Marking private
Since my blog is being scooped up at my workjob as an RSS feed, I marked some pretty non-work stuff as private. So all of you with private access can still get access, it just won't get republished.
By: Colonel Nikolai
Sat Jul 05 13:13:24 CDT 2008
For General Release
Filed under: technology->software->inventedhere
Permalink
•Comment•
Crude hits $144 bbl
Wow!By: Colonel Nikolai
Wed Jul 02 18:01:58 CDT 2008
For General Release
Filed under: politics->peakoil
Permalink
•Comment•
[Restricted Posting]
[The real content of this entry is not accessable to the public. If you have insider access, you need to log in to read this posting.]By: Nobody
Thu Jun 26 10:34:40 CDT 2008
For Restricted Release
Filed under: wouldnt->you->like->to->know
Permalink
•Comment•
Basic Assumtion of Analytical Design
Gary and I were talking about analytical design today. He had a really good point: good analytical design doesn't ensure good analysis; not even a little bit. He's right, but only if you drop the basic assumption that intellectual analysis is based on natual law.
If you examine the Minard graphic (page 124 of Beautiful Evidence) from the perspective of what audience is consuming it, it can support diametrically opposing arguments. Consider a French audience consuming the chart and concluding the same as Minard himself: that war is a colossal waste. But say we consider a Kruschev-era Soviet audience looking at the same graphic, unadulterated. One may well conclude that this chart could support a Soviet military build up, ostensibly the opposite of what Minard intended!
I think the important thing to remember is that a well-executed analytical design does not per force ensure that an analysis of it yields one type (or even one well-defined range!) of conclusions. Beyond the problem of cognitive biases, the big assumption is the adherence of a principle that the product of good analytical thinking will hew to natural law; that the basis upon the rock of natural law is presupposed in analytical thinking. I don't think that assumption holds very often; wish it did! I think it is important to reveal this fundamental assumption in analytical design.
By: Colonel Nikolai
Wed Jun 25 21:29:07 CDT 2008
For General Release
Filed under: technology->analyticaldesign
Permalink
•Comment•
Presenting Data and Information: A One Day Seminar with Edward R. Tufte
Yesterday I went to an all-day seminar on Analytical Design by the reigning master of the field, Edward R. Tufte. It was really neat; he's really passionate about his work and has published many books, the four that are the biggest hits with nformation Technology people and Graphic Design / Visual Artist people are The Visual Display of Quantitative Information (which is in it's Second Edition now) Envisioning Information, Visual Explanations and Beautiful Evidence. If you take the seminar, you get all four books presented to you when you register.
Here are my notes from the course:
"The Information is the Interface", meaning that the information itself is the final criteria of its own presentation. There is no meta-interface, there is no "design" that exists outside of the rules that govern analytical thinking alone.
This necessarily dictates the overarching strategy of analytical design: "Do whatever it takes" to telegraph the information. Therefore the test of good analytical design is a content response, not a response to the design itself. If someone says (taking the US cancer rate geographic distribution maps as an example) "Wow, my house is in a county that's way above average in cancer rates!", then you know you have done a good job. On the other hand, if someone says "Wow, pretty maps! I like the color!" your design is probably in trouble.
"Do not prejudice the mode of display." Don't display the information in a format that is convenient for your presentation, convenient for your organization or for any other non-analytical bias. Never do a lowest-common-denominator presentation. This will merely guarantee that -- if you thought your data would be difficult for people to understand in it's unadulterated form that you felt you had to modify it -- it will never be understood.
"Simple tabular data is underrated." A good table will beat a bar chart, a pie chart, etc. (It's worth it to note here Tufte didn't show ONE bar or pie chart except to show a bad example of analytical design the whole day). A good model of tabular data presentation and analysis would be any sports section of the Washington Post's articles of post-game NFL/NBA analysis. Most of these reports are 10-100 times better than any Power Point presentation template.
Performance Graphics. Most business data analysis is based on performance. So begin by organizing your data into performance stanines or whathaveyou. NOT alphabetical. Start with a Supergraphic. A supergraphic is nothing more than a broad, flat view of the information concerning what's relevant to your presentation. "Information is better presented when a lot of data is displayed adjacent in space, not stacked deeply."
Two major issues in analytical design:
- Putting multiple dimensions of information onto the two dimensions of the computer screen and / or paper.
- The way we measure and value the data has much to do with the resolution and throughput of the data as it is / can be presented
Since Galileo, the first assisted eye data collection process he did represented a 10x or 11x enhancement of the human eye, we have achieved approximately 45x enhancement of the human eye.
Map/Data "Legends" are mostly hangovers from a time when graphics and text were separate processes requiring separate techniques and skills to produce. There is no such excuse now for this methodology. Legends cause problems in data translation. Solution is to use real annotations with text instead of legends.
"Maximize content reasoning time; no detours, no "celebrations", no incoherent transformations; minimize content translation time."
1570 English edition of Euclid's Geometry used a 3-d folding model pasted into the folio. Bringing something from the real world into your analytical design will do wonders to invoke a content response. Tufte goes through an actual hospital bill, line item by line item of a very sick "Mrs. K___" who dies after 26 days in an ICU. There is lot of drama and deep analysis of the document, which is simply annotated in English text.
At its heart analytical design is about:
- making smart comparisons
- focusing on explaining cause and effect relationships
- showing multi-variable information
- integrating into a coherent gestalt words, numbers and diagrams
- documenting itself: What is the display about? Who is the author? What role or title does the author hold? When was the work done? What are the data sources? What assumptions are present in the display? What is the scale of the measurement? Who was the publisher?
The basis of any efficacious methodology in analytical design is about taking "fundamental intellectual tasks" and turning them into commensurate "design principles". This "symmetry of production and consumption" of rules is almost unique to the analytical design field. Galileo: "Theories are to be tested with hard evidence, not endlessly speculated with rhetoric." Wrt the celestial observations he made, he coined the term "Visual Certainty" ("Oculat(r?)a Certitudine")
User Interface Design. The big problem with the GUI is the bifurcation of the production of disparate elements of a document into different applications. This process breaks down or fragments analytic thinking.
"Personalization is the last refuge of the Web Design Scoundrel."
"Avoid information hierarchies: they also fight analytical thinking."
Display indifference: Aim for "Whatever it takes" in all possible output devices. Paper printouts have 10-12x the amount of data that a computer screen has. For every 3 or so analysts, have one 11-17 full duplex color printer available to them.
"Graphics should not be a special occasion in display. A graphic need not rule lines, boxes and ideally should not be created in a separate program. They may be used in place of a number or a word."
"An approximate answer to the right question is worth far more than a precise answer to the wrong question." (Quote form the Statistician John Tukey)
If you want to see cutting edge analytical design, look at the scientific journal Nature. "The top 10% of their graphics are the best in the world."
Advice for presenters
Use power point only as a way to project data; a "projector operating system" Don't use bullets. Don't use indents. Don't use logos.
Always hand out a Supergraphic beforehand. This is typically an 11x17 full overview of the entire relevant data in your work that has the fidelity, if not the precise resolution, of the complete body of information.
Begin with a coherent, prose summary of the work. This is a substantial one. Use English. Don't use "bulleted grunts" or formulate it in the style of a "mission statement". Explain the problem. Explain the problem's relevance. Explain a proposed solution.
"Practice improves performance" Use a video camera to knock out mannerisms in speech and body language. Show up early. Don't tell jokes to "break the ice". Use a handout (supergraphic) Focus on the technical report.
By: Colonel Nikolai
Tue Jun 24 13:14:37 CDT 2008
For General Release
Filed under: technology->analyticaldesign
Permalink