A survey of online tools for public deliberation

A number of tools have been proposed for public deliberation, especially if one goes beyond online tools and attempts to review processes as well: from the historical Athenian agora to the modern Open Space technique. The aim is not to be exhaustive, but to attempt to identify what we might learn from a few existing tools and techniques.

In this page, I will first attempt to survey the online tools, which I will group in four categories:

  1. Newsgroups, Chat rooms and Web forums

    The first attempts to use computers as collaboration tools on the internet were news groups, followed by chat rooms and web forums. In all of these systems, multiple participants write messages of various lengths, which are instantaneously broadcast to all participants. The medium attempts to be as transparent as possible, resembling a noisy room with no moderation; but because messages are written, it is possible to read and follow many threads of conversation which would be utterly confusing if they were actually spoken simultaneously. Additionally, the computer medium usually archives the conversation, and may allow for some extra modalities such as quoting, text searches, hyperlinks, or (usually limited) edition of posts by their authors after broadcasting; but the conversation dynamics is not normally controlled by the tool.

    • Moderation

      It is very common for these debates to be the seat of very heated debates; the partial anonymity afforded by computer-mediated communication, especially with nicknames, seems to suggest to some people that civility is optional. Many forums appoint moderators to oversee the conversations and ban verbal abusers from posting. In extreme cases, for low-bandwidth newsgroups, moderators will approve individual messages. (This is also aimed at spamming.) In general, newsgroups users try to maintain a reputation in the community, and a combination of forbidding anonymous posts and banning flamers' accounts (even if they could in theory register under a new nickname) is usually sufficient. (Suler, 1998 [http://www.rider.edu/~suler/psycyber/badboys.html]Suler, J. (2002). The Bad Boys of Cyberspace: Deviant Behavior in Multimedia Chat Communities. In The Psychology of Cyberspace, http://www.rider.edu/~suler/psycyber/badboys.html (article orig. pub. 1996 with Phillips, W., in CyberPsychology and Behavior, 1, 275-294.))

    • Collective moderation

      Some very large web forums, such as slashdot [http://slashdot.org/moderation.shtml], which have to deal not only with uncivil users but also a large number of posts that do not add much to the discussion ("me-too" posts), will distribute the moderation task to many users: comparatively experienced users are given "moderation points" that they can spend to put the most interesting posts in focus, or to (optionally) hide the most egregious ones from most readers. However, though this allows for skimming highlights, it is found that sanguine, entertaining but ultimately ill-informed posts are often encouraged by this practice. Still, distributed moderation is one of the most serious large-scale attempts at online collective opinion formation.

    How appropriate are these tools for collective deliberation and decision making? First, at least in the software community, forums have shown themselves very efficient at building communities around diffusion of knowledge. The recreational aspect is also important to most users, though that role is now facing competition from social sites or more graphically rich forums such as Second Life [http://secondlife.com/]. However, a key characteristic of these forums that make them inappropriate for decision-making is their ephemerality. Though individual interventions are recorded, and can be referred to, the number (and short length) of interventions makes it extremely rare that a given usenet post is referred to after the few days following its posting; with chat rooms, the lifespan is counted in seconds. Web forums lie somewhere in between, but quoting and reference are virtually unheard of. These posts are written on the spur of the moment, so as to make sense in context, but accessing that context is daunting at best, tedious at worst. Also, the links between ideas are often implicit, and retracing them also requires archaeological patience. As records of a zeitgeist, they will no doubt be invaluable someday; as documents on which to base decisions, they fall far from the mark.

  2. Text repositories, Content Management Systems, and Wikis

    As opposed to forums of various kinds, which are modelled on conversations, many tools take document writing as a basic process. Collectively authored documents are especially frequent in organizations, and before word processors, elaborate conventions have evolved for co-authoring and edition on paper. Some word processors now allow to track changes or add editorial notes in a document, which makes that process that much easier. But a lot of impetus has come from software development; a software project is, in many ways, a text written by a team, and incoherence due to coordination errors can be fatal. So software developers have created tools that facilitate many people working on a collection of related documents.

    Again, specifics vary: many tools keep a history of which exact changes were made, and by whom. Some allow multiple parallel versions of the text to co-exist, and to be compared. Most will prevent conflicts caused when two writers attempt to edit the same document (or document section) at the same time, by forbidding the second author to integrate the changes without review; but a small number of tools, instead, will show to all readers the process of the document being edited in real time, character by character. This fine-grained interaction changes the dynamic considerably, and the document then doubles as a chat room. A few tools actually provide a separate chat room to discuss edition decisions without interfering with the main text. This extreme case aside, text repositories are basically best suited for small closely-knit communities with an established common purpose.

