Basement
:U:!'lV-mUirt'ii»*)rt'.!iII
"X
V.
^
r„ r?_c.
OS.;
!K"T
rT*
JMAR 16 19S8
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
THE ELEMENTS OF NEGOTIATION
MANAGEMENT
J.
WP
1956-87
D.
Nyhan & Dhanesh
K. Samarasan*
November, 1987
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
THE ELEMENTS OF NEGOTIATION
MANAGEMENT
WP
Nyhart
D.
J.
&
Dhanesh K. Samarasan*
November, 1987
1956-87
•The authors
respectively.
are,
Engineering, Sloan School of
Professor of
Management and Ocean
Management and Department
of
Ocean
Engineering, and doctoral candidate. Department of Ocean Engineering, MIT.
The work described
at the
MIT
Project
in this
paper
on Modeling
The following provided support
is
for
part of a multi-disciplinary effort
underway
Negotiation Management.
for this
work and
are gratefully acknowledged;
School of Engineering, MIT; Project Athena, MIT; National Institute for
Dispute Resolution; Program on the Processes of International Negotiation
the American
Academy
of Arts
at
and Sciences under a grant from the Unisys
Corporation.
This working paper
is
do not copy without
one
in a series
of manuscripts in their draft form. Please
the permission of the authors.
and should be directed
to the attention of the authors.
Comments
are
welcome
rlARIES
CONTENTS
I.
IL
The Opportunity and The Need
The Elements
of Negotiation
1
Management
UL Models and Modeling For Negotiation Management
The Simulation
IV.
of the
Problem
6
8
Dealing with Risk and Uncertainty
Selecting Norms and Procedures
The Computer as a Third-Party Neutral
11
The Model
14
in the
Middle
9
12
Collaborative Model-Building
Zooming and Bridging
14
Other Issues
16
Understanding the Impacts of Models and Modeling
Enhancing
S)Tithetical Skills
and Qualities
15
17
17
The Relationship
19
A
21
to Integrative Bargaining
Broader Definition of Negotiation
The Use and Non-Use
of
Models
V. Potential Applications in Substantive Contexts
VI.
4
26
29
Dissemination
30
References
32
THE OPPORTUNITY AND THE NEED
I.
Negotiating, or treating with others in order to reach an agreement,
is
ubiquitous. Everyone negotiates. Bankers engage in arbitrage over percentage
points. Engineers reluctantly
up
plea-bargain to speed
and outside the
firm.
make design
And managers
a trial.
The verb
trade-offs with clients. Attorneys
negotiate daily, both inside
comes from the Latin
"to negotiate"
negotiari,
meaning
"to transact business," underlining the universal character of the
process.
Governments, public
individuals,
and
just
special
officials,
interest
about anyone with any business
groups, private
to transact, all negotiate
regularly.
As
and resolving disputes
differences
&:
widened favor
the idea of negotiation gains
Verma, 1983; Reich,
1987], a
generation of ideas,
needed. This paper describes work being undertaken
Modeling
put the
to
skills
at the
Management to develop these
power of the computer directly
for Negotiation
and
tools,
means
managing
of
example: Lindblom, 1965; Kochan
[see, for
new
as a
and
MIT
tools
Project
ideas, skills,
the
in
is
on
and
hands
of
negotiators and their third-party facilitators.
There are many contemporary problems where the concept of
management
negotiation
involve
the
as elaborated here can play a major role.
implementation
of
engineering
large
systems,
Some
and
are
characterized by technical complexity, a diffusion of decision-making
authority,
and the recognized need
major public issues, share
many
heated by policy considerations.
for a quick settlement.
of the
Still
same
Others involve
characteristics,
and are often
others require the efficient and uniform
handling of large numbers of similar conflicts or the effective use of the
negotiating experiences of others. Unfortunately, not a great deal of
been done
to alleviate the difficulty of dealing
work has
with negotiation in a rapidly
changing and technologically oriented world.
In
Even
a
many
negotiations, the interactions
meager understanding
of arcane
and
specialist.
When
technical
of the issues
knowledge
is
that is
decisions have to be
made
between parties are complex.
contingent on acquiring a depth
beyond the reach
of
all
in these cases, parties
but the
nominate
and depend on
on which
technical experts
who
require a wide variety of information
such information
to base their calculations. Often,
and harder
A
to collect.
greater part of the decision-making
when
all
have
parties to the negotiation
to define,
must then deal
And even
with the management of risk and uncertainty.
circumstances,
hard
is
in
the best of
sufficient data
and
confidence in that data to come to an agreement. Nature does not always
cooperate or conform to expectations: agreements that are politically feasible
often
With the large problems
technical reasons.
fail for
that are today faced
almost routinely, the consequences of a negotiation that
When
when
be disastrous.
to occur,
as
technology, or
way
can
and widespread damage occurs or threatens
irreparable
an arms control agreement
when
fails in this
is
undermined by new
the construction of a nuclear plant has to be delayed
because emergency evacuation plans are unacceptable, further negotiation
is
generated.
Another
particular
end
difficulty
when
arises
conflict
prolonged, with no
is
in sight. Traditionally, violence or the threat of violence has
caused the more threatened side to
conflict.
a
feel a
desperate longing for an end to the
But nowadays, even in purely commercial disputes no protagonist
has the time or the resources for an extended
stint in the trenches.
change already endemic and ever accelerating, settlements
more value
quickly have
stable,
if
not friendly, relations
the courts, the traditional
with
to the parties.
litigants,
high cost of
litigation,
often, the
also important to
is
means
And
have given
movement.
Still,
maintained
when
be reached
prompt resumption
all
sides. Unfortunately,
as
more
are granted access.
and the sometimes higher cost of waiting around
a shot in the
there remains
a dispute
is
of
of resolution in such cases, are jam-packed
and the queues are lengthening
litigation,
that can
With
a
arm
The
for
to the alternate dispute resolution
deep concern
that
due process be
taken out of court to be settled. Such concern
underscores the need for a consistent and principled framework for the
ordering of negotiations.
The problems
authority to
'final'
whom
are only exacerbated
when
the conflicting parties defer.
there
is
no acknowledged
The co-existence of multiple
decision-makers tends to slow a negotiation
down
until
it is
perceived
as being less than efficient. Yet even with this multiplicity of participants
and
decision-makers, frequently not
agreement are effectively
between equity
at the
parties
all
who
will be affected
by an
negotiating table. The perennial tensions
and effectiveness
in participation, efficiency in execution,
resolution clearly present a difficult problem, but one that
in
must be addressed
nevertheless.
In short, these three features - technical complexity, a
need
for quick
and careful disposition, and the absence of unquestioned decision-making
authority - are
common
to
new approach toward
for a
many
of today's conflicts,
and emphasize the need
negotiation. But even in simpler cases,
parties seek a quick resolution of their differences,
new
ideas
and
where two
tools
would
be welcome.
The framework we
the
management
are developing
of negotiations.
is
and with
synthesis
is
a
clear vision of
supplemented by a
how
ii)
be a systematic guide
these ideas
and
provide
to
management with new
to
ideas and
be used. The
tools can
collection of software, the different parts of
which share the following primary
i)
to
The investigators intend
practitioners in the field of negotiation
tools,
meant
functions:
the simulation of the problem through joint model-building;
the calculation of risk
and uncertainty;
selection of decision-making procedures
iii)
the
iv)
the optimization of settlements
by
and norms; and
a third party
computer.
Starting out with these primary functions for the computer, the concept of
negotiation
management
is
aimed
constructs that relate to negotiation.
at the synthesis of
various theoretical
The investigators have developed out of
these constructs a set of working hypotheses about the role of models and
modeling
in negotiation
and
conflict
management. This paper
be a reasoned argument in support of that
set of
hypotheses;
is
it is
research findings or the analysis of a newly created database.
much
of the thinking underlying the efforts
Modeling
for
Negotiation
experimental intent.
underway
Management and
is
a
at the
first
designed to
not based on
It
MIT
comprises
Project
on
statement
of
II.
