separation is probably needed both in message labelling and time. in
other words, it's not enough if each message is clearly labeled according
to its purpose (though this would help immensely); it's also necessary
to discourage indefinite discussion in any of these phases, and to have
clear transitions from one phase to another.
This has been tried. Terry Winograd and Fernando Flores set up a company
that tried to apply the language-action model to organizational messaging,
using categories that split the messages into the "action" associated
with the message. Many of their categories match the ones you have above.
One of my old college roommates, Bud Vieira, did some work on the
system with Flores and discovered that when users self-labelled their
messages, almost everything got labelled as a comment, whatever
category language-action theory might have assigned.
It's interesting that authors don't want to think as carefully about
the labelling for a message as the message itself.
What I had in mind for a prototype was:
- a user interface via which messages are submitted, which reminds the
user what phase the discussion is currently in prior to the message
being submitted, and which insists that the user specify the phase
of the message being submitted. ("comment" is not a phase)
it would accept messages that are out-of-phase but the presentation
of messages to readers would be separated by phase. presumably
the current phase of the discussion would get the most attention.
- a means by which a moderator/chair could reclassify already posted
messages or threads as out-of-phase. for visibility, the message
would be tagged as "reclassified".
All of this would be experimental, of course, so we shouldn't
expect any of this to work on the first try. But we're going
to need to allow WGs to do some experimentation if we want to
find better ways to do our work over the net.
Keith