Gregory
E. Kersten and Sunil
J. Noronha
Centre for Computer Assisted Management
Carleton University
Submitted for publication. Do not replicate any part of the contents of this page or link to it.
Abstract. Support for international negotiations requires integration of decision-theoretic approaches with communication facilities, and different visualization modes. In addition, NSS should also be tailored to different cultural and educational backgrounds of their users. While there have been studies on cross-cultural negotiations involving simple game or economic models, there have been no experiments with NSS in international and cross-cultural contexts. At the same time the emergence and quickly spreading use of the WWW and elctronic commerce indicates the potential of NSS supporting commercial transactions accross borders. This paper presents INSPIRE, the first Web-based negotiation support system that has been tested and used in teaching and training in several countries. The architecture of INSPIRE, which heavily relies on the net-centric computing paradigm and object oriented design is also discussed.
Acknowledgements: We wish to acknowledge the contributions of John Bowen, Ying-Hueih Chen, David Cray, Anantha Mahadevan, Kumudini Ponnudurai and Ravi Ramsaran who are the co-developers of the InterNeg site and the INSPIRE system. This work has been supported by the Social Science and Humanities Research Council of Canada, the Natural Sciences and Engineering Research Council of Canada, and the International Institute for Applied System Analysis.
International negotiations have been studied from many perspectives including psychology, sociology, political science, economics, applied mathematics, engineering, computer science and artificial intelligence (Ikle and Leites 1962; Gulliver 1979; Pruit 1981; Raiffa 1982; Adler 1983; Fisher and Ury 1983; Cohen 1991; Kremenuk 1991; Young 1991; Fraser 1992; Kersten 1995). These perspectives have also been incorporated in negotiation teaching. Simulation is an often used vehicle to study and teach negotiation (Adler and Graham 1989; Bazerman and Neale 1991; Winham 1991; Darling and Mumpower 1992; Graham, Mintu et al. 1994).
One of the dimensions adding to the difficulty and complexity of studying and conducting international negotiations is the fact that they involve people from different cultures and with different social and educational backgrounds. Studies of cross-cultural negotiations include analysis of cases involving two or more parties from different countries (Lockhart 1979; Faure and Rubin 1993) and interviews with, and written experiences of, experts and negotiators who participated in international negotiations (Fisher 1980).
Negotiation simulations allow for focused research and for analysis and verification of specific behaviors and interactions. However, simulations are usually conducted in laboratories, in unicultural environments. To obtain results pertinent to international and cross-cultural negotiation, experiments have been conducted in several countries in a manner that allows comparison of results (see, e.g., (Adler and Graham 1989; Roth, Prasnikar et al. 1991; Graham, Mintu et al. 1994; Carnevale 1995). These experiments often involve bargaining records enriched with questionnaires and reports from large samples and many countries (Bond 1987; Hofstede 1989) but rarely involve negotiations between people from different cultures. There has been little comparative research, especially research involving inter-cultural negotiations. There has been no research that has considered the use of computers and other sophisticated communication media in negotiation, in terms of their relative effectiveness across users from different cultural backgrounds, as well as the effect of different cultural group decision making styles upon the design of electronic aids to negotiation.
The objective of the InterNeg project is to fill this void and develop an environment that supports remote negotiations over simple or complex problems, including real-life situations. This will be achieved through the development of a site on the World Wide Web that comprises electronic bargaining facilities, analytical tools, quantitative and qualitative models, summaries, abstracts, annotated bibliographies, research results, surveys, other forms of expertise, etc. InterNeg will allow negotiations in multiple locations by many people at the same time and with no additional cost to negotiators. By its very nature InterNeg has the potential to become accessible to everyoneólay people and experts alikeóand provide them enabling tools and information to interact more directly and to bring negotiation expertise to lower level in social and organizational hierarchies.
The InterNeg site, and a prototype of one of its Web-based negotiation support systems, INSPIRE, have been operational since July 1996. In the first eight months, over 300 people from India, Canada, Finland, Portugal, U.S., Hong Kong, and Korea have negotiated through InterNeg. Because it is Web-based, inter-cultural negotiations are carried out as easily as intra-cultural negotiations. The negotiations include offer exchanges together with messages. Participants can conduct negotiations anonymously, not knowing their counterpartís country or background. There are facilities that help the user evaluate the goodness of an offer, review the history of a negotiation, and determine whether an agreement can be improved in a way that satisfies both sides. InterNeg provides systems and information that can be used at any place that has an Internet connection.