THE ELEMENTS OF NEGOTIATION MANAGEMENT
Negotiation management, as described in this paper, has several
essential elements:
i)
the use of computers, modeling,
as
an integral part of negotiation and other forms of dispute
resolution and conflict
ii)
and communication technology
the joint use of
management;
models and modeling by negotiating
parties,
by
third-party neutrals, and by students of negotiation;
iii)
the use of
models and modeling
to serve better the introduction
of integrative bargaining or joint problem-solving in negotiation;
iv)
the combination of a synthetical process
based analytical tools
v)
to
develop
agreement on the
of the agreement
viii)
a coherent
outcome;
the use of these tools in different stages of dispute
from the construction of
vi)
and various computer-
a
consensual database, through reaching
critical issues, to
and
management,
settling
monitoring implementation
subsequent disagreements;
the application of these ideas in a variety of substantive contexts;
understanding through research the impacts on negotiation of
the use of models
and modeling, thereby contributing
to
both
theory and practice; and
ix)
the dissemination of the
teaching,
knowledge so gained through writing,
and the development of teaching materials.
The further development
of these elements will incorporate current
theory and empirical knowledge from several
fields.
The underlying ideas
from dispute resolution include the theoretical concepts of synthetic and
integrative thinking, a differences orientation towards negotiation,
maximization of
joint utility
and the
through collaborative problem-solving
and
integrative bargaining. Similarly, from the modeling sciences, the effort
draws upon systems dynamics, game theory, optimization,
retrospective pattern analysis,
Figure
1
and various ideas from
summarizes the elements
statistical
and
artificial intelligence.
of negotiation
management
graphic form. The remainder of this paper elaborates on these elements.
in
Figure 1 - The Elements of Negotiation Managen^ent
Project on Modeling for Negotiation
Management, MIT S 1957
MODELS AND MODELING FOR
III.
NEGOTIATION MANAGEMENT
Although
negotiation
is
largely
is
it
not a
new
(Nyhart) led an effort
at
unfamiliar, the idea of using
models
in
one. In the late 1970's, for example, one of us
MIT through which
a
computer simulation
of the
technology and economics of deep-ocean mining helped to bring consensus
in the negotiations of the
Third United Nations Convention on the
the Sea [Nyhart, 1984; Straus
et ah,
Law
1984; Antrim, 1985]. Since then, there
of
have
been several discussions and examples of the use of computer methods
in
negotiation [Sebenius, 1981; Kraemer, 1985; Nyhart and Goeltner, 1987]. These
methods can be divided
(ii)
and
simulation models;
(v)
into five major categories:
(iii)
(i)
pattern-seeking models;
agreement-seeking models;
(iv)
syntheses of the above. Although this categorization
comprehensive or
free
from overlap,
does serve
it
potential for
computer applications
management,
as laid out in this paper,
in this area.
is
to point
teaching models;
is
not necessarily
out the breadth of
The concept
of negotiation
not meant to be a substitute for the
use of these and other existing paradigms. Instead, the approach taken
is
to
build on what has gone before.
Simultaneously adaptive and developmental, the work
Project
combines
in
an integrated package of software as
and methods as are necessary
for the
management and
many
at the
MIT
different tools
resolution of a
wide
range of disputes and controversies. The different parts of the software share
the following primary functions:
i)
ii)
the simulation of the problem through joint model-building;
the calculation of risk
and uncertainty;
selection of decision-making procedures
iii)
the
iv)
the optimization of settlements
Each of these components
is
by
and norms; and
a third-party
computer.
described in detail below. There are, in addition,
three other functions being explored: the provision of expert advice to a
negotiator, the analysis of large
numbers
of similar
problem
cases,
and the
facilitation of the overall process of negotiation. In this paper,
however, only
the four primary functions are discussed at any great length.
In the description of each of these four functions, the roles
models and the process of modeling can play
facilitators are illustrated.
in Section
which
in aiding negotiators
Although these roles are discussed
and
in greater detail
IV below. Figure 2 summarizes them here as a preview.
Negotiators and facilitators can use models and modeling to help
them:
1
Establish a consensual database as a foundation for agreement;
2
Assess the impact of uncertainty and change in the database;
3
Set a
4
Probe inequalities of power, influence, and access
5
Track the formation of coalitions and
6
Formulate
7
Practice cross-cultural negotiating styles
8
Provide communication links for bargaining and discussion;
9
Communicate
common
research agenda to collect useful additional data;
new
to data;
clusters;
procedures for making collective decisions;
and
tactics;
effectively while protecting secret information;
among
10
Discover the areas of potential agreement
11
Suggest settlements that minimize the need for compromise;
12
Evaluate the substantive impact of proposed settlements;
13
Enhance
14
Monitor the implementation of the
15
Reach an amicable solution should disputes
initial
sub-groups;
agreements through Pareto-optimization;
final
agreement; and
arise
anew.
Figure 2 - Models and Modeling in Negotiation Management:
A Summary
The Simulation
An
Problem
important part of negotiation management
methods
to
interest.
To
is
the use of
computer
simulate the relationships between substantive variables of
date, the v^^ork has focused
dynamics. Speaking
now
on the application of systems
more complex
in general terms, the
system, the less intuitive will be
shown
of the
the substantive
behavior. Forrester and others have
its
be discovered in a group setting
that this failure of intuition can
through the process of jointly building a simulation model, and including
within
it
and structure of the group
the policy goals
example].
The usefulness of
group can take an active
participation
the process
is
enhanced
modeling
role in the
and surfacing of assumptions
deepened understanding of the substantive
[see Forrester, 1969, for
if
the
effort [Senge,
members of the
1985]. The more
that take place, the
more
realities. In addition,
there
is
a
the personal
and active involvement of the participants reinforces
their
ownership of the process. Therefore, a very high premium
placed on easy-
access simulation in the context of negotiation
is
sense of
management: those with the
authority to negotiate most frequently do not have the time to master very
complex modeling techniques.
is
In
any
case, a transparent
technology /interface
preferable so that the parties can concentrate on the conceptual issues in
controversy, rather than on the intricacies of modeling.
As long
features of
as the stipulation of easy access
many
tools of systems
their interests
real
is
observed, the essential
systems can be clearly and effectively modeled with the
dynamics
in
such a
way
that negotiating parties can identify
with the various attributes of the model. Kreutzer, for instance,
has used such a modeling process as an eye-opener in the context of arms
races [Kreutzer, 1985].
model-building
when
making decisions
No
a
doubt, there
group
is
is
a subtle shift in this process of joint
negotiating, as
or policy choices.
To
opposed
to
a lesser extent, the
when
same
it is
shift
jointly
can be
detected between intra- and inter-organizational negotiations. In the language
of negotiation, as explained in a later section, the shift relates to the difference
between integrative and distributive bargaining.
Irrespective
of
the
extent
integratively or distributively,
to
when
which negotiators are bargaining
negotiations focus on convoluted
technical
issues,
or are complicated by a need
relationships between parties, the joint
and
It
And
it
at least
allows the parties to focus on
on
inter-relationships rather than
unrealistic private assumptions.
preserve long-term
development of a model holds
three important advantages [Samarasan, 1987].
the issues
to
fosters a
their positions.
It
exposes
good working relationship
By asking what-if questions, by performing
comprehensive sensitivity analyses, and by conjuring up relevant scenarios
and pursuing them through simulation, negotiators can enhance
among
negotiators.
the
considerably
the
of
level
substantive questions.
We
understanding with which they approach
are designing a
Meta-Modeling - which
new
process - one
will create a negotiating
we have
environment
in
called
which
collaboration through the use of the computer will be not only possible but
natural [Samarasan, 1987].
Dealing with Risk and Uncertainty
A
different substantive issue that often arises in a negotiation
degree of certainty with which information
many
cases,
there
is
is
is
the
held by the various parties. In
incomplete information about variables that are
deterministic in nature. In other cases, the variables themselves are only
probabilistically defined.
concession
is
When
the eventual
outcome of
not immediately obvious, the
a particular gain or
management
of risk
and
uncertainty becomes an important part of the negotiator's role. To this end,
software for decision tree analysis
is
included as one of the tools available for
negotiation management.
Since decision tree analysis
is
well-known, only a brief explanation
presented here in the context of negotiation management.
A
decision tree
is
is
a
graphical representation of the various options open to decision-makers, and
them by chance. By taking into
associated with each option, and the
the various outcomes that might befall
account the expected rewards and costs
multiple probabilities or probability distributions associated with each
random
variable, the
method allows
a negotiator to calculate the expected
value of each possible combination of options and outcomes [Raker, 1986].
