Retrieval + Use (was Re: [Sigia-l] Information Visualization)

[Karl Fast]

> > But haven't we had enough proof of concept demos and working models
> > of infoviz products to see how the retrieval part works?
> 
> Absolutely not, and that's precisely the problem I am referring to.

I sorta agree with both of you.

Have we had enough proof of concept demos?

Yes and no. There certainly are a lot of demos, but I don't think
quantity is the issue.

To my mind, more valuable questions are: Do we have a good
understanding of why these demos do or do not work? Have we learned
enough from them that we can confidently build effective
visualization apps? Do we have theories and models that explain why
certain techniques do and not work in various contexts?

No. We understand some. We're not completely clueless. But some
fundamental problems remain.

And this is why I claim that infoviz is more of a research problem
than a commercialization one. Criticizing it for it's lack of market
success misses the point.

One of the basic challenges for infoviz is to connect form to
meaning, and then coupling that with appropriate interaction. That's
non-trivial. The infoviz literature often talks about moving from
'slow reading' to 'faster visual perception.' This is only part of
the story. The other element is to develop deeper understanding.
There is a large cognitive element to this work.

The intention of visualization is 'amplify cognition.' Right now we
don't really understand how this all works, though we do know that
the right representation can dramatically alter our ability to
understand a concept, complete a task, etc.

> It's trivial to knock out a whiz bang infoviz demo when you *don't*
> connect to large datasets.

I don't think it's "trivial," though it's certainly much easier.


--karl