Hi Steve,

I'm intrigued to hear about your experience with speech recognition software.  For a long time I've wondered if tools like Dragon NaturallySpeaking were fast and accurate enough to let a dialogue mapper shift to spoken input, and from your report it sounds like it is.  (I've always thought it would be more fun if you could do meeting facilitation as pure performance art, sort of like Laurie Anderson :-)).

Have you tried it with Compendium?  Do you speak the node type you want or use a hot key?  How do you do the mouse input stuff like linking?

So my recommendation to the Compendium developers is one I have made a couple of times before.  Include an "Objective" node type into Compendium.  And some sort of mechanism for easily folding in the coupled pros and cons.

The topic of extending the IBIS grammar to add Objectives as a "primitive" node type has come up recently in the context of a new IBIS mapping tool called bCisive: http://bcisive.austhink.com/ This is a very slick piece of software from Austhink (the makers of Rationale) that is in many ways easier to use than Compendium (for example, when you drag a new node into the map the location where you drop it determines what node it links to, and the rest of the nodes in the map shift to make room for the new one).  bCisive has the basic IBIS elements (Questions, Ideas, Pros, Cons) plus many others, such as Options (a more mature Idea), Decisions, Tasks, References, and Facts, plus a set of node types for more rigorous reasoning about the validity of claims: Evidence, Rebuttal, Challenge, Explanation, Example, and so on.  (But I don't think there's an Objective node type in the current version.)

The discussion I was having with the folks at Austhink was: how do you decide what should be in the set of primitives for a mapping tool?  Some people might need a Requirement primitive, for example, or a node type to represent Definitions.  So designers of mapping tools have to really think about how to balance between having a rich set of primitives (to provide lots of expressive power) and a thin set of primitives to make the system easier to learn and the maps easier to read.

But at the end of the day IBIS has an ace up its sleeve in the expressive power department, because of the unlimited expressive power of questions.  If I need to map objectives during a session, I create the question "What are the objectives?" and, voila, the nodes that I hang from that question are Objectives.  This is, by the way, the intention of the Criterial question, one of the 7 most important question types in design/problem solving maps.  (Check out the section "The Criterial Leap" on page 160 of Dialogue Mapping.)  Thus the question "What are the criteria?" leads you and the group to more abstract versions of the pros and cons.

The same is true of any primitive node type you need, including Pros and Cons!  We could drop Pros and Cons as primitives in IBIS and instead hang the questions "What are the Pros?" and "What are the Cons?" to each Idea responding to the main question.  What are the Requirements?  What is the definition of X?  What are the Facts?  What is the Evidence?  You get the idea.

I suggest that the litmus test for potential new primitives is: Are there specific semantics associated with that primitive which the software can support?  I would argue that we need to add an "ActionItem" primitive to Compendium, for example, because it could have additional fields in the node for "Who?" and "By when?", and when such a node was created the tool could automatically dispatch an email to the owner of the actionitem, and perhaps even enter it into their calendar.

What do you think?

Cheers,
Jeff


At 05:39 AM 5/22/2008, you wrote:

Jeff,

I think you're retrospective on dialogue mapping is very insightful.  It's funny, I just attended an Expert Choice training seminar as an observer.  And Dr. Foreman, the software inventor made some of the same comments about the AHP.  Facilitators like myself are really excited about the AHP construct because it seems so intuitive.  And students are initially energized but then become deflated after one or two less than satisfactory implementations.

I think there is commonality of the implementation barriers among the different facilitation techniques and it's essentially the human interface that connects a group to the construct.  Sometimes it's because of the limited skill of the facilitator but other times it's the methodology itself.   I've come up with two recommendations for Compendium.  One for users and the other one for developers.

The user recommendation is simple but has been a dramatic improvement in collaboration quality for me.  And that is the use of Speech to Text software to support facilitation.  The most popular speech to text system is Dragon NaturallySpeaking.  It is less than $150 at Amazon.  Once the facilitator is disengaged from the keyboard, the conversational flow is much more natural and fluid because there is not the staccato back and forth of inputting verbal content.  The facilitator only need develop a few smooth transition steps for managing the speech software.

One reason why an IBIS map becomes so dense is because of the numbers of pros and cons that are generated and the fact that there isn't a mechanism now to conflate them.  The AHP also uses a pro and con technique to elicit decision drivers.  But then adds an additional step to convert the pros and cons into objectives.  Generally the pro of one alternative will be a con of another.  Coupled pros and cons can be conflated into a single objective.

For example if in a car purchasing dialogue a Mercedes is an alternative and so is a Toyota, high price will be a con for the Mercedes and low price will be a pro for the Toyota.  In the AHP process, the pro and con are conflated into the single neutral objective "minimize price".  Doing that potentially halves the number of pros and cons in an IBIS map.

So my recommendation to the Compendium developers is one I have made a couple of times before.  Include an "Objective" node type into Compendium.  And some sort of mechanism for easily folding in the coupled pros and cons.

I think this additional step would provide two substantial benefits.  First, they would visually de-clutter many maps enabling a more complete understanding of the decision context.  Secondly, as the facilitator worked the group through the consolidation step, the explicit decision drivers (the objectives) would surface with much more clarity.  That is when the group would have their A-Ha moment about what really matters to them.  And that is when the process itself would be validated.  (As a dispassionate facilitator with catholic sensibilities about techniques, my ideal would be to use Compendium to initially define and bound the decision space.  And then transfer that content to an AHP model for actual prioritization.)

I know that you can create customized nodes in Compendium.  But I am proposing or formal extension to the IBIS process as both the theoretical and practical improvement.  Jeff, rather then cross post this at the Compendium site, I'll let you take it up those guys if you think it makes sense.

Regards,

Steve Mack

Dr. Jeff Conklin
CogNexus Institute ... Building Shared Understanding of Wicked Problems
http://cognexus.org Phone: +1-707-256-3425  Fax: +1-707-256-3903
1037 Juarez St., Napa, CA  94559    USA