Generally speaking, a tree flows from
issues
under discussion
left to right,
in logical or chronological order.
representing the
Nodes
in the tree
can be either decision nodes or chance nodes. From decision nodes emanate
specific options that are available to the negotiators.
From chance nodes,
branches emerge that represent specific alternative states of nature not v^ithin
the control of the negotiators. Each terminal
the combination of
all
the root to that node.
calculated through
in the tree
The
probabilities that are used v^ith a given
node
node are
and marginal
probabilities, according
The optimal
set of decisions is then
in the tree.
backward induction.
Especially useful for those involved in a negotiation
ability to
corresponds to
decisions and events encountered along the path from
selected from the set of prior, posterior
to the position of that
node
is
the subsequent
analyze quickly the sensitivity of the optimal set of decisions to
changes in assumptions about probability and expected value. Bargaining
positions that are favorable under
some assumptions may
irrelevanr or even have a negative effect
and
their constituents
positions that might
on the
satisfaction of the negotiators
under other assumptions. Information,
seem
crucial at first
may
actually be
interests, or
be found to be dominated by
other information, interests, or positions. Decision tree analysis, therefore,
can be used in negotiation management to scope the discussion and to focus
on the
details that really matter.
The use
not
it
new
of decision trees
may be new
to decision analysis. Raiffa, for
case study [1982]
whether or not
how
the
to negotiation,
example,
illustrates
it is
research described in this paper
is
is
A
in deciding
The premise underlying
that the joint use of the
and
the
method by those
a special case of
its
use
decision tree that includes explicitly
all
the
a valuable extension,
by any single decision-maker.
obviously
with a detailed
method might be used by an attorney
to settle litigation out of court.
involved in a negotiation
but
options available to one party in a negotiation will also include those options
that are available to that
one party
in cooperation
with the others. Later
discussion of the relationship between integrative bargaining and the joint
use of models in negotiation management elaborates on this point.
10
Selecting
Norms and Procedures
The discussion now leaves the substance
questions of process.
Two
of the four
with process. The
Mechanics (SAM), and
is
first
model,
SAM
different
is
Management deal
Simulated Agreement
one
called
is
model
of the negotiation process
no single 'negotiation process'
clearly
for negotiation
and group
interaction. Insofar as
theory deals with the logic of multi-lateral decision making,
theoretic simulation of the
game
equilibrium,
to
actually implements a library of algorithms that represent
ground rules
insofar as
to
for Negotiation
a descriptive
[Samarasan, 1987]. Since there
and turns
components of the software being
developed by the Project on Modeling
explicitly
of negotiation
dynamics of a negotiation.
On
SAM
a
is
game
game-
the other hand,
theory has traditionally dealt with a mathematical search for
SAM
is
an outcast
to this tradition.
As many observers have pointed
out, there
is
great difficulty in
predicting analytically and narrowly the outcome of a negotiation [see, for
example. Lax
surmount the
&
Sebenius, 1983]. In a sense, both Myerson and Axelrod
difficulty elegantly
and extract some interesting conclusions by
modeling only very simple or highly abstracted interactions [Myerson, 1979;
Axelrod, 1984]. But in general, there are simply too
many ways in which they can vary.
SAM, therefore, does not function
outcomes.
a
When
as
decision-making authority
many
and too
variables,
an analytical predictor of stable
is
shared
among
the
members
of
group, whether they be managers distributing assignments, contract
administrators negotiating compensation, or Senators seeking consensus, the
types of agreements that emerge are defined qualitatively in large part by the
ground
rules, or
these group
facilitators
norms
of fairness, that are in effect [Gulliver, 1979].
norms can be coded
can then use
to formulate
SAM
new procedures
in
SAM
Many
of
algorithms. Negotiators and
to investigate different conceptions of fairness,
for
making
collective decisions,
and
to track the
formation and disbanding of coalitions and clusters in the course of a
negotiation under different conditions [Samarasan, 1987]. The conditions
which can be varied
in the simulation include,
distribution of influence
among
among
other things, the
the parties, the control of the agenda,
11
and the
availability of substantive data as well as confidential tactical information.
SAM
is
local
dynamics of
primarily meant to be used jointly by the parties to understand the
given some a
and not
their negotiation,
priori
evaluation of
The Computer
to predict its global
outcome
utility.
as a Third-Party Neutral
The fourth component of the software under development for
negotiation management, which we have called ONDINE (for "ONeDlmensional NEgotiation"),
an unobtrusive computer environment that
is
allows negotiators to communicate and bargain without the intervention of a
human
facilitator,
and therefore without revealing
each other basic
to
information about their preferences [Goeltner, 1987].
If
there
only one issue being negotiated,
is
of this software,
computer
is
appropriate.
ONDINE
I,
the
first
version
deals primarily with the concept of the
It
as a third party in the process of negotiation.
It
is
designed
to
support a two-party, single-issue negotiation, and sets up an environment in
which the computer functions as a neutral advisor
parties towards agreement.
The two
with the program separately.
sides
ONDINE
to
both sides, guiding the
communicate with each other and
channels the bargaining and the
I
ordinary communication separately but simultaneously. In addition, where
actual bargaining takes place but
that the parties can agree
Goeltner,
1987].
is
stalemated,
on beforehand
Because
of
the
ONDINE
I
offers several rules
for the allocation of benefits [see
unobtrusiveness
of
the
ONDINE
environment, the negotiation can proceed in a natural fashion without the
distraction of the
ONDINE
computer
computer getting
in the
way.
n, the second version of this software, retains the role of the
as a third party in the process but
between two
parties over
more than one
is
expanded
to
cover negotiations
issue (paradoxically but naturally,
its
name has not changed much). Once the parties have agreed on the identity
and number of issues to be negotiated, they give ONDINE II information
about their positions on the issues, but do not divulge to the other side
information about their reservation prices or alternatives to agreement.
12
After considering the offers and the reservation prices of both sides,
ONDINE
informs them
II
none, the solution
is
there
if
any room
is
have reservation prices
either they
trivial:
for negotiation. If there
is
that are
incompatible, or they have misrepresented their reservation prices, or they
have not made
In
any
their
most generous
ONDINE
case,
some combination
offers, or
simply requests
II
new
of the above.
information.
there
If
is
genuinely no room for compromise on a particular issue, no amount of
negotiation will solve the problem. Given this information, the parties are
freed to focus their efforts in
the dispute rather than
If,
specified to
Using the
role.
ONDINE
has two related qualities that distinguish
total of the satisfaction
ONDINE
utility functions that
II
it
from those not
derived by both parties
its
begins to
both sides have
two sides
each agreement in
i.e.,
is
this set
in the set:
the highest that
ever be, according to the specified utility functions; and
could improve
II
calculates, while the
negotiate, the set of Pareto-optimal agreements,
sum
perhaps on the substance of
indeed room for negotiation,
is
in confidence,
it
fruitful areas,
on the bargaining.
however, there
play a more active
more
(ii)
position except at the expense of the other.
(i)
it
the
could
neither party
If
shown
these
optimal agreements, the parties can compare their negotiated outcome with
the ones on the Pareto-optimal frontier,
agreement from the optimal
be informed that
their
set.
Or,
agreement
if
and perhaps
select
an improved
they so choose, they can simply elect to
not on the Pareto frontier, and then
is
continue their negotiation.
There are undoubtedly problems with the use of
software.
With
simulation,
understanding, especially
simplification.
if
With decision
overlooked or probabilities
With SAM, the
may be
cultural
so great that a
rationally.
And,
it
finally,
manipulate the process
the
model
may
not
all
these pieces of
always
advance
provides either no simplification or overtree
may
analysis, possible
outcomes may be
be impossible to estimate meaningfully.
and organizational differences between the
parties
common concept of fairness cannot be achieved
with ONDINE, second-guessing and the attempt to
may be inevitable. Research designed to provide a
better understanding of these problems
13
is
outlined below in Section IV.
The Model
The paragraphs above
of the
computer
marked emphasis on the
will reveal a
in negotiation
Middle
in the
management. Straus and Bazerman
attention to "the battle of the print-out" in
models can, and indeed have, been used
and even
to bully.