This paper describes the development of the InterNeg site and its current and future services, the architecture of the INSPIRE negotiation support system and the initial experiences with INSPIRE. It sets a basis for future research on the impact of computer and communication technologies on negotiations and international communication within and between cultures. It also aims at providing directions for the development of Web-based negotiation support tools for real-life negotiation with and between human and artificial agents.
2. Negotiation support via the Web
InterNeg is a computer-based environment that builds upon two emerging technologies: net-centric computing and negotiation support systems (NSS). Together, these technologies are integrated into a suite of resources that facilitate negotiation, training and research.
2.1 World Wide Web: technologies and services
Recent developments in computer and communication technologies have contributed to two trends that are critical to this research project:
The principal feature of the World Wide Web is that it allows people from different locations and time zones to communicate and to use previously inaccessible computational resources. While the Webís greatest use currently is as a powerful source and means for dissemination of information, it is increasingly being used as a means for remote execution and control of complete software systems, thus adding another dimension to the value it delivers.
The Web permeates research and education. Its services include access to techniques and tools which are used in research (e.g., statistical packages and mathematical programming software), access to data and information for processing and analysis, and generation of massive amounts of data. In education, the ability to access and run remote programs and databases allows its users to extend classroom and laboratory boundaries across geographical and time zones. It allows instructors and students to retrieve and use resources from remote sites. However, to our knowledge, the Web is not used in studying and teaching negotiations. Extensive search of Web resources showed that there is a very small number of sites devoted to negotiation. None of these sites use the Web for research or training activities that are accessible to students, researchers and practitioners.
2.2 Negotiation support systems
Negotiation support systems (NSS) are computer-based software tools typically used for training and research in a laboratory environment (Kersten 1985; Jelassi and Jones 1988; Alemi, Fos et al. 1990; Fraser 1992; Andriole 1993; Fang, Hipel et al. 1993; Bui 1994; Holsapple, Lai et al. 1994; Rangaswamy and Shell 1994; Balakrishnan and Eliashberg 1995). In teaching and training NSS are used to:
While there is a large and growing number of NSSs, they are rarely used in real negotiations (Nyhart and Samarasan 1990; Teich, Wallenius et al. 1994; Kersten 1997). Moreover, systems that have actually been used in real negotiations are not of the NSS type. Rather, they are traditional model-based decision support systems allowing for what-if and sensitivity analyses, and simulation of potential effects of the contemplated compromises.
Two current trends that may potentially lead to widespread use of NSS in real negotiations are:
(Zartman 1993) argues that decision analytic methods based on multi-attribute theory, simulation modeling, statistical analysis, and cognitive mapping have shown their usefulness. The INSPIRE system uses decision theory and supports construction of utility functions. The growth of Web-based commerce systems may be illustrated by the recently created corporations (e.g., UNIBEX) which provide businesses and other organizations with the electronic market and tools for exchanging information, and structuring and recording negotiations.
3. InterNeg: resources and services
The InterNeg project began in 1996 with the development of a simple Web-based negotiation support system prototype. The software, called INSPIRE (InterNeg Support Program for Intercultural REsearch), has been tested and enhanced since the Summer of 1996. In July 1996, the system became fully operational. The host InterNeg site also offers other services. It is the home site of two journals, International Negotiation and Group Decision and Negotiation where basic information such as the table of contents, abstracts, and calls for papers are maintained. Beginnersí handouts, information on how INSPIRE can be used in different university courses, and examples of student assignments are available as well as several articles, an extensive bibliography and information about programs and organizations involved in negotiation.
The InterNeg site is a source and repository of negotiation-related resources and its home page is at http://interneg.carleton.ca/. The site is organized into five departments in addition to general information about the site, its history, users and developers. An overview of the site is presented in Figure 1.
Figure 1. The basic structure of the InterNeg site.
Each department has a different focus:
INSPIRE is the first system designed to conduct negotiations on the Web. It is a support system based on analytical models rooted in decision and negotiation analysis (Kersten 1985; Kersten and Szapiro 1986; Lax and Sebenius 1986; Sebenius 1992; Rangaswamy and Shell 1994). Developed in the context of a cross-cultural study of decision making and negotiation, the system has been primarily used to conduct and study negotiation via the World Wide Web as well as in the teaching of information systems, management science, international business, and English as a second language.
INSPIRE views a negotiation as a process occurring in a particular context. It comprises a series of activities beginning with pre-negotiation which involves preparation for negotiation, proceeding through the actual conduct of the negotiation during which messages, arguments, offers and concessions are exchanged and evaluated by the parties until an agreement is reached, and finally, implementation of the agreement. It is usually inappropriate to assume that reaching an agreement is the goal of the negotiation, as is often assumed in low-context societies (Hofstede 1989; Cohen 1991) such as the American. Indeed, in many high-context cultures such as the Japanese, an agreement is viewed as merely the beginning. Revision of the contract and re-negotiation are integral aspects of the negotiation process.
