Chapter 1: Introduction
Intelligence Amplification
Overview of Book

Intelligence Amplification

The view of collective intelligence as a form of behavior made possible by some form of mediating computer system places it within the general tradition of intelligence amplification (IA). This perspective takes the position that computer systems can be developed that partially mirror human mental functions; thus, by increasing the capacity or speed of operation of those functions, these systems can thereby increase or amplify the mental capacity of the human user working with them. As a result, quantitative increases in specific functions may produce qualitative differences in intellectual behavior, making the computer a necessary but not sufficient tool for enabling this mode of thinking.

Vannevar Bush (1945) is generally credited with originating the idea of intelligence amplification. Writing before the first commercial computers were developed, Bush described a hypothetical desk-like device he called the memex that would be implemented using microfilm technology. It would permit a human user to store vast quantities of data, add new information, but, most important, add cross-references at the bottom of any microfilm page that could be instantly followed to some other page. Thus, the human user could construct large networks of semantic relationships within the memex, drawing together vast quantities of data and then quickly and associatively move from one intellectual context to another. One could argue that the book - or at least a library of books - could similarly extend the capacity and precision of human long-term memory and that books do, in fact, include similar cross-references. BushÕs innovation lay in the speed with which associative links could be followed to access new material - a second or two versus the minutes or even hours required to move from one printed volume to another.

It makes sense to talk about BushÕs memex as an amplifying device in the following sense. He identified several key architectural features of human intelligence - long-term memory, semantic relationships, and associative access - and then provided within his memex - at least in theory - their operational counterparts, but with greater capacity (the microfilm store with its embedded semantic relationships) and comparable speed (associative access). Thus, Bush believed his device could amplify a specific set of human mental functions. No one has yet built a complete memex as Bush described the device. However, using more familiar computer technology, Doug Engelbart was the first to build a memex-like system (Engelbart, Watson, & Norton, 1973). In recognition of the goal to supplement human intelligence, Engelbart called one version of his system Augment. Today, many of the features first described by Bush and first built by Engelbart are routinely found in contemporary hypertext systems, some of which are discussed in chapter 3.

Just as IA systems make possible a type of mental behavior that would not be possible without them, so, I suggest, a particular type of collaboration support system may enable a form of collective mental behavior that would not be possible without it. These systems, I suspect, will be based on principles analogous to those for IA systems, but with important distinctions and extensions. We normally form collaborative groups for two reasons. First, the task is too large and/or there is not enough time for it to be done by one person. Second, no individual possesses all of the skills and/or knowledge required. However, when we (necessarily) assemble a group to overcome these problems, we inherently create other problems. Because the intellectual construct being developed by the group is likely to be too large to be known in its entirety by any one individual, it may lack intellectual integrity. Rather than being a structure that is deeply principled and elegantly simple - as we expect of the work of our best individual minds - it may emerge as an awkward assembly of incongruous pieces. Indeed, we have come to expect this of groups, as indicated by the old joke that a camel is a horse produced by a committee.

A computer system that can help groups approximate a CI will have to include, as a minimum, functions that help them perceive and address the overall structure and integrity of their work. It must include tools to help groups establish and maintain the internal consistency and coherence among the various information products they produce through the individual hands of their various members. Thus, it will have to amplify intellectual skills that are (relatively) strong in individuals but less so within groups. It may also include additional tools to facilitate access and version control, communication, joint work, and other group behaviors. But it cannot neglect the more basic requirements of intellectual integrity, coherence, and consistency.

Go to: previous section - next section - table of contents