Re: [ba-ohs-talk] IBIS question
It seems to me that some of the discussion below begins to suggest
something along the lines of Kelly's Personal Construct Theory. If one
wants to register opinions, and I've thought about this a bit, one just
gives an 'argument' window that has a slider on it, center neutral, and
degree of bias in favor or against a proposition, or as a selector between
different propositions. (01)
It then seems to me that when you do this, you might lose the ability to
state justifications. (02)
At 05:12 PM 1/28/2002 -0800, you wrote:
>Some comments and suggestions:
>> > ...I'm not sure how to use IBIS to capture the tradeoffs
>>> and the conditions under which one would redo the decision.
>>I note, too, that among related choices, one is forever doing pro's
>>and con's that are the inverse of each other:
>I think what you are saying is that you would like to use a "relationship
>expressing relative merit" between two Ideas according to some criterium.
>To use your example:
>> Choice A:
>> con: Takes longest (compared to choice B & C)
>> Choice B:
>> pro: Takes least amount of time (compared to A & C)
>> Choice C:
>> ???: Takes a medium amount of time.
>Instead of representing it as pros and cons, you'd like to say:
>A is worse than C (according to criterium "amount of time it takes")
>C is worse than B (according to criterium "amount of time it takes")
>the logical next step would be to add Criterium "beauty" for example ;-)
>C is worse than A (according to criterium "beauty")
>I don't know how to show this graphically, but the semantics certainly
>exist and we should be able to represent them.
>Come to think of it, decisionmaking often involves isolating/defining
>"value coordinates" such as "speed", "cost", "customer satisfaction" etc.
>When we do compromises, we usually try to understand relative merit along
>those "value coordinates" (e.g. arguments like "A takes longer than B but
>the customer will be much happier") and then tradeoff between them ("it's
>more important that the customer is happy than that it does not take long")
>What if one extended IBIS by those two concepts, "value coordinate", and
>"relative merit according to value coordinate" (which is sort of a ternary
>One might be able to show those as a table, with "value coordinates" being
>the columns, and Ideas being rows? The value of the cells being either
>numeric, or more likely, --/-/0/+/++ reflecting the relationships between
>Is that a wild thought?
>R-Objects Inc. (04)