INSPIRE currently addresses the preparation, conduct and post-agreement re-negotiation aspects of the whole process, i.e.,
In addition to the above three major functions there is a range of smaller support features. Also, during the offer exchange the user may re-evaluate issues and options and modify his or her utility function.
4.1. Preparation
During the preparation phase each user individually performs activities that enable him or her to comprehend the problem, the main negotiable issues and options, the possible offers (packages) and criteria. This phase also involves specification of preferences leading to the construction of the userís utility function. These activities are depicted in Figure 2. While users may communicate with their partners, the communication at this stage is limited to unstructured messages.
The currently implemented technique for construction of utility functions is based on conjoint analysis, in which the utility of a given package is determined from the user's preference orderings over a set of factorially designed alternatives (packages) (Green 1974; Rangaswamy and Shell 1994). A hybrid (compositional as well as decompositional) approach is used and it comprises three steps:
where Rating(P) is the total utility of a package P, uij is the utility associated with issue i and option j, and xij is a binary variable indicating whether the given option is present in the package.
There is a large number of packages that could be presented, and we need some way of selectively presenting just a few packages for the user to rate, yet obtain reliable utility values. This is a problem in the design of fractional factorial experiments. One of the most compact and effect designs is the orthogonal design, in which the packages are chosen such that the X matrix is orthogonal. INSPIRE uses the information obtained in the issue and option ratings steps to select the set of orthogonal packages presented to the user for the package rating step. Given the ratings for these packages, the weights uij are computed that minimize the error terms using linear regression.
Since the utility of every possible option is considered explicitly, the utility function for a given issue can be non-linear. This is an advantage since people usually do not have linear utilities as they traverse a given range of values. By default, issues are assumed to have "discrete" options, i.e., only a small number of explicitly listed options are considered to be meaningful as outcomes of the issue. These are also called salient options. However, some issues can also be "continuous" in the sense that any intermediate value can be meaningful. This is typical of quantitative issues such as the price of some commodity. In such cases, the utility function within an issue is assumed to be piece-wise linear, i.e., linear interpolation (or extrapolation) is used to compute the utility of intermediate points between salient options.
4.2. The conduct of negotiations
The conduct of negotiations is divided into four standard stages namely climate-setting, presenting, mid-point bargaining, and closing. These four stages are not clearly distinguished in INSPIRE. However the systemís two modes of communication, that is structured offers and free-text messages allow the users to perform activities corresponding all the stages.
Offers have a predefined format, that is they contain names of the issues and options (issue values). While constructing or analyzing an offer, users automatically obtain its utility value. An offer may be accompanied with a message, which allows for argumentation and backing. Users may also send separate messages in order to, for example, set the climate, request explanations, or press their counterpart for a reply. These and other activities that can be performed during the conduct phase are listed in Fig. 3.
To support users in reviewing the negotiation and its dynamics, the system groups together all the past messages and offers, including utility values. In addition, a graph displaying negotiation dynamics is also available. It depicts all the offers made by both parties over time and the userís rating scale (see Fig. 11 for an example).
During negotiation users may review and revise their ratings, effectively updating their utilities, as indicated in Figure 3. We have observed that the graphical facility and offer scores become a focal point with some users and they tend to revise their preferences frequently, apparently with the objective of getting a satisfactory graph with high final scores. We speculate that this kind of preference migration may reflect the usersí cultural background.
4.3. Post-settlement re-negotiation
Once a compromise has been achieved during the conduct phase, INSPIRE checks it for efficiency (Pareto-optimality). This is the stage when the system acts as a mediator and takes into consideration the utilities of the two parties. Negotiation ends if the compromise is efficient. Otherwise the system computes efficient packages and displays several of them to both users. The displayed packages include those which increase one partyís utility alone, as well as the mid-point solutions.
The list of the post-settlement phase activities is similar to that of the conduct of negotiation phase and is displayed in Figure 4. The systemís additional activity is the computation, selection, and display of efficient offers. A notable distinction is that in the post-settlement phase, users cannot revise their preferences. This is because in this phase the system uses the preference information to determine and display efficient packages. If either party changes their utility structure, the current efficient solutions may become inefficient; in particular, the last-reached inefficient compromise may turn efficient, effectively terminating the negotiation in a way that would be considered unexpected by the counterpart. Apart from confusion, unilateral transformation of the efficient set under consideration can undercut the acceptability of the mediation process.
