The path for eNegotiation systems is typically straight across the top.  This means that a confirmed dispute goes through a process of negotiation[7] until a resolution is achieved.  The path for parties using a mediation[8] system might go across the middle, while the path for an arbitration system[9] would be across the bottom.

A comprehensive system encourages parties to start with negotiation. If the dispute is not settled with negotiation, the process can progress to mediation and finally recommendation or arbitration, until the dispute is resolved or failure to resolve the dispute is reported. Some systems are designed to resolve disputes that occur online while others are useful for any type of dispute, regardless of where it originated. Parties can resolve their dispute exclusively online or use a process that may also include face-to-face meetings.

Short general descriptions for each of the selected systems are given in Appendix A. Detailed technical descriptions of these systems in table form are contained in Appendix B. The information in the table has been compiled on the basis of information available on the Internet and from recent email interviews with representatives of some of the systems.[10]

We believe ODR systems face five main challenges as they attempt to present an effective medium for online dispute resolution:

1.      Problem representation

2.      Preference elicitation

3.      Effective communication

4.      Neutrality provision

5.      Degree of automation

The following section briefly discusses each of these challenges. Automation, related to all of these challenges, is a common thread, and is subsequently summarized at the end.

Challenges for the Construction of Online Dispute Resolution Systems

Problem Representation

Earl Nightingale once said that a well-defined problem is already half solved.  Designers of ODR systems do well to keep this in mind as they decide how to represent disputes. They must choose whether to provide a highly structured or a less structured environment.  Less structured systems simply identify the parties and then allow them to converse in free form. A human mediator would typically attend in this situation.

Systems such as Family_Winner, Inspire and SmartSettle, which provide a high degree of analytical support to enhance negotiation, provide parties with a highly structured environment. In a highly structured environment, in addition to identifying the parties, the issues are identified along with associated variables. In structured environments the variables may be measured or optioned, private or published and related to each other with constraints or formulas. Highly structured environments like Negoisst and SmartSettle also keep track of versions.

The simplest ODR problems involve two parties and a single measured issue, usually monetary (e.g., insurance claims). Of the systems represented in this study, CyberSettle and Mediation Room offer blind bidding as a structured process for solving this simple single-issue problem.

The most complex problems are those with many issues, and many parties with conflicting objectives (e.g., environmental disputes). In more complex problems, interaction among the variables requires non-trivial mathematical modeling that can only be performed with sophisticated systems. SmartSettle is the only system with complex problem modeling capabilities. SmartSettle also offers a multivariate type of blind bidding that can be applied to problems with any number of issues and can even be used in complex multiparty problems.

As users’ experience with ODR systems is expanded, designers of ODR systems will become more adept at problem representation. This will be further enhanced if such experience is used to build databases of good solutions. ECODIR uses predefined structured forms to assist and expedite problem representation. Such reuse of good information from past solutions will make the process increasingly more efficient.

Effective Communication

Effective communication is an essential requirement for successful dispute resolution. There is no one method of communication that works better than all the others. The best communicators use a balance of all the various methods available to them. When using a particular ODR system, users are constrained by its structure. A highly structured environment may have the advantage of leading parties quickly to the ideas for which they are searching. However, if there is little flexibility, parties may be prevented from effectively expressing themselves.

Currently, the primary default means of communication for all ODR systems is the written word, in an asynchronous mode. The Encyclopedia of Educational Technology (Hoffman, 2004) gives a detailed explanation of the differences and the pros and cons between asynchronous and synchronous methods of communication. Each method has clear advantages.

Practical and cost effective video as synchronous communication will eventually be a great asset to online communication.  Although some research systems are already experimenting with video, none of the systems represented by the analysis in this paper have implemented any type of video component. A voice component is also practical and very useful. The Mediation Room features built-in voice conferencing.  Several others (such as ECODIR and SmartSettle) encourage the use of a mix of systems including the telephone to assist resolution of disputes. The simplest means of synchronous communication is instant messaging. A number of systems support this functionality.

