My book Visualizing with Text is nearing publication (early November!). One goal of the book is to appeal to both researchers (with structured text with logical arguments), and designers (with many examples and pictures). I really like books with lots of images. Like the quote from Alice’s Adventures in Wonderland, in the title, I prefer well illustrated books. And I don’t like reading about visualizations where you only get little teeny snapshots: after all the primary subject is visualization!
Text is good to explain, structure and provide context, but pictures are important to a designer: to see how the conceptual description is actualized; to see all the little design decisions that contribute to the whole; to see the anomalies and things that could be improved; to provoke design inspiration; and so on. Arguably, Jacques Bertin was successful with Semiology of Graphics 50 years ago because he presented a theory supported with many examples and illustrations. My ideal goal is to have a 50/50 split between text and images. But, measuring text to images is a bit tricky:
It turns out that the publisher and I count images differently. We can easily agree on page count (274), but my publisher currently counts 146 illustrations and I count 250 images (+/- a few to be sorted out). What? How can we differ so much?
I think the publisher is counting image captions. On the otherhand, I’m counting the individual images that might be referred to in a single caption, for example here’s figure 1.3. It has a single caption but two different images: left is a book from 1497, right is a different book from 1589. I count this as two images (the images come from two completely different sources).
Then, there are parts of the book where I discuss using visualization techniques inline with text. Since, I’m advocating inline visualization, these are simply text richly formatted within the flow of the prose. In this case, there is no image caption, so they don’t count at all, whereas I count it as one image (I had to write the code to crunch the data to format the text).
But there’s more nuances. What about a couple words formatted within written text, simply to facilitate cross-referencing to an image? I don’t count those. And what about an image that’s a composite of many teeny snapshots? I count those as one image. And what about a table that’s got some visualization formatting associated with each cell based on data attributes? That might count as a table and not an illustration. Again, I had to write the code to crunch the data to format the text, so again I count that as an image.
So, overall, I get to around 250 images, out of 274 pages. Given that the Table of Contents, Preface and Index don’t have images, that gets close to the 50/50 split.
Types of Pictures
Returning to the topic of pictures, a few other stats are useful. The pictures split approximately 50/50 between pictures that are my own vs. pictures of other visualizations. I think it’s good to provide a grounding based on real-world precedence. For the pictures from other sources, I’ve tried to include URL’s to them, so readers and educators can easily find the online versions where available.
With regards to my pictures, some are new unique visualizations, some are sequences of pictures, such as showing the same visualization using different attributes, or zoomed in or so on. There’s about 80 different examples of visualizations with text that I created in the book. Some have been published before, on this blog or in research papers.
But I wanted to create some new content (why buy a book with old pictures). So there’s new, unpublished visualizations that will be making their first appearance in the book, including scatterplots of cars, an adjacency matrix of dialogue, a syntax diagram, a massive textual stem & leaf diagram, assorted tables with visualization characteristics, a data comic, some expressive lengthening examples, and a topic model visualization. Here’s eight examples taken from the draft:
Larger versions of these pictures will be available when finalized and placed on the publisher’s website with a CC license shortly before the book is released.