Apart from utility modification, users can perform the same actions during the post-settlement stage as during the conduct of negotiation phase.
5.1. Client-server decomposition
The traditional view of a negotiation (or group) support system is that of a desktop application: each user has one copy of the software on their personal computer, which communicates with the other users' copies over a network (typically a LAN), usually in synchronous mode (i.e., with both parties simultaneously logged on). Figure 5 depicts how INSPIREís process model, conceptualized as a negotiation support system, has been translated into its implementation structure as a Web application. The system uses the client/server model of distributed systems to partition the main components.
The connection between the two sides can be either direct or through a common server program. Conceptually there is no reason to require the server program unless the concept of a third-party conflict resolution service (or facilitator/mediator/arbitrator) is intrinsic to the group support methodology provided by the system. All services not involving a third party, e.g., preference elicitation, offer analysis and construction support, etc., can be implemented locally within each user's desktop application (and indeed it is desirable to implement such features locally for privacy and security reasons---the user's preference information, for example, should not be accessible to anybody else). Only the objects explicitly exchanged during communication (e.g., offers and messages) and information required by the third-party facilitator (e.g., preferences for Pareto-optimal analysis during post-settlement) need be transmitted outside the desktop. However, this neat conceptual partitioning based upon functionality (which is useful for presentation to a lay user) disappears when translated into the physical implementation of INSPIRE; it is replaced by a two-component: frontend and engine model (see Fig. 5), that more closely reflects the realities of current net-centric computing technology.
Three major factors affect the INSPIRE design:
One other factor that significantly influenced the design is the fact that the negotiations supported are asynchronous: since the two parties negotiating with each other typically reside in far away countries with different time zones, it is rare for both sides to be simultaneously logged on. Therefore, INSPIRE is designed to interact independently with each user, saving the state resulting from each user's actions in a form that can be retrieved when the counterpart logs on some time later.
Returning to Fig. 5, we see that while the functionality of the facilitator is implemented entirely in applications on the Web server side (as expected from the conceptual organization), the functionality of a desktop negotiation support system has been distributed between the browser on the user's desktop and the application programs on the INSPIRE server. However, examining the systemís design in terms of the "browser side" and "Web server side application" implementation components is too fine-grained to convey its modular structure. The appropriate higher-level abstraction is a division of the system into "frontend" and "computational engine" components.
5.2. The frontend and engine
The distinction between INSPIREís frontend and the engine corresponds roughly to the distinction between a traditional knowledge based system's visible user interface and invisible internal model base and reasoning engine. The frontend comprises dozens of HTML/JavaScript pages as well as Java applets. They reside on the INSPIRE site but are displayed or executed on demand on the user's host machine. The engine is a collection of programs in C++. These programs also reside on the INSPIRE site, but they are executed on the INSPIRE host itself and invoked via the Web server and the CGI protocol. The basic underlying mechanism that connects the two segments is the CGI protocol, but since this protocol is too primitive to directly handle negotiation-related objects, a new high-level tag-style mechanism has been provided on top of CGI, as explained in the next section.
One of the major goals that this architectural division addresses is that of supporting collaborative group development of INSPIRE by people with varying levels of familiarity with Web document presentation languages (HTML, JavaScript) and programming languages and protocols (C++, Java and CGI). One of the most important reasons for the explosive success of the World Wide Web is the ease with which anyone can develop Web pages. INSPIREís frontend, though sophisticated in organization and content, is amenable to design intervention by anyone with basic word processing skills and this has enabled people with wide ranging skills including behavioral scientists and decision analysts to directly modify and contribute to the design. This is possible because the frontend's interface to the engine resembles an extended set of HTML tags. The engine on the other hand comprises several thousand lines of C++, and along with the Java applets is accessible to development only by relatively experienced programmers.
All of INSPIREís implementation is object-oriented, and each piece of functionality listed within the two major components in Figure 6, is provided by a group of object classes that is loosely coupled with the rest of the system. Therefore each of them has been implemented fairly independently of the others. At the heart of the system lies one segment, labeled "methodologies" on which we focus here and in Section 5.4.
Figure 6. INSPIRE's main components
In this context, "methodologies" is an umbrella term for the sequence of activities performed by the users in conjunction with the system. This is the part of INSPIREís architecture where it looks at the state of the negotiation and dispatches a page from the frontend to the user's browser or performs a short sequence of computational activities and typically displays their results. This is obviously an important part of the system and implicitly controls the invocation of much of the other functionality of the system.