These motivations
to
joint
use
[1985] call
which computers and computer
persuade, to dissuade, to confuse,
for the
use of the computer
to
fail
address one of the basic reasons people negotiate rather than fight or bury
their
it
heads in the sand - namely,
to achieve a cost-effective solution.
Because
focuses on joint problem-solving rather than sophisticated argument,
management
negotiation
in the middle. Especially
is
predicated on the assumption that the model
where
objective information
is
required to deal with
the scientific or technical complexity of the issues, a joint
effort
may be
to the
advantage of
is
and open modeling
all parties.
Collaborative Model-Building
But the concept of negotiation management includes more than joint
modeling.
actually
instead of using a model that comes pre-built, negotiators
If
engage
in joint model-building, the benefits of using
compounded. To use
earlier language, the
a collaborative effort
technical experts
good reason
is
use of ordained models
even better [Nyhart
and advisers understand
all
&
models are
is
good, but
Dauer, 1984]. Even
if
the
the issues completely, there
for the negotiators themselves to jointly conceive, build,
is
and
experiment with simple models of reality so that they can better appreciate
the strengths
joint
and weaknesses
of other parties' positions
and arguments.
model-building effort could precede actual bargaining, or
parallel, the
it
could run in
two concurrent processes informing each other [Sebenius,
Samarasan, 1987].
14
A
1981;
Zooming and Bridging
The
need
practical
to
minimize the additional knowledge and
that are required for for the use of
management
in negotiation
stipulation of simple
price, if at
points out again the logic behind the early
and easy-access modeling. But simplicity comes
at a
accurately.
the substantive issues in a negotiation
cases,
are
not
understood and cannot simultaneously be modeled simply and
sufficiently
may be
computer models and related technologies
all.
many
In
skills
The
natural or perceived hierarchy in a particular substantive area
intimidating. In other cases, several separate
models might
eich describe some of the issues, but cannot be used together
exist that
to
form
a
comprehensive picture, because of vastly different underlying assumptions or
motivations [Samarasan, 1987].
In the language of negotiation
amenable
to simplification
a vertical
movement through
refers
through zooming and bridging. Zooming refers
to
the hierarchy of complexity, v^hereas bridging
horizontal link between hierarchies dealing with parallel
a
to
management, these problems may be
complexities in different substantive contexts [Straus, 1986]. The ability to
zoom
is
important because
it
would allow modelers and model-users
specify their requirements in convenient levels of detail.
And
to
the ability to
bridge effectively between different substantive models, whether ordained or
built
by the negotiators,
negotiators,
and
compensating for
to
make
too
definitions
The
abstract
their
it
would allow
narrow
specializations.
A
facilitators,
each other, thus
cautious note:
it
is
possible
of these ideas, since they are tied so closely to semantic
categories.
ability to
zoom and
and synthesize
relationships.
useful because
their expert advisors to relate issues to
much
and
is
As noted
in
bridge effectively implies the prior ability to
order to grasp the structure of complex
in Section
IV below, these topics are
research questions being studied by the Project.
15
among
the
Zooming and Bridging
The
need
practical
to
minimize the additional knowledge and
that are required for for the use of
management
in negotiation
stipulation of simple
price,
if
computer models and related technologies
points out again the logic behind the early
and easy-access modeling. But simplicity comes
at a
at all.
In
many
the substantive issues in a negotiation
cases,
The natural or perceived hierarchy
accurately.
are
not
understood and cannot simultaneously be modeled simply and
sufficiently
may
skills
in a particular substantive area
be intimidating. In other cases, several separate models might
some
eich. describe
exist that
of the issues, but cannot be used together to
form
a
coiiprehensive picture, because of vastly different underlying assumptions or
r otivations [Samarasan, 1987].
In the language of negotiation
through zooming and bridging. Zooming refers
amenable
to simplification
a vertical
movement through
refers
a
to
horizontal
management, these problems may be
to
the hierarchy of complexity, whereas bridging
between hierarchies dealing with
link
parallel
complexities in different substantive contexts [Straus, 1986]. The ability to
zoom
is
important because
it
would allow modelers and model-users
specify their requirements in convenient levels of detail.
And
to
the ability to
bridge effectively between different substantive models, whether ordained or
built
by the negotiators,
negotiators,
and
compensating
to
make
too
definitions
The
abstract
much
it
would allow
narrow
specializations.
A
facilitators,
each other, thus
cautious note:
it is
possible
of these ideas, since they are tied so closely to semantic
categories.
ability to
zoom and
and synthesize
relationships.
useful because
their expert advisors to relate issues to
for their
and
is
As noted
in
bridge effectively implies the prior ability to
order to grasp
in Section
the
structure
IV below, these topics are
research questions being studied by the Project.
IS
of
complex
among
the
Other Issues
A
few
comments on
brief
the development of the software are in order.
Consistent with the objective of providing software that
and
easily accessed
by non-experts, the
STELLA''"'^ (a product of
is
once powerful
at
High Performance Systems,
Inc.,
Lyme, NH)
systems dynamics simulation environment, and DA:TA''""
TreeAge Software, Brookline,
1987;
and Raker,
1987].
MA)
of
ONDINE
for a
product of
Samarasan,
were developed largely
by the investigators, but the second incorporates
a
Pareto-optimization
routine conceived by Blass and Raiffa [see Goeltner, 1987].
In
(a
for decision tree analysis [see
The two versions
by the investigators from
paper uses
effort described in this
SAM
was written
scratch.
addition to the development and application of these primary
pieces, the
investigate
MIT
Project
on Modeling
and possibly put
to
work
for Negotiation
in
Management
will
an integrated negotiation support
system [see Antrim, 1987] a wide variety of other technologies and generic
'productivity' tools. Included in this last category are such software items as
and spreadsheets. Expert systems have also been
outliners, idea processors,
examined
for applicability to negotiation [see Boose, 1984;
Loofburrow, 1986].
More work has been done under the rubric of CSCW (Computer-Supported
Collaborative Work). This work includes such topics as small group
communication [Shimanoff,
1984], the 'decision
electronic brainstorming (Colab) [Stefik
being examined.
et
room' [Konsynski, 1987], and
ah, 1987].
These areas and others are
The rationale and impetus
investigators' efforts stem
from the simple rule
for
to
this
aspect
of
the
avoid reinventing the
wheel.
With regard
to
hardware, the Project currently operates on two distinct
species of computer. STELLA'^'^,
for the
Inc.,
Apple
Macintosh''"'''
is
Digital
this
personal computer
(a
ONDINE
are designed specially
product of Apple Computer,
Cupertino, CA). The Macintosh interface
described in this paper.
and
DA:TA™, and SAM
is
ideal for the purposes
takes advantage of MIT's
designed for use with a VAXStation
W^
X Window system
workstation
Equipment Corporation, Marlboro, MA). As used by the
workstation runs the UNIX"""" operating system
16
(a
(a
product of
investigators,
product of
AT&T
Bell
Murray
Laboratories,
sounds more complicated than
and
This combination of computer environments
Hill, NJ).
it
really
is,
and gives the software both power
flexibility.
UNDERSTANDING THE IMPACT OF
MODELS AND MODELING
IV.
Among
Negotiation
i)
ii)
the intellectual goals of the
Management
on Modeling
synthetical skills
and
understanding
for
are the following:
how models and modeling might enhance
qualities in negotiation
management;
how models and modeling might
and application of
relate to the
integrative bargaining;
understanding the potential usefulness of models and modeling
in a
iv)
Project
understanding
applicabilit)'
iii)
MIT
broadened definition of negotiation; and
understanding where and
why models and modeling may
or
may
not be used in negotiation.
These research topics are elaborated upon
Enhancing Synthetical
in this section.
Skills
and Qualities
Successful negotiation must be both synthetical and analytical. The
value of analysis in negotiation has been pointed out by Lax and Sebenius
[1986].
It
has also implicitly been taken into account in the concept of
negotiation management, by the discussion on the uses of computer models
in Section
On
IE above.
the other hand, synthesis, or the combining of elements to form a
coherent whole,
negotiation.
is
When
largely an unexplicated concept, especially in the context of
an adjective
is
needed, the word "synthetical"
17
is
used
in
this
draw
paper, to
away from
innovation, and
and coherent
attention towards the idea of creative
which connotes an
the aUernative "synthetic"
inferior artificiality.
