Representation in Cyberspace Study

Berkman Center for Internet & Society at Harvard Law School


Deliberative Polling As a Model for ICANN Membership
James S. Fishkin,
Chair, Department of Government
The University of Texas at Austin

I. Introduction:  The ICANN Membership Problem

The White Paper calls for ICANN to "preserve, as much as possible, the tradition of bottom-up governance of the Internet," specifying that "Board Members should be elected from membership or other associations open to all or through other mechanisms that ensure broad representation and participation in the election process."  The ICANN Bylaws meet this challenge with a two-track Board selection scheme:  nine of the Directors are chosen by the three technically-oriented Supporting Organizations;  the other nine Directors are to be elected by an at-large membership.  To satisfy the White Paper, these two tracks, taken together, must be open, representative, and participatory.

In studying the various models that have been proposed for the at-large membership structure, it has become apparent that these values of openness, representation and participation are in tension with each other.  For example, an entirely open membership risks capture by highly motivated special interest constituencies, thus sacrificing the value of representativeness.  Indeed, each of the three White Paper values can be paired with a related risk:  unqualified openness can leave the membership vulnerable to fraud and capture;  complex schemes to achieve representation can lead to the Balkanization of broad communities and the entrenchment of special interests;  efforts to enable meaningful, ongoing participation by all members risk the creation of a membership behemoth that is unwieldy, sluggish, gridlocked, and ultimately irrelevant.

The challenge, then, is not simply how to make ICANN's membership open, representative, and participatory – ICANN must find a way to balance the values of openness, represenation, and participation so as to minimize the risks of fraud, capture, Balkanization, entrenchment, gridlock, and irrelevance.

We propose a membership model that accomplishes this goal.  Our model will achieve near-perfect representation of the global Internet community, will operate openly and transparently, and will effectively eliminate the dangers of fraud, capture, Balkanization, entrenchment, gridlock, and irrelevance.  These goals are accomplished, however, at a cost:  the model does not facilitate ongoing, universal participation in the membership structure.  For the reasons set forth below – for example, the fact that interested individuals will have numerous opportunities to participate in ICANN governance through the supporting organizations and elsewhere – we believe that this cost is small in comparison to the dramatic advantages the model offers to ICANN and the Internet community.
Our model proceeds from the premise that ICANN's membership should be genuinely representative of the entire Internet community, not merely those directly self-interested portions capable of mobilizing to demand and win representation.  Put another way, the problem is how three groups are to be connected: a) the relevant public, b) the membership and c) the board.  ICANN's Board must be accountable to the membership; the membership, in turn, must be representative of the relevant public. The central issue is whether the membership will be representative of the relevant public and whether the board members elected by that membership will be accountable to the membership. Conventional mechanisms of election can be employed to guarantee the latter connection.  But to rely on the voluntary self-selection of members virtually assures fraud and efforts by special interests to capture a disproportionate share of the membership.

II. Critique of Existing Models:  Capture, Fraud, Verification, Voting Mechanics, Representativeness

For the moment let us assume that the relevant public is the world community of internet users.  Arguments might be offered that everyone will, in the end, be affected by the internet and that, therefore, all persons are part of the relevant public. But for the moment, it seems practical and relevant to limit the account of the relevant public to those who actually use the internet. As that public expands, the membership structure adopted needs to feature a mechanism for updating itself to represent the users of the internet on a reasonably current basis.

Given this characterization of the relevant public, the danger for any self-selected membership model for ICANN is that the membership will bear little relation to the relevant public. First, it is likely that if anyone in the world can simply volunteer for membership, then the composition of the membership will be subjected to organized efforts at  “capture” —thousands of persons will be recruited to join who actually are enlisted by lobbying groups, special interest organizations or corporations, directly or indirectly. Instead of the “grass roots” public, the membership will consist largely of an organized or synthetic public, what American  lobbyists have come to  call “astroturf”. More generally,  Norman Bradburn of the University of Chicago has coined the term "SLOP,"  meaning “self-selected listener opinion poll,” to describe any effort to represent the wider public with a self-selected group. SLOPs almost always fail to be representative because, in general, those with special interests or those who are organized are more likely go to the trouble to select themselves. If the membership of ICANN were a SLOP, it would very likely be unrepresentative of the broader public of internet users and, in addition, it would provide a dubious and difficult constituency for any board to attempt to represent.