In general, users always need to know "What can I do next?" and the answer depends on the context and any rules provided by the system's designers. In other words we have to deal with states and contexts.
Based on the state of a negotiation, differing sets of activities are proposed to the user. Web pages are very good at representing and presenting context. So methodologies can be implemented by a dispatching system that determines which page to present based on a given situation. One of the principles learnt from the INSPIRE implementation is that a rule-based structure works surprisingly well for implementing methodologies. In fact the whole section comprises a series of if--then--else statements in C++ and stands out from the otherwise characteristically object-oriented code in the system.
This part of the architecture can be summarized as follows: pages in the frontend represent contexts; states are represented by (mostly Boolean) expressions in the engine. While the latter is obvious, it helps in understanding the former to observe that each page displays text and images describing (part of) the current situation and gives the user options (clickable links, prompts and other widgets) to perform only those activities that are relevant to the given context.
5.3. The coupling mechanism
The Web pages that comprise the user interface are generic in the sense that they can be reused for arbitrary negotiation cases. However, an unchanging Web page cannot reflect the dynamics of a negotiation. The dynamics inherent in the process manifests itself in two ways:
In short, Web pages have to be constructed on the fly by the programs in the engine to display dynamic or adaptive responses. However, hard-coding the contents of Web pages into the software would be a very poor approach to system development as it requires program recompilation for the slightest change, suffers loss of readability and maintainability, and prevents non-programmers from directly working on the contents of the pages.
It is clear that a means must be found to generate only those parts of a Web page that absolutely have to be computed by the engine (e.g., information that is specific to the users or the negotiation, utility values, etc.) and retain the rest as a proper Web page. The solution we found is to support new HTML-like tags in the Web pages that are understood by the INSPIRE programs. These special tags begin with <INEG> and end with </INEG>; for example <INEG>user_id</INEG> denotes the name of the user accessing the INSPIRE system. A Web page containing such <INEG> tags is called a dynamic page (because it adapts to the situation as explained in a moment), whereas a page without these tags is called a static or normal HTML/JavaScript page. Dynamic pages are not meant to be directly accessed on the Web via their URLs; rather, dynamic pages are meant to be processed by the INSPIRE engine before being displayed to the user. When any program in the INSPIRE engine pipes out a dynamic page, it substitutes each <INEG> tag by something appropriate (in the above example, the user's name). This simple "macro substitution" mechanism enormously simplifies the interface between Web page developers and program developers. All that is required to keep the user interface and the engine fully synchronized with each other is a well-designed catalogue of <INEG> tags and their semantics.
There are four categories of <INEG> tags in INSPIRE: identification, version management, history (state) recapitulation, and negotiation object embedding (e.g., offers, messages, rating tables). For example, <INEG>issue_rating</INEG> generates the HTML table used during the preparation stage to rate the issues <INEG>offer_construction_box</INEG> generates a table containing menu selections for each of the issues, by which the user can construct an offer; <INEG>engine_url</INEG> identifies the location (URL path) to the engine programs on the INSPIRE site, enabling them to be freely moved around.
5.4. The main components
The DSS architecture has traditionally been neatly divided into three components Dialog, Data and Model, i.e., the "DDM paradigm", with clearly defined roles for each component (Sprague and Watson 1997). We found this architecture restrictive and inflexible. The DDM concept does not sufficiently address the need for flexible use of multipleócomplementary as well as competitiveóanalytical methods. The traditional DSS architecture limits communication between the systemís components and also the componentsí communication with external systems in ways independent of any pre-specified control mechanism. In the development of INSPIRE an effort was made to construct software objects that can be used and reused in different conditions and in different configurations. Moreover, we aimed at a system which in future can be expanded and use new and additional components. Thus, we were looking for a system which allows for:
The main component of the INSPIRE system that is instrumental in obtaining the above characteristics is the Methodologies and control (M&C) component.
Figures 7-8 depict the INSPIRE architecture in two main modes: the support of a single party in the negotiation (individual support mode) and the mode in which the system determines and presents efficient solutions (joint support or mediation mode). At a high level, as discussed in Section 5.2, the architecture comprises the engine and the front-end. Currently, the frontend functions as a Dialog component with limited computational capabilities. However, with the introduction of Java applets and coupling with local client applications, the frontend functions will encompass some of the functionality of the Model and Data components.
The Model component is represented by several separate objects used for preference elicitation, utility and history construction, and also other objects which are activated by the M&C component at different stages of the negotiation process. The M&C component of INSPIRE decides on the systemís behavior; it takes information from the frontend (Dialog) component.
