🚀 go-pugleaf

Summary: I believe that we should aim, in designing groupware, to assist
  communication between people, not just for 'better, faster decisions'.

I do not hold with the idea that computers support decision making
by storing all the 'pertinent' topics for consideration by people.
Phrases like "achieving consensus by considering all pertinent
points of view" imply to me a nice rational, objective,
decision making process.  I dont believe that such things can exist.
I also dont believe that they should.

Consider the work of Terry Windograd and Fernandes Flores in their
book 'Understanding Computers and Cognition'.  In this work, Winograd
and Flores argue that there is no objective world which can be measured.
Rather, that the world is created through the language which we use to
communicate.

When people examine the world, they do so through a filter of
their own experience.  We find simplifications of the world in
our language, it is easy to believe that these simplifications
are real.  Take for example the notions of truth and falsehood.
It is easy to believe that a statement is either true or false.
They are not, of course; consider self-reference: "This statement is false"
or nonsense: "The grunky bling on the pimple grog".  How can
we measure the validity of statements like this?  When we try to
cast them into inappropriate moulds it is not surprising that we have
problems.  It is easy on reflection to beware of statements like the
self referential one above, but when we stumble on it in other
contexts we find it very difficult to step back and see it for what
it really is.

Simple statements like: "Ice exists naturally" seem to be quite clear
yet they stand in a background which controls their interpretation.
Try that statement on eskimos, who have no single word meaning
'frozen water'.  Try it on people who work with large amounts of
'dry ice' and who may have a different default interpretation
of what 'ice' is.

Many democratic systems use a questionnaire or a set of weights which
a number of people respond to.  These systems ignore the background in
which the questions were written.  We will see claims that getting
people to fill out these questionnaires does tend to achieve consensus.
Consensus is important not because it shows that people agree but
because it makes people believe they agree.  I would argue that the
only reason that a consensus is found is because people responding are
willing to shift their viewpoint to that dictated by the stream of
questions.  In democratic countries, people are often quite happy to
believe that the ability to vote for their leader every few years gives
them freedom and control. That's good; it keeps the people happy.

We should be very careful when building systems which 'support
group decision making' because many of these systems may hide
their biases too well.  Just as it is hard to step back from the
statement: "This statement is false" it is incredibly difficult to
step back from a questionnaire and see what is happening to your
opinions.  How do we argue with software which, after much crunching and
calculating, says: "Well you all agree, based on the results of the
questionnaire that we will do X."  If X is 'sell, sell, sell' and we
are in the share market and everyone uses the same program the effect
could be interesting.  Of course, that would never happen.

Martin Leadbeater            E-mail: Martin.Leadbeater@edgar.sait.edu.au
Academic Computing Service
SA Institute of Technology                        Phone:   +61 8 3433435
The Levels, SA, 5095, Australia                   Fax:     +61 8 3496939