Posted by
Stephen Guerin on
Feb 02, 2004; 2:47pm
URL: http://friam.383.s1.nabble.com/Roger-Critchlow-Applied-Complexity-Lecture-Feb-11-1-30p-tp519125.html
FRIAM Group Applied Complexity Seminar
SPEAKER: Roger Critchlow
TITLE: Why Johnny Can't Negotiate
LOCATION: Feb 11 1:30p Santa Fe Institute Medium Conference Room
http://www.santafe.edu/sfi/events/abstract/164ABSTRACT: Diplomatic and other high stakes negotiations deserve some of
the computational analysis lavished on automated multi-agent systems in
the past few years. A negotiation should be a combinatoric search problem
where negotiators search for the agreement which provides the greatest
benefit to all parties. Positional negotiation strategies reduce a
negotiation to a one-dimensional line search for the least cost
concessions from initial positions. Negotiators, as practitioners of
combinatoric search, should be aware of the computational issues that
apply to search problems, namely the curse of dimensionality, the no free
lunch theorems, and the consequences of bounded rationality. Knowing that
the number of possible agreements in a negotiation might be greater than
the number of seconds in a human life, knowing that there is no guaranteed
better way of searching the possibilities than a random walk, and knowing
that each step in the search will have a finite cost, one might conclude
that we are doomed to failure, that as our disagreements grow in
complexity we are fated to be buried by them.
The key to crafting effective search strategies is to know the lay of the
land. Knowing whether we are searching flatlands, rolling hills, or rugged
badlands makes the difference between success and futility. The topography
of negotiation is determined by the preferences of the negotiators.
Understanding how negotiators determine their preferences in complex
negotiations may allow us to elicit preferences in just enough detail to
find good agreements, to design negotiation strategies which are optimal
for the preferences, and to design implementation mechanisms which are
least likely to be abandoned. And understanding how negotiators do
determine their preferences may also lead us to better ways to determine
preferences, ways which are more efficiently evaluated, more easily
communicated, or more productively exploited to make negotiations work.