The negotiation process supported by INSPIRE can take several different forms. Negotiations can be sequential, that is parties discuss about one issue at the time, parallel, and mixed. They also can be conducted with or without messages. Other options, which are not yet implemented, include the addition and/or deletion of issues and options, addition of separate objects (e.g., price lists, pictures) to the messages, voice and video, etc.
A major problem with the DDM is that it doesnít have a plug-and-play philosophy. The system should be viewed and implemented as a collection of loosely-associated smart objects (the DSS as toolkit metaphor). DSS and NSS design always must be open-endedóamenable to change, but most DSSs are weak in providing a methodology for change. We view system change as an upgrade in the attributes or behaviors of individual objects (occasionally, modules), independently of other objects or modules, rather than one synchronized change across the whole system.
As an example, consider the problem of making INSPIRE qualitative-information capable, i.e., able to handle missing information such as incomplete preference ratings. This can be implemented by a series of independent changes: upgrade the PreferenceStructure object classes, which deal with storage and retrieval of preferences, to accept holes in the data; then upgrade the UtilityFunction objects to interpolate for missing preferences instead of asking for additional information; then upgrade post-settlement algorithmic classes to deal with interval ranges; then experiment with alternative preference elicitation strategies that optimally use qualitative information. A good (sufficiently general and abstract) set of object interfaces and loose coupling between components are critical to achieving this kind of open-endedness.
One of the main characteristics of the INSPIRE system is its expandability. Figure 7 illustrates the loosely coupled architecture of INSPIRE. This is also depicted in Figure 8, where some objects are introduced and other removed due to the change of the systemís mode of operation (from individual support to mediation).
Figure 7. INSPIRE's architecture: Individual support mode.
It may appear that Figures 7 and 8 depict two different (though similar) systems. This is of course not the case, but reflects the ability to use the same components to perform similar functions but for different users, or the use of systemís other components which previously were not available. For example, the same engine can be used by each user separately (in the individual mode) to rank offers or it can be used in the mediation mode and order offers using both utility functions.
In the mediation mode the system can be viewed as comprising three Engine and two Frontend entities. This mode is entered only when the users achieved a compromise. The components involved in these activities are depicted in Figure 8. Note, however, that the smaller and larger components with the same names are "clones" of the same program but used for different users or different other components. At this stage, the system verifies the efficiency of the compromise usingófor the first timeóthe utilities of both simultaneously. If the compromise is non-efficient, the system asks users whether they want it to search for efficient solutions. Upon obtaining a positive answer from both users, the system determines efficient solutions and selects and displays five solutions for each party.
Once the negotiation is concluded, INSPIRE informs the users and asks them if they agree to provide their counterparts with their negotiation dynamics graphs and other graphs that describe the negotiation dance (Raiffa 1982). The graphs are displayed only if both parties agree.
6. An example from a real negotiation
The illustrations in this section are intended to provide the flavor of an INSPIRE session. An example of one of the user activities during the pre-negotiation stage is given in Fig. 9. Having read a description of the caseówhich specifies the four issues being negotiated (the purchase price of the bicycle parts under negotiation, the delivery and payment schedules, and the return policy for defective parts) and the options available for each issueóthe user is requested to compare each of these options against the others and specify their relative importance. The ratings supplied by the user during this and subsequent steps are used to construct a utility function.
During the conduct negotiation stage users construct offers, analyze counter-offers, send and receive messages, and review the negotiation dynamics. Figure 10 is a snapshot of the offer construction screen; it illustrates how the users can communicate either by plain messages or structured offers, and how the score attached to an offer helps select a good offer.
Figure 11 is a graph of the dynamics of the negotiation process; such graphs are generated on the fly throughout the negotiation, and the one in the figure was generated at the very end of the negotiation. The user names displayed on the graph ("misty" and "ldias") are pseudonyms adopted as INSPIRE login names by the users in order to protect their anonymity. The little numbered triangles denote offers; the X axis shows the time at which each event occurred and the Y axis represents the score associated with an offer. Note that though the offers of both parties are shown, only a single utility function (mistyís) has been used to evaluate all of them. (For two reasons: each partyís preference information is private and unavailable to the other side; nor is it meaningful to make interpersonal utility comparisons).
Figure 11. Graph of negotiation dynamics.
