The Purpose of Visualization
From Visualization Sp06
Lecture on Jan 17, 2006
- Chapter 1: Graphical Excellence, In The Visual Display of Quantitative Information. Tufte.
- Chapter 1: Information Visualization, In Readings in Information Visualization. Card, et al. (handout)
- Decision to launch the Challenger, In Visual Explanations. Tufte. (handout)
Maneesh - Jan 16, 2006 11:09:09 pm
Card et al. briefly mention Larkin and Simon's study of people solving physics problems with and without the use of diagrams. Their paper Why a Diagram is (Sometimes) Worth Ten Thousand Words explains how diagrams aid cognition in more detail. Definitely worth a read.
jheer - Jan 17, 2006 12:33:10 pm
Tufte's discussion of "Graphical Excellence" floats between theory, guidelines, and design sense--a set of tensions that underlie much of the practice of visualization. How well do these separate amongst the actionable, implementable, or inspirational? What aspects can we build into tools and systems? What concerns currently escape formalism, and how might we address those in a constructive fashion?
Bryan - Jan 19, 2006 03:45:46 pm
Regarding the Challenger disaster: I wonder who ultimately made the decision whether to launch: politicians, administrators, engineers, scientists? It seems to me that the ability to absorb and synthesize badly presented information depends on training. Engineers are probably best equipped to handle information in the form of the fax that was recieved, but my guess would be that administrators and politicians ultimately made the decision to launch. I could imagine a scenario where NASA engineers were convinced by the faxed data, but failed to make the case to administrators because of the format it was presented.
Maneesh - Jan 19, 2006 04:49:11 pm
Right. As with many beauracratic entities it not at all clear how the final decision to launch the Challenger was actually reached.
Tufte has gone on to analyze the presentations created during the recent Columbia disaster as well. Once again a few engineers knew that the Shuttle may have been damaged during liftoff, but they were unable to present their information effectively.
Cynbot - Jan 23, 2006 06:17:45 pm
If the decision was ultimatley going to be political, could a good visual presentation have made any difference. Is there a way to create a scientific argument that actually invokes the subtext of a situation such as politics, ego, fear, etc. and maybe try to play on those factors to also to make your argument. And should those poltically motivated visualizations be used at all if one is to call themselves a scientist. Or could it be justified?
Ryanaip - Jan 23, 2006 09:46:55 pm
Bruno Latour, who began his career studying the process of science (the politicization of science) and developing Actor Network Theory, has more recently been exploring the process of politics (where politics is taken broadly to include any form of public interaction). Specifically, he argues that the representational tools of politics are far less evolved than those of science. For example, we do have the visual tools to provide scientific intuition about why the Challenger and Columbia disasters happened. But, we don't have good tools for representing the institutional causes that resulted in the two missions being launched in spite of the scientific data that was available. Likewise, we don't have good tools for representing the political history of statements like "there are weapons of mass destruction in Iraq" which might help to assess their validity. I think that visualizations incorporating such "subtext" will become increasingly important as people seek to better understand the processes that underpin our culture.
IvanTam - Jan 24, 2006 08:03:02 am
Regarding Cynbot's earlier comment: I don't think one should of oneself as any less of a scientist for viewing visualizations of data. However, a good scientist (or any critical thinker, for that matter) should develop a good sense for what information the visualization is and isn't conveying and make a decision on whether or not further investigations into the visualized information is necessary. Comparing the account on the Challenger launch decision and Tufte's chapter on Graphical Integrity in The Visual Display of Quantitative Information, I think a case can be made that even those who are trained releatively well in quantitative methods (the "rocket scientists") can produce visualizations which do not convey information free of distortion.
Nchentan - Jan 25, 2006 11:12:47 pm
Regarding Information Visualization Reading: on page 20, about metadata, the author comments on table 1.11 that if the number were the number of car accidents, a natural whole number, then it's not permissible to interpolate, given latitude and longtidue. I have a comment on this. Some natural number quantity may still make sense to interpolate. For example, the population living withing 1 km^2 of the location given by latitude and longtitude. It is reasonable that the interpolation still can give a good estimate of population at a given unobserved point. It seems that the issue is more whether the number is related smoothly to location, not whether the number is continuous or not.
Maneesh - Jan 29, 2006 10:32:54 pm
Ryanaip - It would be interesting to figure out ways to visualize the kinds of political processes you mention. Does Actor Network Theory provide enough structure to build visualizations based on it?
Nchentan - The variable population can be a bit tricky. In some instances the term population means a raw count of the number of people and in other cases the term population means the number of people divided by the unit area. In the latter cases population is being used as a shorthand for population density. Population density can be interpolated to decimal values. But, if population simply means count then I don't think it makes sense to interpolate it to decimal values.