E-Media Tidbits

ManyEyes This ManyEyes word tree gives an at-a-glance overview of the kinds of problems covered in a report on US military operations in Iraq.

Yesterday the New York Times published an editorial on the apparent failure of Bush Administration blunders in Iraq: How Not to Plan a War. It pulls together highlights from two recently published and particularly damning Army analyses: After Saddam: Prewar Planning and the Occupation of Iraq (prepared by the RAND Corporation for the US Army in 2005, suppressed until recently) and On Point II: Transition to the New Campaign (by the Army Combined Arms Center).

The nut graf of the editorial: "Reading these two histories, you have to wonder: Where was the common sense? What were they thinking? How could people who laid claim to such sterling resumes get so much so wrong?"

So what, exactly, were the main problems with the US in Iraq? I've been playing with IBM's Many Eyes data visualization tool, so I thought I'd use that to get an overview of the problems.

I downloaded the copyright-free RAND report and used a PDF text extractor to capture all its text. I uploaded the resulting to ManyEyes as a dataset, and I created a "word tree" visualization. (I'll do the same later with "On Point II," but that 720-page document will take longer to process.)

...So what's a "word tree" visualization? You specify a word or phrase, and ManyEyes shows you all the different contexts in which that string appears in a tree-like branching structure. This helps reveal recurring themes in the document, and shows how topics and subtopics are related

Here's my RAND report word tree showing exactly what the problems were. It's interactive: Clicking on related words along the branches of the word tree shed more light on what's being discussed. The word tree I've shared keys everything off of the word "problems" -- but you can create your own visualizations from my RAND report dataset. (I made it public.)

...The Many Eyes word tree visualization is a bit of a blunt instrument, a purely linguistic analysis -- but it's still pretty neat. This approach shows how it can be useful to journalists trying to quickly gain insight on the contents of large, cryptic documents (reports, transcripts, legislation, etc). Depending on how good a "story" a word tree tells, you may or may not want to share it with the public. But it can aid your research process.

Occasionally, however, a word tree gets right to the heart of the story. Here's a ManyEyes word tree created by Wade Rousch in April 2007 based on former attorney general Alberto Gonzales' Senate testimony, keyed off the superficially innocuous word "don't."