Note that while misty makes concessions right till her fourth offer, ldias makes concessions until his third offer; then he appears to make a reverse concession between his third and fourth offer. This worries misty into thinking that ldias has hardened his position, and she quickly accepts his fourth offer. In truth, ldias did not make a reverse concession; his graph (not shown) indicates that his fourth offer was indeed worse for himself than his thirdófrom his point of view! In other words, when he thought he was giving away value to misty in order to reach a compromise, he actually appeared to her to grasp value and take a tougher stance. This pattern is very common in the negotiations observed via INSPIRE and underscores natural misunderstandings as a cause of negotiation failure.
As a consequence of misty ës accepting ldiasís fourth offer which was worse for both of them than his third, INSPIRE automatically enters the post-settlement phase and recommends several better compromises. Ldias and misty continue negotiating, making their fifth offers each, and finally reach an optimal compromise as shown in the figure.
The INSPIRE system has been developed as a tool for research and also used for training. Because of the cross-cultural research function we do not plan to change the system; its specification has been frozen so that we can compare the conduct of negotiation by different users and at different times. However, our experience with this system, both in terms of the research output from analyzing international negotiations conducted through INSPIRE and the experience gained from building and deploying a negotiation support system, as well as the feedback from our users, motivate and inform the design of a new system, INSS (the InterNeg Support System), which is currently under development. INSS will have all the features of INSPIRE and many more.
Using Java applets, users will be able to add new values (options) to the negotiation issues. If, for example, a user begins with five salient values for the price, she may add new price values during negotiation. New values will require to recalculate the utilities for both parties and we will use a simple approximation of piece-wise linear utilities. However, both the user and her counterpart will be able to modify their preferences if they wish to do so. Further, we plan to introduce an option to add values for the discrete issues. In this case both users will have to specify the relative preference for the new value.
Issues themselves may be introduced or removed dynamically during the course of the negotiation. In particular, allowing the two negotiators to dynamically define the set of negotiable issues at the beginning requires (1) value focused analysis to be performed as a pre-negotiation step by the individual negotiators, and (2) a new initial protocol during which the issues are proposed and negotiated by all the parties.
We also plan to enhance the pre-negotiation phase with support for specification of reservation levels and the best alternative to a negotiated agreement (BATNA). Once these two constructs have been elicited from the user, the system will flag any offer or counteroffer that violates either the BATNA or any of the reservation levels. Further planned enhancements include context-sensitive advice; users will be able to obtain information explaining their opponentís behavior, assistance in interpreting offers, suggestions regarding available strategies, etc.
An important factor that we have observed as influencing the negotiation process in INSPIRE is the availability of means for exchanging different kinds of structured objects during communication between the negotiators (e.g., formally specified offers versus free-text messages). INSS will be enhanced with facilities to transfer price lists, balance sheets and other multimedia documents such as pictures and video clips.
Adler, N. J. (1983). "Cross-Cultural Management Research: The Ostrich and the Trend." Academy of Management Review 8(4): 226-232.
Adler, N. J. and J. L. Graham (1989). "Cross-Cultural Interaction: The International Comparison Fallacy?" Journal of International Business Studies: 515-537.
Alemi, F., P. Fos, et al. (1990). A Demonstration of Methods for Studying Negotiations Between Physicians and Health Care Managers." Decision Sciences 21(3): 633-641.
Andriole, S. J. (1993). "Interactive System Support for International Negotiations." Theory and Decisions 34(3): 313-328.
Balakrishnan, P. and J. Eliashberg (1995). "An Analytical Process Model of Two-party Negotiations." Management Science 41(2): 226-243.
Bazerman, M. A. and M. A. Neale (1991). Negotiator Rationality and Negotiator Cognition: The Interactive Roles of Prescriptive and Prescriptive Research. Negotiation Analysis. H. P. Young. Ann Arbor, MI:, The University of Michigan Press: 109-130.
Bond, M. H. (1987). "Chinese Values and the Search for Culture-Free Dimensions of Culture: The Chinese Culture Connection." Journal of Cross-Cultural Psychology.
Bui, T. (1994). "Software Architectures for Negotiator Support: Co-op and Negotiator". Computer-Assisted Negotiation and Mediation Symposium, Harvard Law School, Cambridge, MA.
Carnevale, P. J. (1995). Property, Culture, and Negotiation. Negotiation as a Social Process. R. M. Kramer and D. M. Messick. Thousand Oaks:, Sage: 309-323.
Cohen, R. (1991). Negotiating Across Cultures. Communication Obstacles on International Diplomacy. Washington, DC:, United States Institute of Peace Press.
Darling, T. A. and J. L. Mumpower (1992). "Simulating Process and Outcome for Two-party Contract Negotiations." Control and Cybernetics 21(1): 151-184.
Fang, L., K. W. Hipel, et al. (1993). Interactive Decision Making. The Graph Model for Conflict Resolution. New York:, Wiley.
