Q1 - Ben Fry illustrates the importance of "pre-attentive" information, using the example that the human brain can more quickly parse information that is visual in nature and come to a qualitative understanding. Whereas statistics deals with a set of well defined biases in the collection of data (voluntary bias, response bias, etc), what biases are we likely to encounter in data visualization? Are the biases justified because of our pursuit of a qualitative understanding of the information? At what point do our work become uninformative?
While that may be a side effect, the issue is not that the visual design should be “prettier”. Rather, that the approach of the visual designer solves many common problems in typical information visualization...The issue is about diagrams that are accurate, versus those that are understandable (or subjectively, those that viewers prefer to look at).
Q2 - Ben Fry argues in his example of the Treemap project that the visualization suffers from a number of layout issues. How can we quantify these issues, or are they purely subjective? For instance, the person who created Treemap may have a thorough understanding of the information, and the "visual noise" we consider are only there to help further his understanding. It thus seems like our efforts are focused on giving the lay person a way to come to a quick subjective understanding of the information.
All visualizations seem to require the viewer to have some degree of contextualizing or orienting information. Considering the designer as a lay individual, are there, and how can he/she create visualizations that require a thorough understanding of the subject matter?
Q3 - How does a designer deal with contaminated data? What is the likelihood that any given person will look at a visualization multiple times, more specifically, an updated visualization? Given the difficulty of censoring information, how can we ensure that correct information is disseminated?