Because of the technological ease by which email addresses, web pages, and domain names can shield true identities, the Internet makes it very easy for determined individuals or groups to capture mechanisms of representation that rely on self-selection.  One need only note Time Magazine's experience with its online SLOP asking the Internet community (as well as other readers) to vote for the Man of the Century.  The winner in every category?  Kemal Ataturk, who defeated Winston Churchill among "Warriors and Statesman," and Henry Ford in "Builders and Titans," Albert Einstein and Marie Curie among "Scientists and Healers."  He was even the winner among "Entertainers and Artists" and "Heroes and Adventurers."

In sum, for ICANN's membership structure to be open, representative, and participatory in a meaningful way, it must be able to prevent the easy use of fraudulent identities and to verify the identities of those who vote.  In our view, none of the alternative models detailed in the Berkman Center study incorporates a satisfactory answer to this problem.

III. First Principles:  Representation and Deliberation

For the membership to serve as an appropriate democratic foundation for ICANN, it should a) be representative of the entire public of internet users, and b) provide participants with ameaningful opportunity to become informed on the issues on which they might be consulted.  If the membership is representative, then any internet user around the world would be able to say that persons like that user are included in the membership in their appropriate proportion.  Under such circumstance, the interests and concerns of that user are likely be voiced and heard when the membership is consulted.

One way to achieve both representativenessand deliberation is random sampling followed by an opportunity for the participants to become informed.  If random sampling were employed then every internet user around the world would have an equal random chance of being part of the microcosm that is chosen.  But being representative is not enough. While ordinary internet users around the world are likely to be inattentive and uninformed about complex internet issues –like the mass public of citizens in most countries around the world on most policy questions—a representative membership can be giventhe opportunity and incentive to become engaged in a continuing process of discussion and the exchange of information. A deliberative and representative membership would provide a basis not only for electing board members but also for (non-binding) consultation on difficult issues.

The problem facing the ICANN membership with respect to information is not unlike the problem of mass publics on any complex public policy issue. A great deal of social science research has established that citizens are subject to “rational ignorance”. Why should any individual citizen become expert on complex public policy issues when his or her individual opinion or individual vote is unlikely to make much difference to the issue?  One vote in millions will not affect an election. One letter in thousands will not likely affect a representative’s vote. We would like ideal citizens to spend a lot of time and effort becoming informed,but most ordinary citizens have more pressing issues to deal with as they go about their lives and provide for their families. Hence efforts at direct democratic consultation have been hampered by the lack of any rational incentive for individual citizens to become informed and to discuss the issues so as to form considered judgments on most public policy questions.

Even worse, the answers to sample surveys on many public policy issues may well represent phantom opinions or what Professor Philip Converse of the University of Michigan famously termed “non-attitudes.” Respondents in polls are always reluctant  to admit that they lack opinions on issues, even if the questions concern  issues that the respondents never thought about before. Instead of answering “don’t know” they will frequently make up opinions on the spot—opinions that can be shown to vary randomly from one time to another. Significant percentages of the mass public had “opinions” on the now famous “Public Affairs Act of 1975” even though the act was fictional.  Without significant efforts at the provision of information, consultation of the mass public is likely to yield non-existent opinions or highly uninformed opinions crippled by rational ignorance. For a mechanism to provide meaningful public input, it must, somehow combine statistical representativeness with deliberation.