Faure, G. O. and J. Z. Rubin (1993). Culture and Negotiation. The Resolution of Water Disputes. Newbury Park, CA:, SAGE.
Fisher, G. (1980). International Negotiation. A Cross-Cultural Perspective. Yarmouth, MA:, Intercultural Press Inc.
Fisher, R. and W. Ury (1983). Getting to Yes Negotiating Agreement Without Giving In. New York:, Penguin Books.
Fraser, N. M. (1992). "Conflict Analysis Support for International Negotiations". Support of International Negotiation: Implications for Application, Laxenburg, Austria: IIASA.
Graham, J. L., A. T. Mintu, et al. (1994). "Explorations of Negotiation Behaviors in Ten Foreign Cultures Using a Model Developed in the United States." Management Science 40(1): 72-95.
Green, P. E. (1974). "On the Design of Choice Experiments Involving Multifactor Alternatives." Journal of Consumer Research 1: 61-68.
Gulliver, P. H. (1979). Disputes and Negotiations: A Cross-Cultural Perspective. Orlando, FL:, Academic Press.
Hofstede, G. (1989). Cultural Predictors of Negotiation Styles. Process of International Negotiations. F. Mautner-Markhof. Boulder, CO:, Westview Press: 193-201.
Holsapple, C. W., H. Lai, et al. (1994). Analysis of Negotiation Support System Research.
Ikle, F. C. and N. Leites (1962). "Political Negotiation as a Process of Modifying Utilities." Conflict Resolution 6(1): 19-28.
Jelassi, M. T. and B. H. Jones (1988). Getting to Yes with NSS: How Computers Can Support Negotiation. Organizational Decision Support Systems. A. M. M. R.M. Lee, and P. Migliarese. Amsterdam:, North-Holland: 75-85.
Kersten, G. E. (1985). "NEGO - Group Decision Support System." Information and Management 8(5): 237-246.
Kersten, G. E. (1995). "Simulation and Analysis of Negotiation Processes: The Case of Softwood Lumber Negotiations". Proceedings of the 28th Hawaii International Conference on Systems Sciences, Maui, Hawai: Los Alamitos, CA: IEEE Computer Society Press.
Kersten, G. E. (1997). Support for Group Decisions and Negotiations. An Overview. J. Climaco. Heilderberg:, Springer Verlag.
Kersten, G. E. and T. Szapiro (1986). "Generalized Approach to Modelling Negotiations." European Journal of Operational Research 26(1): 142-149.
Kremenuk, V. A. e. (1991). International Negotiations, Analysis, Approaches, Issues. San Francisco, CA:, Jossey-Bass.
Lax, D. A. and J. Sebenius (1986). The Manager as Negotiator. New York:, The Free Press.
Lockhart, C. (1979). Bargaining in International Conflicts. New York, N.Y.:, Columbia University Press.
Nyhart, J. D. and D. K. Samarasan (1990). "The Elements of Negotiation Managment". International System Dynamics Conference, Boston, MA.
Pruit, D. G. (1981). Negotiation Behavior. New York:, Academic Press.
Raiffa, H. (1982). The Art and Science of Negotiation. Cambridge, MA:, Harvard University Press.
Rangaswamy, A. and G. R. Shell (1994). "Using Computers to Realize Joint Gains in Negotiations: Toward an Electronic Bargaining Table". Computer-Assisted Negotiations and Mediation Symposium, Harvard Law School, Cambridge, MA.
Roth, A. E., V. Prasnikar, et al. (1991). "Bragaining and Market Behaviour in Jerusalem, Ljubljana, Pittsburgh, and Tokyo: An Experimental Study." The American Economic Review 81(5): 1068-1095.
Sebenius, J. K. (1992). "Negotiation Analysis: A Characterization and Review." Management Science 38(1): 18-38.
Sprague, R. H. and H. J. Watson (1997). Decision Support for Management. Upper Saddle River, NJ:, Prentice Hall.
Teich, J. E., H. Wallenius, et al. (1994). "Advances in Negotiation Science." Transactions in Operational Research 6: 55-94.
Winham, G. R. (1991). Simulation for Teaching and Analysis. International Negotiations, Analysis, Approaches, Issues. V. Kremenuk. San Francisco, CA:, Jossey-Bass: 409-423.
Young, H. P. e. (1991). Negotiation Analysis. Ann Arbor:, The University of Michigan Press.
Zartman, I. W. (1993). "Decision Support and Negotiation Research: A Researcher's Perspective." Theory and Decision 34(3): 345-351.