    • Wikis

      Wikis were one of the first widely-deployed systems that allowed non-programmers to edit text documents collaboratively. What made wikis popular was that the resulting documents were available on a web site, and could both be accessed and edited with a browser; also, they allow authors to create new documents, and links between documents, simply by naming them. The openness of wikis, which originally allowed any reader to change any content, has allowed a very distinct communication dynamics to evolve around them. The first Wiki [http://c2.com/cgi/wiki] was developed by Ward Cunningham as a repository for software design patterns, which were an evolving set of abstract best practices distilled from practitioner's stories. The purpose of the site was explicitly evolving from the specific towards the synthetic, and this fitted the natural tendencies of the community whom the site served. In that cultural context, the lightness of the wiki, which imposed no structural constraint on data, became a strength, allowing many different forms of interventions. Some pages are still more synthetic; some bear the traces of debates, above or below the text of their resolution; but stories are rarely contentious in themselves, and remained mostly untouched, and even the lessons learnt from those stories were interpreted within the context of a discipline, and the debates were mostly constructive.

      In the wider cultural field of Wikipedia [http://en.wikipedia.org/wiki/], debates were not always so constructive, and destructive comments were more frequent. Wikipedia first contained the damage thanks to page archives, making page vandalism as easy to undo as to do; and by splitting each topic into discussion and synthesis. Even then, the wikipedia governing board recently had to restrict edition rights. Still, despite numerous storms, the wikipedia is the largest collection of collectively authored documents, and as such a shining example of collective intelligence in action.

      Could these successes be harnessed for collective deliberation and decisions? In many ways, wikis as they exist are a wonderful tool for deliberation, thanks to their highly informal nature; and provided that there are participants in the community willing to do the work of maintaining it, it is possible to maintain a good overview of the main issues. Also, members are required to periodically discard some debris (such as old conversations) from the wiki, to ensure each page goes right to the point (this is known as "gardening"). Though labour-intensive, these activities have been shown to be possible in dedicated communities; but many wikis have failed without a core group of volunteers dedicated to this work. However, wiki's very informality makes them less suitable for further processing of the opinions they contain for synthesis, as would be needed to incorporate votes into the platform. Indeed, wikis' fundamental unit is the page, not the concept; and though it is possible (and customary) to maintain a strict one-concept-per-page discipline in a community, this is cannot be enforced by the tool itself and there is a constant need to train newcomers.

      Nonetheless, wikis teache us many valuable lessons, both technical and social:

      • do not constrain users with formal models
      • people respect stories
      • create a separate space for dissent
      • maintaining a synthesis is hard work
  3. Blogs

    In a way, blogs (web logs) are a simplified way to display articles on a web site; since there is no interaction in a shared text, they could be compared with news groups. However, that would be misleading, as the social dynamic is completely different. Authors maintains articles on their site, and readers subscribe to updates on a per-site basis. As readers chose who they subscribe to, abusive writers are not an issue. Also, as an author's writings are displayed as a whole, they are further encouraged to maintain face, which furthers both civility and coherence. Most blogs allow comments from external reader to each individual article, which usually functions as a mini-webforum, with all attendant limitations; but the main venue of dialogue in blogs is that other readers maintain their own blog, and individual articles are made in explicit reference to an article on another blog. This makes the thread of dialogue more difficult to follow, but encourages more self-contained responses, which are usually better thought out as a result. Text ownership is very important in the blog culture; as we can see from the development of a tool [http://www.cocomment.com/] to bring one person's comments on other blog sites together. (I personally suspect that many active blog authors had been avoiding Wikis because text ownership was unclear.) Finally, search engines specialized for blogs, such as Technorati [http://technorati.com/], help bloggers stay aware of topics of mutual interest.

    We see again how comparatively small changes in the publishing model (pull vs push; text ownership) can change the communication dynamics considerably. It has been claimed, notably by Joichi Ito [http://joi.ito.com/static/emergentdemocracy.html], that the blog communication model is heralding a new form of democracy; and certainly it has succeeded in bringing public attention to many issues missed by traditional media (the blogging community has been called an echo chamber, for its capacity to amplify issues.) However, it is also frequent for blog authors to only subscribe to the blogs of authors of like mind, which minimizes the potential for actually addressing contentious issues. Note that this becomes more of a trend as some blog platforms also include features to display personal tastes and preferences, which encourages connecting with like-minded people, in a way similar to that common on social sites (such as mySpace, Facebook, etc.) Also, though the search engines help propagate ideas in the entire blogging community, the simple tags used for search do not express fine distinctions or nuance. Again, this is not a sound decision platform, though it has achieved a lot to raise the level of deliberation.

