Friday, April 25, 2008


I got an e-mail from Rafe Donahue with a fabulous work-in-progress attached, a piece he has been writing on the importance of not letting summary measures conceal what's interesting about a body of data (for example, distributions). He talks about various ways one could use graphics to display data so that important patterns can be discerned. (Rafe has typeset the piece in InDesign; a second e-mail talked about all the amazing things one can do with InDesign, making me glad I managed to pick up a student edition on the strength of my weekend course back in January. )

I then got an e-mail from a writer suggesting an interview, and asking for my version of my departure from the States a few years ago. I'm perfectly happy to talk about how I came to Berlin, but it's hard for me to see leaving the States as a radical departure from the norm. Having bought Microcharts only a couple of weeks ago I naturally saw this as a good time to play with it: I put together a table showing time spent in and out of the US from birth to now and converted these into a table of miniature pie charts. If I had ever worked out how to get Parallels up and running again I could have taken a pretty screenshot on my Mac; instead I've had to put Microcharts on the Brontosaurus (aka the Sony Vaio), take a screenshot and paste it into Paint, with this somewhat dowdy result:

For purposes of Mac-Windows comparison, here's a nice clean little screenshot of a couple of pie charts cobbled together in Excel on my MacBook, showing amount of time spent in/out of US a) up to the age of 20 and b) thereafter:

A cursory glance at these graphics shows that the thing that needs explaining is the brief return to the US in the last decade. (Long story. Don't ask.)

While the data may or may not show a strong preference for expatriate life, anyway, the presentation certainly demonstrates a strong preference on the part of the author for telling a story using miniature pie charts. We know this not merely from the use of a handful miniature pie charts, but from the fact that the author has gone out and paid good money for an Excel plug-in which generates miniature pie charts. (That's not all it does: it also generates miniature columns, miniature bar charts, miniature line charts, miniature win-loss charts and a couple of others that I forget.)

This is a preference that actually says quite a lot about a character: the character may be unable to explain things in words, not because she is inarticulate but because patterns of data are better presented in a graphic array. We don't see graphic arrays very often in modern fiction, which means a) that we literally don't see patterns of data about characters that can best be presented in graphic array and b) that texts don't represent a mode of thinking that is characteristic of the type of person who thinks in terms of patterns of distribution.

This is actually rather odd. There are other styles of thought and communication that can't get far using words - music is one very striking example. Musicians can play together without speaking a word of each other's languages; it's very powerful. But that's something one could only represent in a medium that made use of sound; you can't get sound off the printed page. Graphic arrays, on the other hand, are made to be seen; we just never see them.


oxfv said...

The Microcharts would strike home with Edward Tufte, the inventor of Sparklines and a great proponent of condensed data presentation. He offers some impressive historical examples of data presentation in his several books.

Oh, never mind, a quick google shows that Microcharts are based on Tufte's work.

BENDIS said...

One other area where shared / agreed upon information about symbols can allow two people who speak completely different languages) to share a very intricate experience is the game of chess.

Andrew said...

I don't mind the miniature pie-charts, but I don't like the grid, since it breaks arbitrarily at 10-yr points and also invites the reader to find irrelevant patterns by reading down the page (just as in, when reading a book, one can become distracted by the diagonal patterns of blank spaces that go down the lines of text).

Ithaca said...

AG - this was not my first choice of graphic, but Excel couldn't do any of the things I wanted to try. Excel would not be my first choice of software (it is not exactually unusual to envisage a graphic and find that it cannot be produced in Excel), but it has been a while since I used R and I would not be able to use it without a lot of backtracking.

I suppose dividing one's life into decades is arbitrary in some sense, but it is one that has some power over the way people live their lives. It's not uncommon to think: I've spent my 30s doing X, am I still going to be doing X 10 years from now? 20 years from now? Or: I'm almost 30, is this what I'm going to be doing with my life?

I think now: I've just turned 50, I've spent the last 10 years dealing with publishing crap, do I REALLY want to spend another 10 years like this? There must be SOMETHING else I could do. And the thing is, when I have this kind of discussion with myself, I think: well, maybe X, Y or Z, but then I'd have to go back to the States. So I can't do X, Y or Z. Isn't there anything ELSE I could do?

So I don't know that ten-year divisions are unmotivated; apart from the first they do correspond to various crises whose acuteness related, precisely, to the advent of the turn of a decade. Since the first consideration in addressing a crisis was always to solve it in a way that made it possible to stay out of the country, it's not surprising that the main feature of a temporal representation of divisions of place of residence is the change from the years under 20 (living wherever my parents happened to go) and above (not).

Ithaca said...

bendis, chess is an interesting example, because it can, of course, be represented on the page, but no amount of cleverness with design will enable the non-player, or the novice, to see on the page what a good player sees.

Anonymous said...