargadon's How Breakthroughs Happen. Christensen demonstrated how even the best-established companies can fail if they don't pay attention to "disruptive technologies" that could make their products or services obsolete. Hargadon argued that breakthrough innovations are created by bridging distinct social networks, defining technology as people, ideas and objects and underlines the importance of keeping ideas alive.

"I combined the two theories," Saarikoski explained, when I questioned him about the origin of his hypothesis: "I had access to many players in the mobile industry in Finland, but I also had access to some players in the Japanese mobile industry. I bridged the two and conducted interviews in which I explained the innovator's dilemma to the person I was interviewing. We discussed the potential for disruptions in his company, and I shared the other disruptive opportunities I had found in earlier interviews. I came up with eighteen disruptive possibilities within the mobile environment and relating to the mobile Internet. The next step was to make sense of the opportunities. The relevance of e-mail emerged from this material. Comparing e-mail with SMS fits beautifully into Christensen's framework -- e-mail is a disruptive possibility in mobile. And the pattern of scale-free networks emerged from the same interview material."

Scale-free network theory grew out of a 1998 paper by Duncan J. Watts and Steven H. Strogatz that demonstrated how a wide variety of networks in the physical world, from neurons to power grids, are made of a relatively few highly interconnected hubs and a majority of weakly connected nodes; this ratio remains the same no matter how large the network grows, hence, "scale-free." The "small world" title comes from the famous experiment conducted by Stanley Milgram, who asked people to send letters to strangers by sending them to someone who might know the strangers; surprisingly, it took only an average of six hops, giving rise to the notion that the entire human population is separated by only six degrees. Albert-László Barabási refined the model: If a network is scale-free, it is also a small world. Scale-free networks like the Internet are very robust in the face of broad attack -- you can still traverse the network in a small number of hops if a large number of nodes are taken out of action, as long as a sufficient number of hubs remains. In such networks, an innovation at one node -- the World Wide Web protocols, for example -- can very rapidly spread to all the nodes, transforming the functional capabilities of the network itself.

Such networks have such an intriguing set of properties, shared by systems as disparate as languages and ecosystems, that an entire field of study is emerging. That's why Saarikoski's theory is significant, if true: what other properties of small world networks are present in or missing from, or could be magnified or exploited by mobile Internet products and services?

Saarikoski became a student of small world network theory: "Barabasi et al proved that the Internet (physical layer) and the Web are scale-free distributed. Chinese researchers proved that e-mail networks are scale-free distributed, Japanese researchers published showing that the traffic on Japanese mobile networks is scale-free distributed and the French researcher Vespiganani and his team proved that scale-free networks are the most efficient type of complex networks.. The natural question than becomes: is a network of SMS messages scale-free distributed (like a network of e-mail messages is) and if it is scale free-distributed, what is the difference in efficiency between SMS and e-mail as the connecting vector? The first question can be solved mathematically, but would require 10 million SMS messages, the second question cannot to my knowledge be answered with mathematics."

"The answer to both questions can be obtained in a much more simpler way with simple arithmetic," Saarikoski claims: "One can look at barriers and thresholds in the network. For example, sending an SMS message costs 7 eurocents, and if you send the same message to 1000 people, the cost multiplies one thousand fold. In the case of e-mail, one message on a GPRS network is 1 cent, and it does not cost you any more if the message has a thousand recipients. There is another way to look at the difference. E-mail, which has been proven to be scale-free distributed, connects -- or if you prefer -- bridges rather nicely with other scale-free networks, e.g. the Web. In short, two scale-free networks add up to be a more efficient scale-free network.

"Think about Nokia's slogan, 'Connecting People,'" Saarikoski responded when I asked him about which properties of scale-free networks are important in mobile social media: "SMS connects people very inefficiently. Those who design future services would do well to search for more efficient ways of connecting people. Mobile e-mail not only connects people more efficiently, but its integration with the Web is key. A Web site creates a superconnected node, which drives networks toward scale-free properties, but information about the Web site has to spread in order for that to happen. I think the role of people-to-people messaging, the online/mobile equivalent of word-of-mouth, is huge in this regard. Look at Skype -- no advertising budget, but it spreads like wildfire. Trust is important in connecting people with commerce online, and people trust their friends. Again, media like mobile e-mail connect people to their social networks, the source of trusted information, and to the Web, where transactions don't flow without trust. Perhaps the most important property of scale-free networks when you are talking about telecommunications is the property that enables them to evolve: innovations in scale-free networks can spread incredibly fast through the entire network. "

Saarikoski believes the industry is in for a change. "Some people -- particularly representing incumbents -- are quick to argue that this change is all about IP (Internet Protocol) and our networks becoming converted to Internet protocols. Do not be fooled by this argument. IP is a transfer-layer protocol. It is not a messaging medium between people. IP is a human-engineered protocol, which makes digital convergence possible. Scale-free distribution is a law of nature. Mobile operators have been obsessively focused on the agreements between a small number of players in a 'value chain.' They need to think in terms of networks, not chains."