William Drenttel | Essays

Edward Tufte: The Dispassionate Statistician III

A year in advance of its publication, a chapter from Edward Tufte's next book, Beautiful Evidence, has been posted on his website under Ask E.T. "Sparklines: Intense, Simple, Word-Sized Graphics," makes a case for the "wonderful possibility of writing with data graphics." Thus, a sparkline (or a word-sized graphic) when embedded in a text, "means like a word and shows like a graphic."

Tufte acknowledges that others before him have embedded graphics within sentences: most notably, he cites the example of Galileo portraying Saturn as "a visual noun" within a sentence of his book Istoria e dimostrazioni intorno alle macchie solari.

Such citations do not quite explain Tufte's obsessive defense of "sparklines" as the ultimate information graphic — what he terms "a context for nuanced analysis — and perhaps, better decisions." Nevertheless, he excels in openmindedness: here, he uses his site to prototype his argument, thus initiating an open discussion and feedback loop. The feedback itself is fascinating, and includes detailed technical corrections, queries about "shade, hue and intensity," grammatical recommendations, and philosophical ruminations about statistics.

Some comments are exercises in pure logic: "'Sparkline/glucose/128' seems to have poor flow. In a traditional statement, 'Glucose 128' is short for 'The Glucose level is 128' or similar. (Your) sparkline has a more awkward interpretation, passively asserting, 'Here is some data, which is on Glucose, and has an end value of 128.' A better arrangement would be 'Glucose/sparkline/128.' The corresponding meaning would be 'The Glucose chart has a latest level of 128.'"

I'm easily persuaded. More importantly, though, I'm captivated by corrections that begin, "Dr. Tufte. This is a lovely chapter. Nevertheless, I have some comments, concerns and, in a few cases, corrections." This excerpt comes from Alex Merz, a biochemist at Dartmouth, who continues: "'Pieces of our DNA remain much the same as that of the first worms in the muck.' This is OK, but could be more correctly phrased: 'Our DNA retains sequences largely unchanged since the first worms.' The pieces of DNA in our genome are copies, not originals. The bit about the muck is cliche."

Other writers go on to dispute many facts cited by Tufte on the topic of DNA. It is to the author's credit that he accepts and, more impressively still, that he responds to such criticism. Indeed, Tufte not only posted his chapter for public critique, but appears to both encourage group discussion and online commentary. Who among us would willingly subject their writing to such "peer review" — let alone, allow our work to be moulded in the process of so doing?

Posted in: Science , Technology

Comments [3]

Pictures inside of paragraphs has always been an interesting idea. Though I feel that it has mostly been a jarring experience ever since picture books of my youth. It seems the just jump out at you too much and demand too much attention. I'm sure they can be toned down but I don't know if I've ever seen effective use of them with out it messing with the clean paragraphs.
Stefan Hayden

Seems like an interesting idea again, Mr. Tufte...

But as always, i take his ideas with a grain of salt. Even skimming over his drafted "sparklines" chapter (which I must admit I first read as "sparkliness"... as in the quantitative shimmering quality of something like a bedazzled pair of jeans), I can quickly see that there are some benefits to use of (his) concept. And some detriment.

What concerns me about anything Mr. Tufte puts forth is that many so-called information designers will go forth, spreading the gospel of (in this case "sparklines") without much regard for Tufte's full story. Suddenly we'll see sparklines everywhere, even where they're not appropriate.

I also think that there are times when a sparkline may be overkill. Take Tufte's example of baseball team sparklines: they provide quick access to a season of baseball, yet if all i want to know is how many games were won and lost, the sparkline is in the way. And, if i really want to know if the Orioles won or lost game 31, am i going to stand to count the tiny tickmarks one by one? Where does the application work, and at what point does it fail - if by only making the user experience more cumbersome? If the sparkline is showing too much information, is this a sin as grave as not enough?

As it goes, I don't want anyone to think I don't appreciate Tufte's work; I happen to own two of his books, and this one is likely to make it into my library as well. But I'll be taking the "sparkline" concept as an idea more than anything else.

Of course, Mr. Drenttel, your point is less about "sparklines" than it is about the willing participation in public critique of our ideas. I think it's rare that someone with a profile as high as Mr. Tufte engages in such a dialog, but I think it lends credibility to his cause. For the lesser known, such an opportunity is rare - but I think we all engage in this kind of dialog with every post here on Design Observer.
Andrew Twigg

From what I've been able to pick up from this article, sparklines seem like they'd be useful in some contexts, and it would certainly be nifty to be able to use this function in a word processor. However, as a synesthete, a word like "Saturn" carries useless but familiar color information that a picture of a planet does not. This is not to say that the concept of sometimes replacing words with pictures does not hold a certain philosophical appeal. Bringing words back to the things they really represent instead of further abstracting them with letters seems attractive, if only on the surface. But I really would miss those synesthetic colors.

Also, from a practical standpoint: I can't read pictures aloud as readily as words, so arranging a sentence of any complexity in my mind in order to extract meaning from it would become a much less automatic process if the sentence contained pictures.
Julia Kriz

Jobs | July 16