markably little about how this process takes place, about what factors make
shared work a pleasant, effective interaction and what factors make it stressful,
inefficient, and ineffective.
Many of the essential properties of effective shared action seem to result from
‘‘accidental’’ side effects of the old-fashioned way of doing things. I put the
wordaccidentalin quotes because I suspect the procedure is not quite as acci-
dental as it might seem, even if it was never consciously designed. That is,
over years of experience, the procedures for performing these tasks have gone
through a process of natural evolution from their original form to their current
shape. Over time, a long sequence of minor changes would occur, each mod-
ifying procedures in a small way. Changes that were effective would be apt to
stay; changes that were detrimental would be apt to die away. This is a process
of natural evolution, and it can lead to remarkably efficient results, even if no-
body is in charge, even if nobody is aware of the process.
It is dangerous to make rapid changes in long-existing procedures, no matter
how inefficient they may seem at first glance. New technologies can clearly
provide improvements over old methods. The old-fashioned control rooms are
indeed old-fashioned. Many of their properties, even the ones people grow
fond of, are accidental by-products of the technology and may even be detri-
mental to the task. New technologies can indeed make life more enjoyable and
productive. The problem is, it isn’t always obvious just which parts are critical
to the social, distributed nature of the task, which are irrelevant or detrimental.
Until we understand these aspects better, it is best to be cautious.
Natural, smooth, efficient interaction should be the goal of all work situa-
tions. Alas, natural interaction is often invisible, unnoticed interaction: We
don’t know it is there until we remove it, and then it may be too late. We do
know that communication is important, however. Listening to the chatter of
air traffic controllers turns out to keep pilots informed about all the other air-
planes along the route: Replacing this chatter with computer messages sent
only to the relevant airplanes destroys this critical aspect of situation aware-
ness, even while giving the benefit of more accurate, less confusing mes-
sages. In a similar fashion, replacing the office clerk who delivers mail from
department to department with a computer-controlled robot also destroys one
channel of communication among departments. Automating factory control or
forms processing can also hinder the informal communication processes among
workers that allow productive, unofficial decision paths to develop within a
company.
The human side of work activities is what keeps many organizations running
smoothly, patching over the continual glitches and faults of the system. Alas,
those inevitable glitches and faults are usually undocumented, unknown. As a
result, the importance of the human informal communication channels is either
unknown, unappreciated, or sometimes even derided as an inefficient and
obstructive, non–job-related activity.
Eventually, the natural process of evolution will work even upon the latest of
technologies. The problem is that in the meantime, if we are too precipitous in
making change solely because it is possible, we are apt to run into difficulties.
When these difficulties occur in commercial aircraft or large industrial plants,
the results can be tragic.
Distributed Cognition 447