Articles

---

The future belongs to small, connected devices that will wirelessly allow the user -- and the technology -- to self-organize, creating something smart out of many small and simple nodes and connections.

In the last half of the 1980s, I was second-in-command in a traditional, mainframe-oriented IT shop. My boss was an old hand with experience back to punched cards and transistors, who used to say that there were two kinds of technologists: Those who had read Douglas Hofstadter's 777-page G�del, Escher, Bach and those who for some reason hadn't gotten around to it yet. The book, referred to as GEB by initiates, is a gem: A whimsical yet deeply learned romp through all kinds of systems theory, exemplified by Bach's canons, Escher's lithographs, and G�del's incompleteness theorem. I constantly recommend GEB to my students. Every second year or so, one will take me up on it and actually read it. It is, in my view, a sufficient but not necessary condition for being considered an intellectual.

In the book, Hofstadter uses many examples to illustrate self-referencing systems. One of the more intriguing ones is that of an ant society -- an anthill. An anthill is a complex, adaptive system (not that Hofstadter uses those words) -- it can build itself, defend itself against enemies, store resources, provide for the young (and thereby perpetuate itself), repair damage and even move itself (or at least its reproductive capacity). All this is behavior that would seem intelligent to an outside observer. However, if you take it upon yourself to study the movements and actions of individual ants, you will quickly conclude they are, frankly, stupid. Driven by instincts and signaled by smells, ants seem to largely wander around aimlessly, pulling at sticks and getting into arguments with one another and other insects.

So, asks Hofstadter, where do anthills come from -- or, more to the point, where in the anthill is the knowledge of how to build an anthill stored? It is not like there is a CEO of ants, telling everyone else what to do. There is, in fact, no single ant or group of ants that has any idea that they are building something much larger than themselves.

The point here, of course, is that you can create complex systems out of simple parts that only vaguely interact with each other, to create something that is highly organized. These are called self-organizing systems, and the art and science of how to create them with computers is gradually coming into its own. From simulations of traffic conditions and ants (see Resnick, 1994) to Internet applications such as KaZaa and Gnutella, we are seeing massive systems created by actions of non-systematic individuals. Gradually, we are beginning to understand how to build such systems -- partly because we can simulate and study them, but also because, fueled by technological advances, we now have the capability to coordinate complex behavior by using technology to create self-organizing, purposeful systems. The power of this kind of thinking is rather extraordinary -- and alarm bells should go off in a number of large corporations as to its effects.

Now, self-organizing systems are coming into your life with a vengeance, driven by two long-term technology trends: The move to wireless technology, which frees up all the little ants to move around, and the miniaturization of processing technology to the point where you literally can take it with you. The upshot is that we can have computers in everything, connected to everything -- and then we can start to develop applications that take advantage not only of the capability to connect everywhere to everything, but also those that can observe the behavior of all the little ants and draw conclusions from their behavioral patterns. Some observations and examples:

Establishment of networks is increasingly facilitated by self-organizing systems. Self-creating networks are variously called Mesh Networks or Ad Hoc Networks (both from The Economist, June 20, 2002) and Automatic Networks (from MIT's Technology Review, May 2002). Establishing networks this way seems to be hard to do for companies (or, more precisely, hard to establish in a way that allows you to make money on it) but thrives on community engagement.

If I were a telecommunications company executive, I would be nervous at the thought of a network-forming technology that can be implemented by individuals: Adoption is essentially a question of buying the box and turning it on, with the network configures itself. Although the commercial service Ricochet (see www.ricochet.com), whose network is built on self-configuring base stations, has had difficulties, these have to do with the willingness to pay for such products and not the technology in itself. Traveling businessmen will need total coverage -- and can get it in other ways. Home users and small companies, however, may be satisfied with the initially rickety performance of adopt-as-you-go networks. I think these technologies -- epitomized by the increasing coverage of WiFi networks, now even installed for free in New York's Bryant Park (see www.bryantpark.com), will be disruptive to the traditional telecomm providers. After all, the technology fits Clayton Christensen's (1997) three criteria for disruptive technologies: It gives poorer performance than the existing product, the best customers don't want it -- and if the existing suppliers (telecommunications companies) started selling it, they would lose money compared to their existing business. As long as there is room for continued improvement of the technology, it can be disruptive -- and I think it will be.