Except for the very simplest of problems, live human intervention is still a very important component for effective online communication. eBay, one of the largest warehouses of disputes, relies heavily on highly capable independent ODR providers like SquareTrade to help resolve disputes where  human involvement is invaluable (Rule, 2004).

Before ODR becomes mainstream, providers of ODR and ADR services need a technique for communicating with each other.  This means that the systems need some fundamental interoperability, particularly when exchanging data between institutions such as courts or brokerage firms that require ADR. One approach being used for more effective communication with the wider system is called ODR XML. XML, an extension of HTML that was introduced in 1990, is now a globally recognized open standard used to tag information on web pages, forms and free flowing text documents. ODR XML is simply an application of XML to ODR, i.e., creating standards for ODR and ADR systems to talk with each other (Keane & Miller-Moore, 2004).

The name “OdrXML” originates in a project created in 2002 by the European Union’s Center for Joint Research (JRC).  JRC developed a data dictionary and schema for online consumer disputes and dubbed it OdrXML (see www.econfidence.jrc.it).  The goal of the OdrXML Technical Committee is to harmonize the work in different countries, and build on the work in Europe so it can interact with applications developed in the US and Australia. 

As one can see from the table in Appendix B, none of the listed ODR systems are ODR XML compliant. In reality, it may be premature to even list such a column in this table since accepted standards do not yet exist.

Preference Elicitation

A simple unstructured online location in which parties communicate and negotiate is the simplest format in which preferences are represented and exchanged in ODR systems. This format is useful for human mediators but becomes difficult when eliciting preferences, since the lack of structure prevents the identification and measurement of preferences. In order to elicit preferences with sufficient precision for effective analytical support, ODR systems must provide structure to the problem representation (e.g., with forms as is performed by ECODIR).

An accurate elicitation of preferences is probably the most difficult aspect of the dispute resolution process when attempting to provide automated decision support. However, when preferences are accurately represented, automating the rest of the process becomes easier.

Quite a few systems elicit user preferences and offer analytical support. SmartSettle offers the most comprehensive support with features including preferred outcomes, bargaining ranges, satisfaction functions, tradeoffs, issue importance and package ratings. SmartSettle has demonstrated the benefits of analytical support extensively in research and is now beginning to do so in application to real cases. An eNegotiation System with comprehensive analytical support has the potential to assist negotiators in two ways; in reaching their agreement more quickly, and in reaching a better agreement, discovering value that would otherwise remain on the table.

The data in Table 2 shows our selected systems organized according to analytical support, summarizing the corresponding columns in Appendix B.

Table 2 – Analytical Support provided by ODR Systems

Neutrality Provision

The majority of ODR systems use human involvement. Thus, ODR providers must take steps to instill confidence in users that the service they receive will be as unbiased as possible. ECODIR, for example, uses humans for selection of neutral independent mediators and plans an oversight panel to assure and monitor impartiality of operations.

Neutrality is most important to systems that provide arbitration. The greatest challenges to such systems are predictability and proportionality (Odams de Zylva, 2001). Users need to be able to trust the system to consistently deliver a reliable fair result. At the same time, the arbitration system needs to deliver the service in a cost-efficient manner that can be scaled up to meet high volume demands. The involvement of live professionals to solve small problems quickly becomes prohibitively expensive if the time they spend is worth more than the benefit of the service. Automation is necessary to counter this problem, especially for small problems.

It is important that ODR processes are fair, impartial and transparent. eBay addresses this by constantly evaluating its efforts against the most prominent standards that have been issued in the ODR field (ABA, ICC, BBB, GBDe, Consumers International, etc.) (Rule, 2004). The quality of neutrality can also be improved with automation. 

Related to neutrality provision is neutral site security. Users of the system need to be confident that their private information will be kept confidential. If a system uses browsers for the user interface, it must also implement a secure protocol for transmissions.