The
intellectual challenge in the area of synthesis
A
discovery.
is
essentially
one of
negotiation process that arrives at an outcome by integrating,
combining, composing, creating, or unifying must somewhere along the
way
include a significant synthetical component. Perhaps this component
is
located in the intuitive skills of an experienced and imaginative facilitator
who, having accompanied the
parties through the negotiation, can
draw
the
interests, stakes,
and issues together into one multi-faceted agreement. Or,
possibly,
be found
it
to
is
among
the negotiating parties themselves, in the
tension between collaboration and competition as they "create and claim."
it
could conceivably grace the mind of the rare 'Renaissance' individual
can rally data and knowledge from
a
wide variety
Or
who
of fields or disciplines to
fashion a creative and original solution to the problems being negotiated. The
answer
is
It
the
not obvious.
could also turn out to be no more than coincidence that the use of
computer
is
a highly synthetical activity.
The computer allows
the user to
hold, arrange, sort, analyze, combine, and present large quantities of data. But
on another
in that
it
concept of negotiation management
level, the
attempts to integrate a coherent
set of ideas
and
is itself
tools,
synthetical,
each of which
can be used flexibly according to the dictates of the substantive context or
process problem.
And
in the enunciation
and development of
management seeks
number of different fields -
foundations, negotiation
S)Tithetically
from a
and behavioral
game
theory,
have individually contributed
The hope
into the 'how'
is
and
to
a
to
eclectically
more
quantitative fields
artificial intelligence
better
-
all
intuitive progress
of
which
understanding of human
form
to reveal
a coherent whole.
new
or
Although
improved ways of
it
is
not expected
that anything in the nature of rigid rules for efficient synthesis will
some
but
that further research in this project will delve deeper
and 'why' of synthesis
combining elements
draw
conceptual
the art of design, the social
sciences, including law, as well as the
of decision analysis,
interaction.
to
its
may
be possible.
18
be found,
The Relationship
to Integrative
Bargaining
Integrative bargaining forms a major part of
as the
dominant paradigm
Lewin
[1987],
it is
in negotiation. In the
what
words
is
widely recognized
and
of Lewicki, Weiss,
the process:
by which parties attempt to explore options to increase the size of the
gain without respect to the division of payoffs. The key to this process is
openness of communication in order to maximize information about one's own and
...
joint
the other's needs
and
persp)ectives, to facihtate discovery of joint gain options.
Walton and McKersie, identifying integrative bargaining
work
[1965], contrasted
each party
tries to
the result that
it
in their seminal
with the process of distributive bargaining,
maximize
its
share of a fixed
what one party wins,
sum
in
which
of available payoffs, with
the others lose.
Several scholars have pointed out that the integrative approach
offer significant
may
advantages over the distributive. Lax and Sebenius [1986]
note, for example, that:
...
integrative bargains
agreement probabilities;
:
(3)
(1) lead to speedier settlements; (2) have higher
reduce the danger that one or more parties will
repudiate the agreement; (4) tend to strengthen the relationship between the
parties thus facilitating later agreements, and (5) contribute to organizational
effectiveness where subunits ... with distinct needs and values engage in intraorganizational bargaining.
If
...
parties value different aspects of the settlement differently,
and
if
their
thinking style predisposes them towards the visualization of trade-offs and
joint gains, then integrative
[Fisher
and Ury,
Beyond
bargaining can be used to maximize joint
utility
1981].
their
usefulness
in
negotiation,
the
ideas
integrative bargaining appear to be responsive to problems
levels. In assessing the
captured
on several
in
societal
present and future place of the United States in the
global order, Reich v^ites [1987]:
... The most complex questions alx)ut our place in a changing world, as they
are processed by the conceptual filter of political mythology, are reduced to a
blunt and binary choice between toughness and charity toward 'them.' ... This
narrow spectrum of choice - assertiveness versus accommodation, discipline
versus conciliation - bounds our political debates, limits how problems are
defined and solutions weighed, and blinds us to a subtler set of options. ... The
common error of both variants is the rigid delineation of 'us' and 'them.' ... The
tension between a basic stance of accommodation or one of confrontation excludes
19
ground of negotiations and collaborations
and comprehend 'theirs.' ...
the middle
interests
that
both assert
'our'
Freely interpreted, Reich might have been describing a social orientation
towards the bargaining styles associated with distributive situations, which a
widened use
of integrative bargaining concepts might ameliorate.
Especially
those
in
problems characterized by technical
social
complexity, a need for early and decisive resolution, and a diffused decision-
making
authority,
a
focused application of the principles underlying
integrative bargaining might be useful. This assertion
such problems are viewed as
a series of negotiations.
agreement requires
a
a cluster of activities
And
joint problem-solving,
is
particularly apt
if
over time, and therefore as
since the problems are so complex, reaching
combining of disparate parts and
solution - a step which
is
activities into a
whole
highly synthetical and integrative. Collaboration, or
therefore,
may be an
effective
approach for these
problems.
In the field of law, a
growing emphasis on preventive law [Browm
Dauer, 1978; Sciamanda, 1986] with
[Nyhart
&
attempt
its
&
link to alternate dispute resolution
Dauer, 1987] provides another conceptual underpinning for the
expand integrative bargaining. The structural options
to
in
preventive law of dealing with the prospects of future disputes by
"presentiating" and resolving, deferring, or avoiding
them
(the latter in the
sense of making agreements that leave minimal possibility of future disputes)
would seem
to call forth objectives closely akin to those associated
integrative bargaining - the effort
to
expand the number and nature of
possible outcomes while treating issues as
jointly, rather
with
common problems
than battled to resolution through
to
be resolved
litigation.
These are among the considerations informing the premise that the
attempt to increase the integrative quality of negotiations
effort. Fortunately,
might help
To
it
would appear
that the use of
is
a
worthwhile
models and modeling
in that effort.
a certain extent, the joint use of models implies a
the tenets of integrative bargaining.
commitment
As Lax and Sebenius point out
to
[1983],
bargaining for competitive advantage often co-exists with a less explicit
bargaining for knowledge. Because knowledge and information are the
essential ingredients in the evaluation of utility, irrespective of negotiating
20
between integrative or synthetic negotiation,
style, there is a natural link
which requires access
modeling
sciences,
to a great deal of
which can provide
The software created
among
for
knowledge and information, and the
that access, or
an approximation of
it.
use in negotiation management provides,
other things, the capability to model different styles of negotiation,
and different concepts of
Using these models, negotiators can
fairness.
compare the various conceptual frameworks and
integrative techniques to their situation.
might allow researchers
test the applicability of
And academic
use of
this
software
to further explicate the principles of integrative
bargaining.
The establishment of consensual databases, from which
draw
data,
and
to
which
parties contribute data,
all
parties can
all
the natural starting
is
point for a collaborative problem-solving process. In the kinds of conflicts
under discussion, the collaborative process
model-building
effort.
Even
if
by each party's expert, there
offers
As
activities,
a parallel
good reason
more
for the negotiators themselves to
rationally separate
good arguments and
Broader Definition of Negotiation
illustrated in Figure 2 above,
it is
believed that in at least fifteen
models and modeling can potentially
management. One way
to
understand these
tentative version of a descriptive
assist
activities is to
framework
in
negotiation
view them as a
for negotiation
management,
around the use of computers, models and modeling. Another way
interpret them, given that differences
them,
by
from bad ones.
A
built
greatly aided
the technical issues are completely understood
is
build the models, so they can
is
is
to see the
framework
and disputes can and do occur
to
in all of
as extending the idea of negotiation far
beyond
beyond
explicit
actual bargaining.
The recognition
that the process of negotiation extends
bargaining - beyond the 'actual negotiation' literature.
divides
it
Susskind and Cruikshank [1986]
into three
set
is
not unfamiliar in the
out a model of the process that
main phases: pre-negotiation, negotiation, and
21
post-
The
negotiation.
claim,
basic trichotomy
and reconfigure - further
settlement follows what
Straus
and Bazerman
-
create,
reflects these ideas. Raiffa's post-settlement-
[Raiffa, 1985].
have also recognized the basic difference between
and bargaining phases of negotiation. From the
the consensus-building
standpoint.