  4. Structured discourse

    Tools that would assist in the deliberation process, whether by counting votes, visualizing the argument structure, or any other function, cannot do so unless they know about the various issues, arguments and so forth brought by the participants. Since computers do not understand text, this requires that participants make their discourse extremely explicit, in a way that is usable by a computer. This requires some skill, and has been the greatest barrier to spread of such tools; yet, many tools have been designed for computer-mediated collaboration, and there is now a better formal understanding of the structural elements of a debate that should be made explicit to a computer.

    • Issue-Based Information System (IBIS)

      One of the first methodologies for adding structure to complex debates was proposed by Horst Rittel [http://www-iurd.ced.berkeley.edu/pub/WP-131.pdf]Werner Kunz Horst W. J. Rittel. «Issues as Elements of Information Systems». , jul 1970. in 1970. As later refined, that methodology consists in classifying statements as either issues (aka questions), options or arguments. (Some variants also distinguish evidence for arguments.) Those statements are displayed as a network, with options relative to issues and arguments either pro or con an option. The network is usually constructed by an experienced IBIS facilitator in the course of a meeting, with participants confirming that the statements, as reformulated, still reflect their views. Users of IBIS systems are usually very enthusiastic about its usefulness in enhancing meetings; however, it has been found [http://kmi.open.ac.uk/publications/index.cfm?trnumber=kmi-05-18]Simon J. Buckingham Shum, Albert M. Selvin, Maarten Sierhuis, Jeffrey Conklin, Charles B. Haley, Bashar Nuseibeh. «Rationale Management in Software Engineering», Hypermedia Support for Argumentation-Based Rationale: 15 Years on from gIBIS and QOC, 111–132. Computer Science Editorial. Springer-Verlag, 2006. that offline access by less-experienced users often confuses the issues.

    • delphis

      Another advanced technique that has been developed for structured group deliberation is the delphi method [http://web.njit.edu/~turoff/Papers/CDSCMC/CDSCMC.htm]Murray Turoff, Starr Roxanne Hiltz, Michael Bieber, Jerry Fjermestad, Ajaz Rana. «Collaborative Discourse Structures in Computer Mediated Group Communications». Journal of Computer-Mediated Communication, 4(4), jun 1999. , originally develeped by Murray Turoff. One key aspect of this method is that of distinct voting scales: arguments can be rated independently on importance or validity; proposed actions or goals can be rated on desirability and feasibility; other scales include rating trends on significance, assumptions on validity, and measurements also on significance. This provides a richer voting structure, and when there is agreement on goals, using this method with a group of experts from very different disciplines has been shown to yield accurate predictions and sound measures. (This is a sound application of the wisdom of crowds [/]Surowiecki, James. The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations. Little, Brown, London, 2004. principle.)

    • Hypertext and the semantic web

      In general, making links between ideas more explicit was the primary intent of the inventors of hypertext; some of the techniques pioneered by Ted Nelson and Doug Engelbart involved typed links which could be traversed in both directions, as well as graphical mind maps. The world wide web as it was developed by Tim Berners-Lee was more modest in scope, and only involved unidirectional untyped links. However, Berners-Lee has since been championing [http://www.w3.org/DesignIssues/Semantic.html]Berners-Lee, Tim. Semantic Web Road map. Online draft. 1998. new tools for explicit semantic links in web pages. These new semantic links, combined with blogs, would allow for a computer to gather information about actual opinions. I believe this direction holds much promise; though the current breed of semantic desktop applications is not yet convenient enough for casual usage by most non-technical people, this area is under serious development. Another issue is that most semantic applications so far have been using data models more appropriate for social software than online deliberation; though some important efforts are also being made in that direction.

Overall, past research shows that deliberation is best treated as a structured form of conversation, and semantic web tools provides us with the raw tools to express this structure. However, since the advent of the world wide web, research in hypertext representation has been very busy with ways to represent very large networks with low structure; what is needed now are ways to represent medium-size networks with more complex structure. It is far from certain that traditional network representation techniques can scale up to debates involving a very large number of participants; and even if that were so, informed decision making still requires the main points of the deliberation to fit in a person's brain. Fortunately, a lot of work has already been done on argumentation visualization [http://www.visualizingargumentation.info/]Paul A. Kirschner, Simon J. Buckingham Shum, Chad S. Carr. Visualizing Argumentation. Springer-Verlag, London, 2003. , and research is still ongoing. Also, methodologies need to be refined that allow transforming informal input into structured arguments, in a way that can keep the participant's sense of contribution. In other words, the need for informed techniques for structural overview is as important as ever, and despite advances, much research remains to be done.

P.S. Since I last wrote this piece, many new argumentation visalization tools have been developed... The following list will not be as organized as what precedes, but I feel I must keep somewhat up-to-date.