[Date Prev] [Date Next] [Thread Prev] [Thread Next] Indexes: Main | Date | Thread | Author

Re: [ba-unrev-talk] Icons for IBIS

Ten or 15 years ago I was really into capturing subtleties of argument 
structure -- different weights, constraint relationships, etc.  Then I 
tried to teach that stuff to business people who were trying to solve 
complex problems -- even the ones who really understood the logic got it 
mixed up (e.g. incorrect link semantics) in the press of sorting out a real 
problem.  So I have a great appreciation for structures and interfaces that 
are simple and intuitive -- the test being: can my clients use them while 
making real decisions without getting all balled up in the logic.    (01)

One of the issues that has come up recently is about the structure of 
rebuttal to arguments.  An awful lot of problem solving discussion centers 
on groups evaluating the validity and relevance of arguments, which means 
that there's a whole "sub-grammar" for debate about argument validity (as 
opposed to pros and cons about options/alternatives/ideas).  My experience 
with clients suggests that anything that is logically adequate will be 
overkill in practice.    (02)

Anyway, I think that generally the sense of an argument (whether it's a pro 
or con) is best treated as link semantics, with the node graphically 
amplifying that semantics wherever it is unambiguous (as QuestMap 
does).  However, I also think that a lot of skill and practice goes into 
wording arguments so that they are semantically clear and unambiguous.  My 
first impulse about transcluding arguments is not to do it -- restate the 
argument in each context to be clear and compelling in that 
context.  What's an example where you really need to transclude just the 
argument, without the Question and Idea that it's a part of?    (03)

Jeff    (04)

At 09:29 AM 4/11/2002 -0400, you wrote:    (05)

>Eric's example below is actually a case where something is being used as an
>argument that is really better stated as an idea/position/answer. "X costs
>$10,000 to buy" isn't stated as a an argument; it's a statement of fact. "X
>is too expensive", OTOH, would be an argument.
>Eric Armstrong <eric.armstrong@sun.com>@bootstrap.org on 04/10/2002
>07:42:33 PM
>Please respond to ba-unrev-talk@bootstrap.org
>Sent by:    owner-ba-unrev-talk@bootstrap.org
>To:    ba-unrev-talk@bootstrap.org
>Subject:    Re: [ba-unrev-talk] Icons for IBIS
>Peter Jones wrote:
> > This question is addressed to those folks out there implementing GUIs
> > for IBIS.
> > If I have a particular node in a discussion that is marked as a
> > positive,
> > and I then transclude that node into another discussion where I
> > want it to play the part of a negative, how do I do that given that
> > I want to maintain the identity of the node across the two discussions?
> > Are the icons independent of node identity?
>Yes, the icon must be independent of the node, for exactly that reason.
>The "for/against" property is a property of the relationship, not a
>of the node.
>For example, the node:
>        X costs $10,000 to buy
>Is a negative response to the question:
>        Q: Should we buy X or implement it ourselves?
>             A: Purchase X
>                  +: X costs $10,000 to buy
>                  +: If we develop it, it will cost $100,000.
>On the other hand, it's a positive if the question is:
>       Q: What product should we buy?
>             A: Purchase X
>                  -: X costs $10,000
>              A: Purchase Y
>                  +: Y costs $1,000
>Similary, if the question is whether to lease or buy,
>that knowledge nugget / factoid could play a positive
>or negative role.
>This issue also brings up the issue of "scales" that we
>visited a while a back. If there are three purchase
>options, X, Y, and Z, and they cost different amounts,
>then a price dimension exists, and each product has a
>point on that scale.
>Now, an IBIS discussion can be carried on without
>reusing data. In such contexts, it makes sense to assign
>positives and negatives to nodes. But when the IBIS
>concept is extended to include access to reusable
>"knowledge", using transclusion or some other mechanism,
>then it is only the data which is transcluded, not the
>+/_ relationship.
>That leads me to think of invariant structure nodes that
>cannot be moved reused, and maleable content nodes
>that can be. The structure node may have the + attribute
>then, because it is fixed in a hierarchy (it represents the
>relationship). It would link to a content node which contains
>the data "X costs $10,000". That content node could then
>be reused in multiple locations, were it desirable to do so.    (06)

--     (07)

Dr. Jeff Conklin             <jeff@cognexus.org>
CogNexus Institute ... Collaborative Display, Collective Intelligence
http://cognexus.org          Phone/Fax: 410-798-4495
304 Arbutus Dr., Edgewater, MD 21037    USA    (08)