[1986]
normally considered negotiation
is
[1984]
made by Lax and Sebenius
Brown and Dauer,
legal
in their concept of preventive law, recognised
these general ideas earlier [1978]. Nyhart
and Dauer linked the
activities of
preventive law and alternate dispute resolution to suggest a spectrum of
foreshadow those expressed here
activities that
At
this point,
may
it
analyze them in some
be worth returning
to the fifteen activities, to
detail. In reality, these activities
parallel streams, a series of cycles, or
simplicity, the activities are presented
1)
[1984].
even a converging
below
in a
may form
several
spiral iteration. For
simple and linear
Hst:
Establish a consensual database as a foundation for agreement
Gaining consensual agreement on the basic
large problem
is
facts
underlying
when
an early task, and one which,
a
achieved,
quickly identifies areas of disagreement on the fundamental
technical uncertainties, points to areas of disagreement over fact,
and
which consensus
indicates the extent to
In doing so,
many
issues
may
already achieved.
is
systematically be
removed
potential points of subsequent adversary confrontation.
of modeling in establishing databases
technical databases are models,
joint
modeling requires
held by the parties. At
and
is
as
The use
fundamental - many
vice versa.
specificity as to values
The process
of
and assumptions
this early stage of the negotiation,
it
also
provides an opportunity for the parties to improve their
communication and
relational skills vis-^-vis each other.
Lax and
Sebenius, in referring to "knowledge bargaining" [1983], attest to
the importance of this stage of negotiations.
2)
Assess the impact of uncertainty and change
As explained
in
the database
in the discussion of decision tree analysis in
Section HI above, information, interests, or positions that might
seem
crucial at first blush
comparison
to
may be found
after
some
be dominated by other information,
22
analysis
and
interests, or
positions.
decision tree models, for example, can
DA:TA''"'^
therefore be used
in
evaluating the database to scope the
discussion and to focus
on those
it
make
details that really
a
significant difference.
3)
Set a
common
Once
research agenda to collect useful additional data
the initial consensual databases are evaluated,
and the
significant deficiencies in data identified wi\h the use of models,
the parties can use this information to establish a
commonly
acceptable research agenda on what nev^ data need to be collected.
In particular, substantive STELLA''"'^ simulation
an instrumental role here: dominated
to
be fully investigated
as
if,
is
models can play
may
possibilities
not need
usually the case, there are only
limited resources for information-gathering.
4)
Probe inequalities of power, influence, and access
SAM,
Referring nov^ to the prior discussion of
parties will not agree a priori
Especially
SAM
if
there are
might help
many
in
data
to
it is
on what constitutes
a fair process.
parties to the negotiation, joint use of
formulation
the
of
decision-making
procedures and norms of fairness that are acceptable
5)
likely that
to all parties.
Track the formation of coalitions and clusters
SAM
can be used by negotiators to
jointly the impacts of
unequal distributions of power,
As indicated
examine
influence,
approach
earlier,
and information on the course of the negotiation; the
is less
abstractly mathematical than similar approaches
taken by other modelers [see Myerson, 1979, for example]. Details
of the tracking
parallel
6)
methods and motivations
are elaborated in a
paper [Samarasan, 1987].
Formulate new procedures for making collective decisions
The suggestion here
is
that negotiators can use the results of
the process simulations that
simulations that
effects of the
STELLA™
norms
SAM
provides and the substantive
provides to understand the practical
that they hold,
and
to begin to relate
contrast political agreeability and technical effectiveness.
understanding of these relationships,
23
it is
suggested,
may
A
and
better
lead to a
or willingness
desire
decision-making
7)
institute
to
a
more
collective
rule.
Practice cross-cultural negotiating styles and tactics
Other researchers have concentrated an
specific
example
them with
its
in a training
inter-cultural
mode,
&
a negotiator can begin to understand
differences
is
that
SAM
SAM
a different
Gercik, 1986]. Using such a
negotiating
tactics,
in
communication, and become familiar
certain extent,
in dealing
and
style,
with them. To a
performs the same function. The difference
makes no attempt
an automated negotiator.
to act as
Provide communication links for bargaining and discussion
Face-to-face
communication between the parties
more dimension
negotiation adds one
there are
many
instances
when
face-to-face
negotiators are geographically far apart,
communication
it
may
often be
There
communication may
instance,
may
it
are
facilitate
also
times
ONDINE
with these issues.
Communicate
to deal
effectively
The obvious and
while protecting secret
critical
remote
information
importance of effective and trusted
examples by writers from such diverse
in
example: Allison, 1971; Kettelle, 1986; Walton
&
[see,
for
McKersie, 1965].
the software currently developed for negotiation
management,
addresses the issue of communication most
24
many
fields as international
diplomacy, arms control, and industrial relations
ONDINE
more
or encourage
channels of communication has been illustrated
Of
when
made without embarrassment.
to
intended
when
down
admissions and retractions
is
is
the emotional interaction: for
allow tempers to cool
be
a
messaging and
practical to negotiate remotely through electronic
teleconferencing.
to
to the interaction. Clearly,
either expensive or inconvenient. In large groups, or
9)
A
knowledge base and combines
knowledge base about negotiators from
a
organization or culture [Susskind
system
effort in this area.
the use of an expert system that captures
is
effective negotiation rules in
8)
useful
literally
and
directly.
It
allows the parties to communicate effectively
without revealing their basic negotiating positions.
10)
Discover the areas of potential agreement among sub-groups
Both
ONDINE
and
SAM
areas of potential agreement
allow the parties
among
to
discover
all
sub-groups. They accept from
the parties information that allows the models to determine the
range of solutions for each issue that
However, what the
parties
is
do with
acceptable to
this
all parties.
information
is
not
prescribed automatically.
11)
Suggest settlements that minimize the need for compromise
One
norms included
of the
in
SAM's algorithm
simple idea that the best settlement
least
compromising on the part of
choose
SAM
norm,
this
the least total
movement
the parties.
all
of
all
parties
issues.
is
the
the one that requires the
computes the settlement
on the
ideal settling points
is
library
away from
The goal here
If
the parties
that results in
their respective
is
not so
much
to
locate that particular settlement, but to illustrate the effect of the
norm on
other
and course
the quality
norms
in the library that
of the negotiations. There are
SAM
can suggest, or the parties
can design their own.
12)
Evaluate the substantive impact of proposed settlements
This
is
perhaps the most obvious of the roles that a computer
model can play
in a negotiation. Countless
adversarial use of models can be found.
examples of the
The suggestion here
that joint substantive modeling, with DAiTA''"'^
for instance, allows
all
is
and STELLA^*^,
parties to see the impacts of proposals,
both theirs and others'. There are often times when such
information,
13)
Enhance
If
them
if
shared, can precipitate agreement very quickly.
initial
agreements through Pareto-optimization
the parties reach an agreement, simple techniques can assist
in
enhancing
its
value [Raiffa, 1985].
ONDINE
uses a Pareto
optimization routine to point out to the parties potential
agreements where
at least
one party could do better without any
other party doing any worse. Especially
25
when
the issues are easily
valued
a
in
common
currency, this technique can highlight
significant joint gains [see Goeltner, 1987].
14)
Monitor the implementation
A
critical yet often
of the final
agreement
overlooked aspect of negotiation
is
the
monitoring of the implementation of the agreement and of the
overall system or set of issues of
was
the core. In this activity,
dynamic picture
parties a
which the negotiated agreement
models can be used
provide
of the implementation. In the ideal
models that were
case, substantive
to
built
during the negotiation,
either as self-standing ones or as front-ends to
more rigorous
models, would be used [Samarasan, 1987]. This double duty
would help reduce
It
would
also
the effective cost of the initial modeling effort.
minimize the unacceptability of the monitoring
models.
15)
Reach an amicable solution should disputes arise anew
Inevitably, differences will arise over the subject matter of the
negotiation.
Some
may develop
of these differences
disputes. Anticipating these developments, parties
advance
to
may
agree in
use the models that have been validated as bases for
objective criteria
in
disputes in
a
non-
where appropriate, adjudicators could
of such models: there
in adjudication
new
resolving the
adversarial setting. Or,
make use
into vigorous
[McGovern,
is
precedent for the use of models
Of course,
1986].
in
some
contexts
these early models might have to be updated before they can
serve a sensible function.