IV. Background:  Deliberative Polling

A. Face-to-Face Deliberative Polling
Deliberative Polling was devised precisely  in order to combine statistical representativeness of the mass  public with deliberation. The idea is to take a statistical microcosm of the public, and after an initial survey, invite its members to engage in a continuing dialogue with carefully balanced briefing materials, opportunities to interact with panels of competing experts and decision-makers and, over time, come to a considered and informed judgment on the issue. The aim, in other words, is to provide an informed and representative input from the public on complex issues.

The process  is based on the work of Professor James S. Fishkin of the University of Texas, Austin.  He and various collaborators have conducted fourteen Deliberative Polls in different parts of the world with random samples of respondents, brought together face to face, to deliberate for a few days. The samples have been representative of the relevant populations and they have undergone large, statistically significant changes of opinion on many policy issues. For more on Deliberative Polling see or Professor Fishkin’s recent book The Voice of the People (Yale University Press, expanded edition, 1997). The Deliberative Polling concept derives from an ancient form of democracy—the deliberative microcosms of five hundred or more citizens in Ancient Athens that were selected by lot for the Council, for legislative commissions and for citizens juries.

The process was the basis for the January 1996 National Issues Convention broadcast on PBS and hosted by Jim Lehrer at the University of Texas. This was a dialogue with Presidential candidates and Vice President Gore and a random, representative sample of nearly 500 US voters drawn from around the U.S. and selected by NORC (the National Opinion Research Center at the University of Chicago). Other national Deliberative Polls have been held in Britain with the television network Channel 4 on crime (1994) Britain’s Future in Europe (1995), the monarchy (1996), the economic issues in the British General Election (1997) and the future of the National Health Service (NHS) on its fiftieth annivesary (1998). In Texas, a series of local Deliberative Polls have been conducted as part of the regulatory process for public utilities in conjunction with the Public Utility Commission of Texas and each of the eight major investor owed utilities in the state. These Deliberative Polls have concerned whether there should be investments in fossil fuels, or renewable energy (wind or solar power) or conservation and whether there should be subsidies for low income customers. A variety of specific public policies have been implemented as a result of these Deliberative Polls.

B. Online Deliberative Polling
We believe that much or all of the process of Deliberative Polling could be adapted to an online environment. Indeed, as we discuss below, we believe that the technique of Deliberative Polling could be adapted to facilitate the creation and operation of an online Deliberative Council – a representative microcosm of the world Internet community that would actively and meaningfully serve as the ICANN membership for a period of one year, until it elected new Board members and was replaced by the next year's Deliberative Council.

There are daunting, but we believe, resolvable issues in identifying a random sample of the world’s internet users.  Once identified, participants would be given a survey of their attitudes and demographics and then they would be invited to participate in the experiment. Since we will have data from all the participants, those who agree to be part of the Deliberative Poll and those who only agree to take the initial survey, we will have the data to judge the representativeness of the deliberative microcosm. As in other Deliberative Polls modest incentives would have to be offered to pay participants for their time and effort in participating in an on-going deliberation. Most importantly, the participants would have to become convinced that ICANN was interested in their views, once they had a chance to discuss the issues. The secret of Deliberative Polling, whether on a highly salient issue such as crime, or on a comparatively low visibility issue such as public utility regulation, is that the respondents must be convinced that their voice in the process will matter. In this way, the incentives for “rational ignorance” can be overcome. The respondents are participating in an important and visible experiment. If convinced that their views will matter, members of the public will go to a lot of time and effort and can become informed and engaged in the discussion of complicated policy questions.

V. Interim Solution:  Deliberative Poll

We believe that the model of a Deliberative Council – that is, a randomly selected, scientifically representative microcosm of the global Internet community – best resolves the ICANN membership conundrum.  However, we also recognize that the idea is sufficiently radical that many will want it to be tested and proven before ICANN considers whether to adopt it as its membership structure.  Accordingly, we propose that ICANN supplement its initially chosen membership structure (whatever it may be) with an experiment in online Deliberative Polling.

