I have spent some time with the integrated text of discussion between
Jay and Victor. I hope others have done the same. Steve did us all a real
favor with his work. In fact, I would like to know how to get from pdf
to an editable file because he provided a next step integration that was
a great tool of thought.
I am having some difficulty with the abstractness of the discussion and
suspect I am not alone. I want to provide my interpretation of a concrete
instantiation of a point made by Victor which begins.... "Even simple experiments with a small collectivity of ....agents show that ...... while convergence
is possible it is fragile (i am simplifying here for lack of time, but the
context is retrievable.
Here is my candidate concrete instantiation
In the 1970's and '80s our research group spent a LOT of time videotaping
interactions and then coding them according to various motivated coding
schemes (speech act theory was one, but not the ohly one). We discovered
what Victor describes in a manner that I belive is still relevant to
xmca members.
We would have two or members of our group code a series of interactions and
put them in categories, and within a few (very time consuming!) sessions,
the coders would become to high levels of agreement. But then one of the
coders would move on to another job, or get sick, or get reassigned for
some reason to some other activity. And, lo and behold, when a new coder
was paired with an old coder, or two new coders tried to replicate, reliability
went to hell in a handbasket.
Our interpreation was (mine is) that the initial coders formed an idioculture
which allowed them to anticipate their joint interpretations, but its was a
culture of two and very fragile. The codes did not cut interactional
nature at its joint, but hacked a usable concensus.
Hence, in reading the discussion between victor and jay, that part of it
is not abstract/empty for me. It is a serious methodological caution based
upon experienced reality.
I do not believe this is a unique experience. Anyone else experienced this
problem in rendering descriptions as "real" and not social conventions?
mike
This archive was generated by hypermail 2b29 : Fri Aug 01 2003 - 01:00:08 PDT