The Use and Non-Use
Although the uses of models discussed
part, not revolutionary in the
new
to the
in this section, is
in this
paper
are, for the
most
world of computer science, they are relatively
world of negotiation. Therefore,
with resistance and rejection.
Models
of
Much
of
concerned with ways
26
their
what
to
is
use can be expected to meet
said above, in Section IE
and
make models and modeling more
Enough
useful and widely acceptable.
now known
is
or suspected, however,
much
about perceptions that lead instead to their non-use, so
so that an
understanding of the obstacles to the increased use of models in negotiation
is
essential.
The kinds
extent, be
many
made by
of criticisms
met by the observation
that the
of the types of disputes for
and decision
the dispute
settlement
intuition of the
lack
of
settlement)
fail to
attempt a model
consequences, to
to fail in
an attempt
And
to
is
III
can, to
some
not a panacea. But in
which the concepts of systems simulation
beyond the
(or
computer
most appropriate,
tree analysis are
is
the authors in Section
i.e.,
where the substance
of
unaided human mind, or where the
dispute
the
of
may prove
to
be
has
far-reaching
more
far
costly than
model.
with regard to the role of
essentially the chance for an
SAM
and
ONDENE
as agents, they offer
improved process. Many writers on alternate
dispute resolution have urged the caveat that certain kinds of problems are
simply not amenable
to decisive resolution
through negotiation [Lewicki
Ozawa & Susskind, 1985; Raiffa,
These problems commonly end up either in
Litterer, 1985;
negotiation.'
At
least in these cases,
management may turn out
Some
to
1982; Fisher
and
&
Ur>', 1981].
the courts or in 'extended
an improved process through negotiation
be the best alternative to stalemate.
other apparently legitimate bases of resistance have also been
anticipated. Lax, for example, has speculated [1987] that the usefulness of
Pareto-optimization software will be a
expect on
initial
From
than one might
lot less significant
consideration.
the perspective of Negotiation Analysis as elaborated
Sebenius [1986], a
critical
element
is
the interplay
between competition and
cooperation in such bargaining. Parties to a negotiation,
joint gains at all, are
by Lax and
if
they are creating
simultaneously creating these gains and dividing them
up amongst themselves - an intertwined
expected to advance their
self-interests.
process. Naturally, parties can be
And
they
distorting the statement of their preferences
may do
and
attempt to do so by
utilities
when using
a
Pareto-optimization routine, providing biased information to skew the
'optimal'
outcomes towards
their true preferences.
27
Lax and Sebenius note
that
this
problem
essentially
is
one form of the Negotiator's Dilemma [Lax and
Sebenius, 1986]:
An
...
by game
due
result,
dilemma
abstract version of a similar
is
currently being investigated
theorists studying bargaining with incomplete information.
analogous
Myerson
to
to the
ones
One major
(1979), is that in highly simplified situations that are
we
discuss, fully honest revelation of private information
by individual bargainers and Pareto efficiency cannot simultaneously be
is a starkly abstracted analogue of the result we describe that
tends to result from the tension between cooperative moves to create value and
competitive moves to claim it individually. ...
achieved. This
Lax believes
that
when
using models, negotiators will
from negotiating into trying
efforts
shift their
immediate
understand the iDlack box' of the
to
optimization model in order to claim more of the joint gains. Extending this
may
argument, he argues that negotiators
often be unwilling to
commit
dispensation of their negotiation to an optimization routine whose role
agreements that maximize
to find
Be that as
it
is
may
outweigh the objections raised above.
that
complex negotiations with many
issues,
and unknowns. The capacity of an optimization model
in that case aid
the analysis because analysis
could flag those situations where -
without
it.
utilities
- no solution space
It
is
joint gains.
the case of very
parties, possibilities,
it
may, there remain times when the uses of optimizing
models may provide advantages
One example
the
exists.
And
if
would be impossible
at least
a basic
according to stated
agreement
is
eventually
reached, a Pareto-optimizing model might then be used to enhance the
agreement
[Raiffa, 1985].
In a sense, the concerns expressed above
different issue than the use of
be about the degree
to
may
in essence turn
on a
computer models. The underlying doubts may
which the negotiation
at
hand
perceived and treated
is
according to the principles of integrative or distributive bargaining.
Nonetheless, they do point out the importance of understanding the
theoretical
as well
as practical obstacles
to
the use of computers
and
information-based modeling.
Different reservations about the use of models
complex or science-laden negotiations.
[1987] interviewed persons
the recent negotiation
who had
may
In a recent study
in
Clements and Sossen
one way or another been involved
and implementation
28
arise in technically
of
a
monitoring plan
in
for
proposed
oil
and gas exploration of the Georges Bank area offshore
England under the Department of the
was anticipated
study
oil
that
some
essentially
At the outset of the study,
of the several existing
and gas productivity,
might have been useful
Interior.
oil pollution,
computer models used
and interaction with
to the negotiators. In fact, the
it
to
fisheries
study concluded that
no computer models were used. The reasons given by the
interviewees for steering clear of modeling obviously informs the
Project.
New
From among them, Clements and Sossen summarized
work
of the
the following:
... Over the course of the interviews, we discovered three reasons for the
non-use of models. First, at the time, models for these sorts of negotiations were
underdeveloped. Better models have appeared, however, subsequent to the
Lease-Sale 42 debate. Second, there was distrust of the models that were then
extant. Scientists and negotiators alike expressed reservations about the
adequacy of the input data, the scientific knowledge embodied in the models,
and the implicit assumptions. Finally, the perception of the role of the
computer at that time limited the willingness of negotiators to employ models.
Whether these concerns
reflect
perceptions of those interviewed
that they exist at
all.
is
objective
less
questions, or
important
at this
the subjective
point than the fact
Further research will adapt the Clements and Sossens
protocol to a general format for inquiry that can be used for investigation in
other contexts.
V.
POTENTIAL APPLICATIONS IN
SUBSTANTIVE CONTEXTS
Use
of the software
and processes described here
is
a
major element of
negotiation management. Use initially provides an opportunity to test and
ultimately evaluates the efficacy of the generic software in a range of
substantive contexts.
At the time of v^iting, the investigators are creating contexts from
situations for use in teaching materials.
Joint
Venture, focuses
on
a
One
case, for
example, the Kherov
U.S.-U.S.S.R business negotiation.
examines the potential exploitation of
29
krill
fact
Another
resources in the context of
prospective Antarctic Treaty negotiations.
underway include
a strategic analysis of the
other modeling efforts
Still
submarine nuclear deterrent
in
an arms control context, the simulated enforcement of new international
garbage disposal rules applying
and the management of
to offshore waters,
disputes arising in shipbuilding. Earlier applied modeling efforts concerned
deep ocean mining [Nyhart
et
al, 1978]
The next research phase
will
and
oil spill
be
to
clean-up [Psaraftis, 1985].
work
directly with parties in
negotiations. Substantive contexts are being sought for this purpose in
which
the quality and the course of the negotiation are a central part of the problem.
It
is
hoped
collaborative
that multi-party
work using
and multi-issue situations can be found
the tools at hand.
VI.
DISSEMINATION
At the heart of the MIT Project on Modeling
Management
modeling
is
for
making
the objective of
the integrated set of
paper available
tools described in this
Negotiation
for
models and
to negotiating parties
and
third-party facilitators, for general use and under appropriate distribution
arrangements.
The methods
a systematic
way
of negotiation
management can and should be taught
in
before conflicts arise, and the software developed at the
Project encompasses this purpose. Therefore, another part of the Project
preparation, testing, and packaging of teaching materials for use at
is
the
MIT and
in
law, management, engineering, and public service professional schools.
Software and written teaching materials providing
instructions will enable students to create their
simulated
facts.
fact
scenarios
own models
and
based on the
Negotiation exercises, comprising materials for simulating
the process of joint model-building will be provided along with appropriate
textual material.
The
intent
is
to
provide a complete set of materials for the
use of students, negotiators, and third-party neutrals.
30
Another means of dissemination, the MIT Research Colloquium on
Modeling
for Negotiation
critique current project
Management, has been established
work and
to
presentations of the current state of the
as
a
to discuss
and
hear scholars' and practitioners'
art.
The Colloquium, which functions
research and teaching seminar, serves as a focal point for such
interchange and for the development of graduate, undergraduate, and faculty
research.