Connection is increasingly always on and through small devices. Wireless networks are further helped in their rapid adoption by the availability of smaller and more useful devices. Intel is unveiling radio-ready chips, aimed to be building blocks for future wireless devices, especially in embedded systems. AMD is releasing a WiFi chipset that will make it easy for computer manufacturers to included wireless connections in most computers as a matter or course. Increasingly, small devices will not only act as sensors reporting to a central computer, but also be able to communicate directly to other nearby devices. While we are waiting for Bluetooth, Apple has released the Rendevous networking technology, available in the latest version of the OSX operating system, which allows wireless service-discovering connections between devices. Oak Ridge National Laboratory is developing miniature implantable sensors, and the American Chemical Society reports of "pharmacies on a chip" and self-destructing nano-scale sensors. Until we get used to it, future teenagers and other mobile phone and Walkman users will be thought deranged because they are staring into space and talking into thin air -- courtesy of their radio-connected tooth implants ( Electronic Telegraph, June 20, 2002).

Companies will use the wireless technology to redesign their supply chains. Fedex, the Army, Eastman Chemical and Ford are all examples of companies that are or have started using wireless technology to increase the coordination capability of their supply chains, particularly when it comes to replenishment of supporting material -- all the little things that must be there to make the bigger things work. Long term, I think we will see a transition towards parts taking responsibility for their own regeneration -- imaging the gearbox of your Dodge Caravan (which tends to go flaky after about 80K miles) signaling the factory that they better send a new speed sensor.

Information about network use will spawn new services. Speeders in Norway have long been able to subscribe to a service that warns them about police speed controls, delivered via SMS (short message service) to their mobile phones. This service (see www.politiet.no) takes its input from drivers reporting speed controls via their phones as well. More benignly, a system that detects road traffic building up by tracking congestion on mobile phone cells is being tested in the UK and Sweden ( BBC News Online, June 1, 2002). When car traffic slows down, mobile phone traffic goes up, as people start to call colleagues and family to report that they will be late -- again. These applications, as indeed SMS itself, are applications that will thrive once adoption of wireless devices have reached critical mass and the network externalities kick in.

History repeats itself Computing has always gone through generations, from centralized through departmental computers to personal computers, laptops -- and now the PDA/mobile phone. Each of these generations has started with "renegade" adoptions that are self-organized: Minicomputers were sold to departments as "calculators" to escape tight centralized computing budgets focused on mainframes. Personal computers were expensed as calculators by corporate knowledge workers wanting to do their own spreadsheets. Most PDAs and mobile phones were initially bought and used individually, sometimes as fashion items, and later adopted for information processing and communication to mobile workers. For each of these generations, there were planned, initial uses within companies -- which served to develop the technology, which then took off on uses unforeseen by their creators.

The new paradigm of wireless, ubiquitous computing will happen in the same way and for the same reasons -- people want freedom and ease of use, the ability to substitute communication for planning. Any technology or telecommunications company that thinks this adoption will be orderly and the applications known in advance should be worried. The future belongs to small, connected devices that will wirelessly allow the user -- and the technology -- to self-organize, creating something smart out of many small and simple nodes and connections.

There will be lots of anthills. And I am quite looking forward to them.

References:
American Chemical Society (2002). "Beyond patches and pills: the remarkable future of drug delivery", online at http://www.eurekalert.org/pub_releases/2002-11/acs-bpa112502.php. Christensen, C. M. (1997). The Innovator's Dilemma: Why New Technologies Cause Great Firms to Fail. Boston, MA, Harvard Business School Press. Hofstadter, D. M. (1979). G�del, Esher, Bach: An Eternal Golden Braid. New York, Vintage Books. Mitchel Resnick (1994) Turtles, Termites and Traffic Jams: Explorations in Massively Parallel Microworlds. Cambridge, MA, MIT Press. Shankar, V. and T. O'Driscoll (2002). How Wireless Networks Are Reshaping the Supply Chain, Ebusiness Research Centre, Penn State. (online at http://www.ebrc.psu.edu/publications/papers/pdf/2002_09.pdf)