Towards the Web-as-Brain Metaphor

Metaphors are particularly useful as models, because we use them to borrow ideas from the familiar in order to understand the unfamiliar. Our models of the Internet affect how we use it, interact with it and with others using the Net. The dominant model for the Internet today uses the metaphor of space, as demonstrated by such expressions as the "information superhighway" and "cyberspace". But it affects more than the way we talk about it - it is also reflected in the way we think about the Web and use it. For example, retrieving information on the Web has become a matter of finding its "location" and "visiting" the site.

However, the static metaphor of space has outlived its usefulness. One case where the metaphor fails, or will soon fail, is in the area of information retrieval. Search engines, such as Alta Vista and Lycos use robots, or spiders, to regularly visit all the pages they know about and compile a record of their content, a so-called index of information. The user simply has to visit a central location where the results of the robots' travels are kept, in order to locate sites containing information of interest to the user. However, the Web is growing at a phenomenal rate, faster than the computing power and storage, and certainly much faster than the bandwidth, which governs how fast the spiders can build a central map of cyberspace.

No conventional spider can hope to keep up with the growth of the Web, as well as keeping the record of already indexed pages up to date. Already, the amount by which same query searches on different engines overlap is a small fraction of the results. The recently deployed Hotbot is a step in the right direction, since it uses collective power of many computers to carry out the indexing and the search.

One answer to the growth of the Web and the indexing problem is to increase the number of search engines - if one spider can't keep up with the growth of the entire Web, perhaps it can manage a portion of it. The implication here is that we should be able to use the Web as its own search engine. Every web server will index the content it serves, in addition to making the topical hot lists of the users connected to the server available (there is a good chance that the users are interested in the topics their machines are serving and have already found relevant links). Perhaps, each time changes in content are made, the web server reports them to some DNS-like service, which can then refer search queries to their proper destinations.

This thought experiment above suggests a new model for the Internet, one based on the dynamic self organization and adaptive characteristics of a complex system, such as a living organism. Here is some support that the "organism" metaphor correctly captures the more interesting aspects of the Internet. The evidence is organized around the central characteristics that define an organism.

• Adaptability - The Internet has some adaptability built into its infrastructure. The TCP/IP protocol attempts to efficiently route data packets by taking advantage of unused bandwidth. Most spiders that index the Web are designed to roam when Internet traffic falls. Similarly, off-line Web browsers take advantage of low off-hours traffic and phone rates. In a not too distant future, millions of intelligent agent programs will perform highly specialized tasks on the Net, from information gathering and management to routine maintenance. These agents will take adaptability to a qualitatively different level as they learn to perform their tasks more efficiently, fulfilling and perhaps anticipating the preferences of users in the changing environment.
• Self regulation - A living organism usually has many ways to maintain homeostasis (equilibrium). These mechanisms bring the system back into the comfort zone when disturbed by environmental causes. On the Internet it is the expensive resources, such as band width, user time and attention, that the system regulates. Though it is easy to find abusers of these costly resources, the Internet makes it easy to fight back and punish abusers, for example, by sending spam mail back to originator and overloading their server machine. This important quality is tied to the very nature of the Internet survival, and it is near to the hearts of its netizens. Some of the most colorful Net folklore comes from the tales of abuse and retribution.
• Growth - This must be the least controversial aspects of the Internet. The number of computers connected to the Internet has grown explosively since 1991, and there is still room for more growth.
• Survival - For years predictions about the end of the Internet were made frequently, that as grows it will collapses under the weight of its popularity. However, the Internet has sustained an explosive growth, without a significant improvement in the infrastructure.
• Immune system - Though there have not been any reported cases of virus infection spread on the Net, the fears of Internet (email, applet) virii never subside. In fact, the true virii are the jokes, such as the "Good Times virus" hoax, which is still making rounds on the Net. The propagation of hoaxes on the Internet (through news groups, mailing lists, email) is probably very similar to the spread of disease in a body and epidemics in a population. The analogy of the Net to biological immune systems was not lost on IBM, which recently announced its new virus detection program. Computers participating in the program ship a suspicious piece of code representing potential viruses to IBM's lab, where human researchers analyze it and devise a fix, if necessary, which is then shipped to every computer that is part of the Immune System network. (Unfortunately, in a case of mixed metaphors, they call their system an Immune System for Cyberspace.)

12/96

revised 12/97

kL