How to raise the level of a debate? How to make conversations more productive? How to break out of "Yes it is!"- "No it isn't!" point-counterpoint haggling? How to take complex, tangled webs and make them comprehensible? How to attack "wicked problems" and begin to solve them?

What's needed is a framework for disciplined argumentation --- something to provide:

There are many structured-thinking frameworks, but few of them are widely known and fewer still are much used in real life. One of the nicest (in my limited experience) is VIMS, the "Visual Issue Mapping System" developed and popularized by Dr. Jeff Conklin. (Jeff was formerly at Group Decision Support Systems, and before that at the Microelectronics and Computer Technology Consortium (MCC); I hear that he's his own company now. See for an essay on VIMS and its ancestry. A form of VIMS is instantiated in some (rather idiosyncratic) software called "QuestMap".)

The power of idea-mapping comes from a simple syntax that organizes arguments. In a nutshell:

  1. Write down a key question ("?"). (Make it brief and open-ended. Seek to begin the question with "Why?" or "How?" or "What?" or "Who?". Avoid multi-clause queries; split them into their fundamental sub-issues. Stay away from questions which can only be answered "yes" or "no". Eschew biased pseudo-questions which imply a single answer or which contain arguments.)
  2. Write down candidate ideas ("!") in response to the key question, and link them to the question with arrows: "(?) ← (!)". (Ideas are potential answers to the question. Express them compactly and objectively. A good idea does not include reasoning or argumentation.)
  3. Write down arguments pro and con ("+" & "-") in response to candidate ideas: "(!) ← (+)" & "(!) ← (-)". (Capture arguments succinctly. Indicate whether they support or refute the idea to which they are linked. Keep arguments atomic, one to an elementary block.)
  4. Repeat! (Formulate more questions and link them onto existing questions, ideas, and arguments. Add new ideas and hook them to questions. Connect additional pro and con arguments to ideas.)

This process results in a map of the conversation, a diagram that captures the essence of a debate. Questions can be linked in anywhere: to another question, to an idea, or to an argument. (Question authority!) Answering ideas, however, only connect to questions. Pro & con arguments only apply to ideas, never to other arguments or to questions. (No question is a bad question!)

These simple constraints give structure to a dialogue. They unravel tangles and result in a smoother, stronger fabric of analysis.

Thursday, September 14, 2000 at 21:09:03 (EDT) = 2000-09-14


There are more papers on wicked problems and IBIS and VIMS at . (The term VIMS has been replaced by the more intuitive "Dialog Mapping.") In the past few years there has been new research on using the IBIS notation to build semi-structured models ... this work is called "Compendium" and there are papers about it (e.g. ,

But, beyond all the papers and research, there is still this very simply, even elegant, IBIS notation for "stuctured argumentation." Questions, Ideas, Arguments. To explore a problem of any size means there will be several Questions, perhaps even dozens or hundreds of them, and for each a set of possible answers (Ideas), for some of these Ideas there may be Arguments that support or object. It becomes a big network of small chunks of knowledge. QuestMap is an (admittedly idiosyncratic) tool for building these networks. Verizon has created a similar tool, written in Java, but still crude in some ways, called Mifflin.

But it seems we need a *family* of tools for IBIS -- tools with a simple command-line interface for getting people started, tools with progressively more elegant interfaces and network navigational horsepower for larger and more complex IBIS maps, tools for capturing conversations in meetings, tools for asynchronous dialog on the Web, and tools that "pipe" these maps into (and out of) various other formats so that the IBIS maps stay integrated with existing tools, data, and practices. -- Jeff Conklin

Jeff, wonderfully stated; I'm in violent agreement yet again with you. Tnx esp. for the pointer to your recent paper ( = "Facilitated Hypertext for Collective Sensemaking: 15 Years on from gIBIS") ... most insightful. --- MarkZimmermann

(correlates: OneThingAfterAnother, TheNewTwenty, Comments on DeliberateOpinion, ...)