Degree of Automation

Automation refers to functions that are performed without human assistance... eBay has more than 100 million users leaving more than a million pieces of feedback a day (Rule, 2004).  Because of this volume much of their dispute resolution effort is devoted to automated, fourth party ODR tools and techniques.  Through contextual, user-friendly intake and convening forms and rule-based case routing processes they help parties define and focus on the specific issues that need to be resolved.  Through intuitive, integrated direct negotiation tools they can coach parties in effective communication. 

In one sense, any online system that parties can use without human intervention, even simple email exchange, would be considered automated. A commonly automated function is that of informing parties of activity in the case. Most systems provide a means of doing this within the ODR system interface. Many systems (e.g., ECODIR) will also automatically notify their users of new activity by sending them an email message.

Inspection of the table in Appendix B reveals that high automation is generally associated with low comprehension and vice versa. This is expected, due to the difficultly of automating for complex situations. The only existing ODR systems that are totally automated are those that mainly deliver information or those that offer a blind-bidding process for simple single-issue two-party cases.

The challenge lies in providing automation that is sophisticated and intelligent. The ideal ODR system would be one that provides all the functions of any intervenor and is fully automated, supported by the ultimate artificial intelligence that could not be improved with any kind of human intervention in any type of problem. In fact, we can expect that such systems will actually outperform human beings, bringing levels of fairness and efficiency to ODR and expanding to improved negotiation of all kinds of problems.

References

Abernethey, S. 2004. Building Large-Scale Online Dispute Resolution & Trustmark Systems. Proceedings of the UNECE Forum on ODR 2003. http://www.odr.info/unece2003/pdf/Abernethy.pdf visited on 2004 July 10.

Bellucci, E. and Zeleznikow, J. 2001.  Representations for decision making support in negotiation. Journal of Decision Support. 10(3-4).

Black, H. C. 1990. Black's Law Dictionary, West Publishing Company, St. Paul, Minnesota

Bordone, R.C. 1998.  Electronic Online Dispute Resolution: A Systems Approach-Potential, Problems and a Proposal, Harvard Negotiation Law Review: 175-211.

Conley Tyler, M. 2004. State of ODR 2004, in Conley Tyler, M. and Katsh,  E. (Eds.), Proceedings of the Third Annual Forum on Online Dispute Resolution, Melbourne, 2004 July 5-6.

Folberg, J. & Taylor, A. 1984, Mediation: a comprehensive guide to resolving conflicts without litigation, Jossey-Bass social and behavioral science series, Jossey-Bass, San Francisco.

Hoffman, R. Synchronous/asynchronous Delivery, Encyclopedia of Educational Technology, San Diego State University. Also at http://coe.sdsu.edu/eet/Articles/synchasynchtech/index.htm visited on, 2004 June 29.

Jennings, N.R., Faratin, P., Lomuscio, A.R., Parsons, S., Wooldridge, M.J. & Sierra, C. 2001. Automated Negotiation: Prospects, Methods and Challenges, Group Decision and Negotiation, 10 (2): 199-215.

Katsh, E. & Rifkin, J. 2001, Online dispute resolution: resolving conflicts in cyberspace., Jossey-Bass, San Francisco.

Keane, J. & Miller-Moore, D.  2004.  Linking Information Technology and Dispute Resolution, Dispute Resolution Journal, 2004 February – April (Vol. 58, No. 1), a publication of the American Arbitration Association, 335 Madison Avenue, New York, NY 10017-4605.

Kennedy, G., McMillan, J. & Benson, J. 1984, Managing negotiations, 2nd ed., Hutchinson Business, London

Lodder, A. & Zeleznikow, J. 2005.  Developing an Online Dispute Resolution Environment: Dialogue Tools and Negotiation Systems in a Three Step Model.  To appear in the Harvard Negotiation Law Review.

