Tools for Enabling Structured Discourse

Copyright (c) 2000-2001 by Rich Morin
published in Silicon Carny, September 2000

The Internet supports many different kinds of discourse, ranging from chaotic to highly structured. Here are some interesting and useful tools that you may wish to try out.

The Internet supports many different styles of discourse, ranging from simple email to highly structured systems for collaboarative document preparation. No single style is suitable for all situations, but a combination of styles can often be made to work in concert.

This article examines some ideas for supporting structured discourse. Be warned; the ideas range from mundane through experimental to totally hypothetical. We don't have any definitive answers, but we have found some interesting questions!

Mailing Lists, Newsgroups, Forums, etc.

Simple email is a very informal and highly flexible way to exchange information. Mailing lists, by adding multiparty distribution, threads, and archiving, can encourage extended discussion and greatly enhance the impact of the contributions.

Mailing lists can also be a fertile breeding ground for arguments. Whether motivated by honest differences of opinion, egotism, or the desire to start fights, some participants seem to find arguments on every list they enter.

Moderated mailing lists address this problem, either by authorizing individual messages or by censoring problematic topics and/or authors. Message authorization is very effective at reducing noise, but it introduces time lags and requires a lot of work. Topic/author censorship works almost as well, over the long run, so many mailing lists have adopted it.

USENET newsgroups and Web-based discussion forums act very much like mailing lists, with a couple of significant exceptions. Because anyone can "drop in", topics can arise in a more spontaneous manner. And, because articles disappear very quickly, the same topics tend to recur again and again.

Although mailing lists and newsgroups might seem ephemeral, this is not necessarily the case. Many mailing lists are archived and even indexed, Dejanews provides a searchable archive for netnews. Thus, your random comments may well find their way into the historical record.

Articles, Papers, Web pages, etc.

Articles and papers, whether published formally or simply posted on a personal web site, are also a style of discourse. The pace is slower, to be sure, but the ideas tend to be more carefully developed than they might be in a mailing list posting.

Some sites provide web-based forums to promote public discussion of published articles. Others (e.g., SunWorld) invite private comments. In either case, the author benefits from hearing the reactions of the readers.


It can be difficult to make detailed comments about an article. How, for example, do you specify the exact context for the comment? Alternatively, how do you indicate the fact that another web page should link to this point in a document?

Ideally, all web pages would have mechanisms that would allow readers to make annotations, create links, etc. This is not the case today, however, and is not likely to be the case any time soon.

CritSuite (sponsored by the Foresight Institute) addresses this deficiency. It provides a mechanism for creating "mediated" versions of arbitrary web pages. The mediated versions can be annotated in a number of ways, without any effort on the part of the original provider.

Although powerful and flexible, CritSuite is very simple to use. Go to the CritSuite web site, then give it a target URL. CritSuite retrieves the desired page, displaying it under a CritSuite banner.

You can then look over any annotations others have made, then (if desired) make your own. Navigation is automagically handled via CritSuite, making any subsequent page available for commenting.

CritSuite annotations can cover either an entire document or a selected phrase. They can also be tagged as to type: comment, issue (disagree), query, support (agree). The annotations appear (on the mediated version of the web page) as pairs of small icons, bracketing the annotated text.

By way of example, I have put a small annotation onto my personal web page ( To view it, go to and then ask the page to show you Alternatively, you can simply use the URL

In the second paragraph, you should see that the word "cats" has been highlighted. To see my comment, click on one of the highlighting tags that surround the word "cats".

Wikis and WikiWebs

Annotations are useful, but a bit limiting. What if we wanted to create a document in a "Bazaar" style, folding in contributions from a variety of authors? Public CVS repositories support this, to a degree, but they are a bit too formal and intimidating.

The "Wiki" concept was created by Ward Cunningham. Wikis make collaborative document trees easy to create, moderate (if desired), and use. To make users' lives easier, Wikis also provide services such as indexing and searching.

A "WikiWeb" brings this notion to the Web, making it available on a global scale. Wikis have been implemented in a variety of ways and are available for a wide range of topics. Some implementations store documents using CVS; others (including Ward's original WikiWikiWeb) don't.

The rules governing particular Wikis vary. Some allow anyone to edit pages; others restrict editing to certain parties. The balance between convenience and security is always a gnarly issue.

Wikis that are backed up (e.g., by CVS) are defended from permanent damage, because any changes can be backed out. Nonetheless, cleaning up after a determined wall-scribbler could still be a trial for the Wiki administrator.

One of the key strengths of the Wiki approach is that it's very easy for authors to create hyperlinks. Consequently, pages in a WikiWikiWeb tend to be very well connected.


IBIS (Issue-Based Information System) was conceived 30 years ago by Horst Rittel (and implemented on an Apple ][ :-). IBIS attempts to solve "Wicked Problems"; ones in which the problem definition is dependent on its solution.

The advantage of the IBIS approach is that it aids participants in defining problems and reducing the kinds of repetitive exposition that are all too common in less-structured forums.

IBIS uses a small set of document types to encode an entire discussion. Forcing the discussion into this format can be both tricky and exasperating, but it can pay off hugely in clarity.

In the terminology used by QuestMap (a commercial IBIS implementation, produced by Group Decision Support Systems), A "question" (e.g., "Which programming language should we use?") encapsulates a single, open-ended issue. An "idea" proposes a possible answer to the question (e.g., We should use Perl."). An "argument" (e.g., "Yes, it's very flexible." or "No, it's too slow.") justifies (or refutes) the position.

As the discussion proceeds, the participants define subsidiary questions, eventually creating a "tree" of argumentation. Each party is able to express an opinion on any topic, so there is little room for repetitive argumentation. A participant can tweak and tune his argument, be he cannot insert it into random parts of the tree.

IBIS under Wiki?

I have speculated about the possibility of using a Wiki to host an IBIS discussion. The Wiki model of pages and links could easily store the various IBIS document types. As the discussions grew in size, however, specialized tools (ala QuestMap) might be needed to navigate and manage the topic trees.

IBIS and Wiki are just two of the tools that are available to support collaborative work. Whole books and conferences, in fact, are devoted to the topic. So, if these ideas interest you, you're in luck; there's a lot of interesting things to read about, try out, and develop further!



Cunningham & Cunningham, Inc. (Ward Cunningham)
Public Wiki Sites
Wiki Wiki Web (aka dejanews)

Foresight Institute


Group Decision Support Systems
IBIS Manual
Wicked Problems
Working Papers

"Issues as Elements of Information Systems"
WP-131, 6-1-70
Horst J. Rittel & Werner Kunz
Institute for Urban and Regional Development
UC Berkeley

About the author

Rich Morin ( operates Prime Time Freeware (, a publisher of books about Open Source software. Rich lives in San Bruno, on the San Francisco peninsula.