Can self-evolving systems gather implicit knowledge?

If readers show sufficient interest, a publisher may subsequently release a bound, hardcopy of that content, but the principal, default format of published content will be electronic. Printed material will be a convenient and attractive supplement to online material, but nearly every publication will be fully accessible through broadband Internet.

Readers will use software to cull through immense collections of published material and assemble pieces from many sources. Content will be finely divided and fully exposed to external indexes. The indexes will allow the content to be repackaged according to the needs of individual readers. Each reader will use a digital container to collect snippets of content that have been gleaned and continually updated specifically for him or her.

Most knowledge refuses to wear a name tag, and easily slips past even the finest index. Self-evolving systems can someday inhabit cavernous digital networks and search out patterns that are too fleeting to be described.

This change in the publishing industry will affect education. Most of the knowledge involved in making judgments or performing complex skills is implicit knowledge, or knowledge that is too diffuse or somatic to be categorized or to be represented symbolically. The cognitive abilities brought to bear on solving a math problem, for example, are only dimly visible to the person solving the problem or to a teacher attempting to transfer the relevant mathematical skills. Much of the labor of education is due to the daunting challenge of transmitting information that cannot be directly referenced by means of symbols.

Because implicit knowledge is too elusive to be mapped symbolically, it also resists indexing. Unfortunately, what that means is that most of the knowledge crucial for the development of human culture is too "low level" to be indexed. This is arguably one of the most important problems of our time, because as the cultural commonwealth becomes more complex through developments in science and technology, relatively fewer people are able to participate in that commonwealth.

At some point, the global digital network will have accumulated so much data that it will itself become as seemingly complex and inscrutable as the physical world reflected by the data. Efforts by software engineers to introduce robots and intelligent search agents will only make the challenge greater. Such technologies will allow the digital network to grow even more explosively, in the same way that freeways eventually led to the creation of suburbs, more cars, and much more traffic than what the freeways were originally designed to alleviate.

Could self-evolving systems, embodied in the digital space of the global network, be used to accumulate implicit knowledge about the network? If the network has itself become a purely dynamic, stochastic system, could systems be cultivated that would interact with that network and evolve useful structures at the interface between human activity and network activity?

Michael Webb, 2000

home ]