Odams de Zylva, M. 2001. Trusted Online Dispute Resolution – The Three P’s model, Word and Bond Policy Paper, 2001 March 19

Ross, G. 2001, Online Dispute Resolution - Why Settle for Less?. Internet Newsletter for Lawyers, 2001 May/June. Also at http://www.venables.co.uk/n0105odr.htm visited on, 2004 June 29.

Ross, H. L., 1980, Settled Out of Court, Aldine

Rule, C. 2004, eBay Director of Online Dispute Resolution, Email communication, 2004 June 8

Thiessen, E. M. & McMahon, J. P. 2000. Beyond Win-Win in Cyberspace. Ohio State Journal on Dispute Resolution, 15: 643

Williams, G.R., 1983, Legal Negotiation and Settlement, West Publishing Co., St. Paul, Minnesota

Appendix A: Short Descriptions of Selected Systems

CyberSettle

CyberSettle is a totally automated double blind bidding system that is targeting mainly insurance claims. See their website (www.cybersettle.com) for more information.

eBay UPI (Unpaid Item ODR Process)

eBay’s Unpaid Item ODR process is eBay’s new system for resolving disputes around items that are successfully won or bought on the eBay site but where the seller has not yet received payment.  The UPI process is entirely automated (except for an optional appeals component at the end of the process) and it consists of a form-based intake and diagnosis, leading into a direct negotiation message thread, where buyers and sellers can communicate on the eBay site without using email.  Different outcomes are available based upon information provided by the parties during the negotiation and data in eBay’s records, but the participants choose the final outcome.

ECODIR

ECODIR (“Electronic Consumer Dispute Resolution”), an European Union backed project run from University College Dublin, Ireland, provides a secure private web-space in which the parties to a dispute can negotiate a mutually acceptable solution using ECODIR’s structured negotiation process. The parties can call for the help of a qualified ECODIR mediator to assist them in solving their dispute if they both wish. Communication is asynchronous and within the website alone - the system notifies the parties by e-mail of new communications.

Family_Winner

Family_Winner uses Principled Negotiation to support trade-off and compensations strategies in a negotiation.  An Issue Decomposition Hierarchy identifies sub-issues. Input is in the form of issues and numerical utilities, which represent the importance disputants’ place on issues.  Family_Winner output consists of a list of allocations the system has made.  Family_Winner’s uses Trade-off Maps (a variant of Constraint Diagrams) to represent trade-off opportunities inherent in the issues of a dispute.  The system uses empirical formulae to allocate rewards for each issue.

Inspire

Inspire is a research system that is easy to use for simple negotiations in which the structure is already provided. Preferences may be represented with satisfaction graphs. Optimization is used to generate suggestions after an agreement is reached. Parties then resume negotiations about whether to accept one of the improvements.

Mediation Room

www.TheMediationRoom.com is a flexible platform enabling online any form of mediation/arbitration and comprising discussion areas, each with rights of access, writing and/or reading strictly controlled. The neutral can, in one area, discuss matters with one party and, elsewhere, hold discussions with everyone.

Blind bidding and on-screen controlled voice conferencing are available, the latter whilst reviewing and adding to the text discussions. Neutrals can use www.TheMediationRoom.com or have a version configured to fit the dispute whether commercial, family or consumer complaint.

Negoisst

Negoisst is an electronic negotiation support system that covers semi-structured communication support as well as document management and analytical decision support. The key characteristic of the system is the integration of these components. Dyadic negotiations are understood not only as a communication process but also as the joint construction of a document, namely the contract. Negoisst generates XML contracts automatically based on the offers made and maintains the relationship between messages sent and contract versions.

notgooodenough.org

NGE provides customers with an online space to seek advice, share experiences and learn accordingly. NGE also uses the media - of which its website is one form - to highlight gripes that are posted on the site. This is done on regular radio spots, TV and in the print media. NGE provides companies with access to consumer feedback from the NGE site. This allows organisations real time access to what customers are saying about them and their competitors.