Given that Deliberative Polling has only been tried in a face to face context, we urge a transitional period of experimentation to see whether this ancient form of democracy—the deliberating microcosm selected by lot – can be adapted to the internet. Over a two year period, large scale experiments in Deliberative Polling can be conducted with internet users, and the results can be compared to a those from a trial experimentation with a self-selected membership model.  After two years, a decision would have to be made by the board as to which model to institute for election of future at large members of the board—either the combination of organizations and the Deliberative Council or one of the self-selected models of at large membership.

An online Deliberative Poll will provide an important complement to whatever initial membership structure that  ICANN adopts.  By assessing the informed views of the entire global Internet community on the key issues confronting ICANN, the online Deliberative Poll will (1) better inform ICANN about the needs and wishes of its global constituency, and (2) provide a significant check on the success or failure of its initial membership structure.  If the initial membership structure has been captured or is otherwise unrepresentative of the global Internet community, the results of the online Deliberative Poll will indicate as much.  By proceeding with an experimental online Deliberative Poll, ICANN can secure not only better inputs, but also a measure of insurance against the very real dangers that confront its membership structure.

VI. Potential Membership Structure:  Deliberative Council

As we outline above, Deliberative Polling offers a model of democratic consultation that solves some basic problems facing ICANN’s institutional design.

First, by using a random and representative sample of internet users, it solves the problem of the connection between the relevant public and the membership.  If the membership structure consisted of a Deliberative Council – a random sample of say five hundred  to a thousand persons, renewed on an annual basis -- this sample could not be captured by organized self-selection or volunteering, because it would not consist of volunteers. It would certainly include persons with intense interests, or indeed, persons with any particular points of view, but only in their proportion to the total population of internet users.

Second, it combines representativeness with deliberation. Such a Deliberative Council would become informed about the issues and could provide a mechanism for thoughtful and representative consultation by the board over the course of its term each year. The idea is that the Deliberative Council would perform two functions — (1) elect the nine at large Board members and (2) provide a mechanism for public consultation on difficult issues for which the Board desired thoughtful and informed public input.

Of course, the mechanisms by which the Deliberative Council is randomly chosen from the global community of Internet users must be  open and transparent and subject to the highest standards of social science, to prevent the misimpression that any group or entity is packing the Council with its partisans.  More than other models, the Deliberative Council will depend on openness and transparency for its legitimacy.

Each Deliberative Council will require an extensive effort to prepare its members for discussion of the issues. Advisory Boards representing diverse and competing stakeholders and competing expert opinion on the topic will collaborate to produce a public briefing document which serves as the initial basis for discussion. This document will be publicly available and carefully scrutinized for balance and accuracy. Once an agreed briefing document has been developed, it is sent to members of the sample. The members of the Council are  randomly assigned to small groups for discussion with trained moderators. While it may be advisable to keep the names of the members of the Deliberative Council confidential (so that they can not be overwhelmed by lobbying by outside special interests), it will be important for purposes of transparency to make all the discussions public. During the deliberations, the small groups arrive at key questions they wish to ask panels of competing experts and decision-makers. Activists, stakeholders, experts of various sorts would all have prominent roles to play, both in the formulation of the materials and in the continuing dialogue with the members of the Deliberative Council.

We note at this point that organizations may have an independent claim for representation, separate and apart from individuals.  The model of the Deliberative Council is flexible enough to allow for some of the at large Board members to be elected by the Deliberative Council, so as to represent the entire mass public of internet users, while the remaining at large Board members are elected by organizations. One method might be rotation:  in any given year, the Deliberative Council would elect two Board members, while a body representing organizations elected one;  the next year, the numbers would reverse, and so on.. Furthermore, the fees paid by organizations might be dedicated to subsidize the Deliberative Council membership representing the mass public. We envision that ICANN will pay the members of the Deliberative Council to insure a representative input from all the world’s internet users.

Last modified Feb.22, 1999 by Diane Cabell

Berkman Center for Internet & Society