One
of the goals of the project
is
educate Ph.
to
D
students in the
uses of modeling in negotiation management, in the context of an integrated
approach. In this regard, the Colloquium will continue to be an element of
that effort.
^it»iHHHHHHHHk*******»*»*****
The
long term goals of the
Management
models
in
materials,
Project
negotiation,
design
the
negotiation
the
of
creation
of computer
methods
formal
of
practice
possible
a
negotiation.
optimization,
software
instruction
the assessment of the implications
and supporting
in
negotiation
of the use of computers
and decision-making.
The experience gained so far suggests
the
Modeling for Negotiation
in
include the development of principled guidelines for the use of
management, and
in
MIT
and
teaching
deeper
of
negotiation
understanding
of
that
crucial.
is
complex
But more generally, the computer
so
that
negotiations
are
robust.
31
the use of the
more
will
and
in
The computer makes
technical
help in
efficient,
computer
issues
in
organization and
and agreements more
REFERENCES
Allison, Essence of Decision. Boston: Little,
Brown, 1971.
Negotiation" in Coastal Zone and
Continental Shelf Conflict Resolution, Nyhart, ed. Cambridge: MIT Sea
Antrim, "Computer Models as an Aid
Grant Publication
MITSG
85-28,
to
MIT, 1985.
Antrim, "Computer-Assisted Negotiation: A Report on Research Sponsored
by the Unisys Corporation," American Academy of Arts and Sciences,
1987.
Axelrod, The Evolution of Cooperation.
Boose,
New
"A Negotiation Advisor," ETS
Services,
York: Basic Books, 1984.
Application Series, Boeing
Computer
BCS-G2010, 1984.
Brown & Dauer, Planning
Process. Mineola,
New
by Lawyers: Materials on a Non-Adversarial Legal
York: Foundation Press, 1978.
&
Sossen, Computer-Aided Negotiation: The Role of Computer
Models in Resolving Offshore Resources Disputes. Joint Master's thesis,
Alfred P. Sloan School of Management, MIT, 1987.
Clements
Fisher & Ury, Getting to Yes: Negotiating Agreement
Boston: Houghton-Mifflin, 1981.
Forrester, Urban Dynamics.
Without Giving
In.
Cambridge: MIT Press, 1969.
A
Decision Support System for
Two-Party Single-Issue and Two-Party Multi-Issue Negotiations,"
Working Paper 1958-87, Alfred P. Sloan School of Management, MIT,
Goeltner, "The
Computer
as a Third Party:
1987.
Gulliver, Disputes and Negotiations:
York: Academic Press, 1979.
Kettelle,
OR
A
Cross-Cultural
Perspective.
New
"Systems Analysis and Mathematical Modeling in Arms Control,"
Spectrum, Vol. 8 [1986].
&
Verma, "Negotiations in Organizations: Blending Industrial
Relations and Organizational Behavior Approaches," Working Paper
1418-83, Alfred P. Sloan School of Management, MIT, 1983.
Kraemer, "Modeling as Negotiating: The Political Dynamics of Computer
Models in Policy-Making," Advances in Information Processing in
Kochan
Organizations, Vol. 2 [1985].
A
Workshop for Exploring
Arms Races. Systems Dynamics Group, MIT, 1985.
Kreutzer,
Personal Computer
32
the
Dynamics
of
Lax, remarks made as panel member. Panel On Systems Approaches to
Negotiation, Conference on the Problems of Constancy and Change,
Budapest, 1987.
Lax
&
Lax
&
Sebenius, "Negotiation and Management: Some Analytical Themes,"
Working Paper 1-784-034, Harvard Business School, 1983.
Sebenius, The Manager as Negotiator.
Lewicki
&
Litterer, Negotiation.
Homewood,
New
York: Free Press, 1986.
Illinois:
Richard Irwin, 1985.
& Lewin, "Models of Conflict, Negotiation, and Conflict
A Review and Synthesis," manuscript in process, 1987.
Intelligence of Democracy. New York: Free Press, 1965.
Lewicki, Weiss,
Intervention:
Lindblom, The
Loofburrow, remarks made as guest speaker, MIT Research Colloquium on
Modeling for Negotiation Management, MIT, 1987.
McGovern, "Toward
A
Functional Approach for Managing Complex
Law Review, Vol. 53:440 [1986].
Compatibility and the Bargaining Problem,"
Litigation," University of Chicago
Myerson,
"Incentive
Econometrica, Vol. 47 [1979].
Nunamaker, Applegate, & Konsynski, "Facilitating Group
Experience with a Group Decision Support System,"
Management Information Systems, 1987.
Creativity:
journal
of
Nyhart, "Negotiating Conflict Over Marine Resources: The Use of Multi-Party
Models," Environmental Assessment Impact Review, 1984.
Nyhart, Antrim, Capstaff, Kohler, & Leshaw, A Cost Model of Deep Ocean
Mining and Associated Regulatory Issues. Cambridge: MIT Sea Grant
Publication
MITSG
78-4, 1978.
Nyhart & Dauer, "A Preliminary Analysis of the Uses of Scientific Models in
Dispute Prevention, Management, and Resolution," Missouri Journal of
Dispute Resolution, 1986 [originally presented at the Conference on
Coastal Zone and Continental Shelf Conflict Resolution, 1984].
&
Goeltner, "Computer Models as Support for Complex
Negotiations," Working Paper 1955-87, Alfred P. Sloan School of
Nyhart
Management, MIT,
1987.
Ozawa &
Susskind, "Mediating Science-Intensive Public Policy Debates,"
Working Paper 85-4, Program on Negotiation at Harvard Law School,
1985.
Damage and Liability from Oil Spills," in Coastal Zone
and Continental Shelf Conflict Resolution, Nyhart, ed. Cambridge: MIT
Sea Grant Publication 85-28, MIT, 1985.
Psaraftis, "Assessing
Raiffa, The Art
and Science
of Negotiation.
Cambridge: Harvard University
Press, 1982.
Raiffa, "Post Settlement Settlements," Negotiation Journal, January, 1985.
33
Raker, "Calculating Litigation Processes", National Law Journal, April
14,
1986.
Reich, Tales of
A New
New
America.
York: Times Books, 1987.
Samarasan, "Meta-Modeling and Negotiation Management," Working Paper
1957-87, Alfred P. Sloan School of Management, MIT, 1987.
Sciamanda, "Preventive Lau' in the Corporation," Alternatives
Cost of Litigation, Vol. 4 No. 9 [1986].
Sebenius, "The
Computer
Management, Vol.
1,
1
the
High
Journal of Policy Analysis and
as Mediator,"
No.
to
[1981].
Dynamics, Mental Models and the Development of
Management Intuition" in Proceedings of the Keystone Systems
Dynamics Conference, Keystone, 1985.
Shimanoff, "Coordinating Group Interaction via Communication Rules," in
Small Group Communication, Cathcart, ed. Dubuque, Iowa: William C.
Brown, 1984.
Stefik, Foster, Bobrow, Kahn, Lanning, & Suchman, "Beyond the Chalkboard:
Computer Support for Collaboration and Problem-Solving in Meetings,"
Communications of the Association for Computing Machinery, January,
Senge,
"Systems
1987.
Straus,
"Computer-Assisted Negotiation," American Arbitration Association,
1974.
Straus, Koh, Nyhart, Richardson, & Sebenius, "Computer-Assisted
Negotiations: A Case History from the Law of the Sea Negotiations and
Speculations Regarding Further Uses,"
Vol. 426 [1984].
New
York Annals of Science,
Straus & Bazerman, "New Techniques and Cognitions for Effective Dispute
Resolution," in Coastal Zone and Continental Shelf Conflict Resolution,
Nyhart, ed. Cambridge: MIT Sea Grant PubHcation 85-28, MPT, 1985.
Susskind
&
Cruikshank, Dealing with Differences: Practical Approaches
Resolving Important
manuscript, 1986].
Public
Disputes
(forthcoming)
to
[unpublished
Gercik, remarks made as guest speakers, MIT Research
Colloquium on Modeling for Negotiation Management, MIT, 1986.
Susskind
&
&
McKersie, A Behavioral Theory
McGraw-Hill, 1965.
Walton
34
of Labor Negotiations.
New
York:
^53U 056
MIT LIBRARIES
3
TDflD
DDS 132 DD3