Scenario Builder

Scenario Builder is a generic application that can quickly create an ODR application (among many other types) by capturing important knowledge of the process structure. This simple yet powerful system automatically generates webpages for an Intranet or Website to guide people through that process. An example of an ODR system created with Scenario Builder is DisputeInfo. DisputeInfo is an artificial intelligence system that can give standard advice to disputants about many types of community disputes. The data given in Table 1 represents the type of system that is possible to build with Scenario Builder.

SmartSettle

SmartSettle is an eNegotiation system that empowers negotiators to find better agreements in a shorter time, reducing cost and stress in simple or complex cases. It clarifies tradeoffs, helping people understand how they become satisfied on all types of issues while a patented neutral server keeps private information confidential and secure. The SmartSettle Internet network connects multiple parties located anywhere in the world and uses optimization to quickly transform conflicting objectives into fair and efficient solutions.

SquareTrade

SquareTrade offers companies like eBay a family of services and technologies from creating trust to resolving conflict. If there is a dispute, SquareTrade offers a structured process that proceeds from direct negotiation to mediation and finally an optional recommendation for a resolution if the dispute is not resolved sooner. Trust is build with the SquareTrade Seal program. If a customer finds an item on eBay there is greater trust if they see the SquareTrade Seal. Conflicts are easier to resolve in an atmosphere of greater trust.

Word & Bond

Word&Bond is an online dispute prevention and claims management system for all types of eCommerce disputes. The system appoints an i-arbitrator who is required to declare his/her independence of the parties. Once started with the process, neither party can resort to court. Rulings are made within 14 days and sent to both parties by e-mail in the first instance.  The original certified i-award is sent on thereafter. The claimant/defendant, then has 21 days to satisfy the requirements of the ruling.

Appendix B: Technical Descriptions of Selected ODR Systems

(Explanation of categories are listed below the tables.) 

Explanation of Categories in Above Tables

Client Type (fat, thin): Fat clients require software installation or downloading of a plug-in. Thin clients require only a browser.

Resolution Modes (info, blind bidding, plain negotiation, mediation, arbitration): Five categories are used to describe the resolution modes of the various systems described in this table. Info represents the function of simply providing information. The next two categories are negotiation modes; either blind bidding or plain negotiation, where negotiation may be facilitated or not. The last two modes are mediation and arbitration. Mediation is defined here as neutral intervention for the purpose of easing any friction that may be preventing parties in a negotiation or dispute resolution process from coming together, e.g., making suggestions to solve impasse if necessary. The function of facilitation is one of simply guiding the process, leaving as much control as possible in the hands of the negotiators. A facilitator may reflect on the comments of participants to ensure that all viewpoints are heard and understood. S/he helps achieve closure and consensus through a variety of techniques, including brainstorming and creative problem solving. The main difference between mediation and facilitation is that a facilitator offers no suggestions or any other content to the process.

Info (yes, no): This category indicates whether or not the system provides significant useful information that can help parties resolve their dispute. No means that the system assumes that parties come to the system with enough knowledge or must rely on other systems to provide that information.

Blind bidding (single area or multiple areas, univariate or multivariate, no): Single area or multiple areas refers to the number of blind bidding areas that can be associated with the same case. Univariate or multivariate refer to the number of issues or variables that can be handled simultaneously within one blind bidding area. Univariate blind bidding is the common form of double blind bidding that is offered by many ODR providers. In this case, the offers and demands are blind and the result is determined by splitting the difference if the final offer and demand are close enough according to some pre-determined formula. Multivariate blind bidding involves negotiations with packages comprised of two or more variables (issues). In multivariate blind bidding the acceptance of the package is blind.

Plain Negotiation (facilitated, auto, both, no): Options in this category describe whether or not parties using the system can negotiate directly with each other making plain proposals as to how to resolve the issues, i.e., where the proposed decision values are known immediately (as opposed to blind bidding).

Mediation (human, auto, both, no): Human means that mediation is provided with human intervention. Auto means that mediation functions (e.g., suggestions) are fully automated without requiring human intervention. Both means that the system is able to function either way. No means that no mediation functions are provided.

Arbitration (binding, recommendation, informal, no):  Binding means that binding arbitration is build into the process. Recommendation means that non-binding arbitration is built into the process as an optional step. Informal means that parties have the option of determining what type of arbitration will be part of the process.

Parties (two, multiple): This criteria indicates whether the system is primarily designed to handle two-party or multiparty negotiations (not including the neutral).

Issues (form, number): We are concerned here with the number of issues that can be handled by the system and the form in which they are presented to the user.

Form (free, structured, both): A free form area is like a chat room. Structured means that a structure is built or provided to identify and define separate issues.

Number (single, multiple): Most types of blind bidding systems can handle only one single issue at a time. Most other types of systems can simultaneously handle multiple issues.

Document Management (formal, free, filing, all, no): Filing refers to the storage and organization of documents of various formats within the system. Formal refers to a structured process in which the parties are guided through the process of constructing their contract as it evolves through various versions. Free refers to a system in which the parties determine their own process of constructing the contract. All refers to a system that provides all the functionality described here.

User Communication: This category is concerned with functions that help users communicate with each other.

Timing: (synchronous, asynchronous, either): This category describes the timing of messages between the participants. The default mode of communication on the Internet is asynchronous. Synchronous means that both parties need to be present at the same time (like a telephone without voice mail).

Change Notification (email, interface, both, no): Method by which parties are notified of new information. Interface means providing the user with a message on the screen (e.g., popup).

Voice Conferencing (built-in, phone, no): Verbal communication within the system.

Video Conferencing (yes, no): Transmission of images of the person speaking.

Caucus Area (yes, no): Private discussion area within the system for members of one party or between a party and the neutral.

Offline (automated, manual, no):  Storage and processing of information locally from face-to-face meetings or when not connected to the Internet and later synchronization with the server.

Neutral Site Security (encryption strength, yes, no): Security of most systems depend on browser functionality, which typically offers a maximum of 128-bit encryption. If an ODR site offers secure document transmission, the URL on the website will begin with HTTPS://. The additional "S" on the end of the familiar HTTP indicates a secure channel to the server. Every secure page should be secured with a digital certificate. This may be shown via a "site certificate" that sits on the secure pages. You may view this certificate by clicking on the image of the closed lock or the solid key on the bottom bar of your browser window. A small frame displaying site security information will appear. If a particular ODR system offers a higher grade of encryption, it is specified here. A system dependent on plain email would be an example with no security.

Analytical Support: Analytical support is functionality that supports the representation and analysis of preferences provided by the parties.

Complex Modeling (yes, no): Complex modeling refers to problem simulation with mathematical formulas that represent constraints or specified relationships among issues.

Tradeoff Analysis (yes, no): Preference analysis and representation of relative importance among the issues.

Optimization (yes, no): Generation of suggestions based on the preferences of parties.

ODR XML Compliant (yes, no, coming): This category indicates whether or not the system exports or publishes case information using ODR XML standards readable by other systems. It is expected that the answer will be universally no here because standards do not yet exist. Coming means that implementation is contemplated as soon as standards are established.

Patented (yes, no, pending)

Automation (%): The value in this column is an estimate of the degree of automation. It is based on the subjective judgment of the author based on information provided about the subject system. The degree of automation is represents the replacement of human intervention in terms of time spent. For example, if a system is 100% automated, then trained users are able to use the system without any intervention of a live human in all resolution modes that it is built for. If a system was only 50% automated, then it is estimated that automation saves 50% of intervenor time compared to systems providing the same functionality that are not automated.

Comprehension (%): The value in this column is a rough estimate of how comprehensive the subject system is in terms of the types of cases that the system is able to handle. For example, if the system is built to handle only a particular type of case, e.g., simple single-issue cases its comprehension is rated low. If the system is able to handle a broad range of cases from simple to complex including the ability to model the complexities of those cases to the satisfaction of the users, then the comprehension is rated high.