Review: Isabel Meirelles, Design for Information
When I’m asked for a good book about visualization, I usually try to change the subject. There is no book I really love, they all have their issues. But thanks to Isabel Meirelles, I can now give a straight answer: Design for Information.
In the interest of full disclosure: I was sent a free copy by the publisher to review. However, favorable treatment on this website cannot be bought, and most certainly not for the $40 this book retails for.
Meirelles is associate professor for graphic design at Northeastern University, who has done research on visual representation and metaphors. She has not published in the visualization literature though, and her credentials are clearly those of an artist or designer (creative work) rather than a researcher or CS professor.
The book is beautifully laid out and designed, and works well as a coffee-table book. But it has a lot more depth than many other books, coffee-table or not.
What sets this book apart from any other I have seen are two things: a clear separation of theory and case studies, and clear, concise, in-depth explanations of visualization techniques.
You Can Explain Visualization, Who Knew?
Perhaps my favorite part of the book are the little diagrams that explain visualizations by drawing over them or by adding a little schematic version to show how it works. I don’t understand why this is so rare, it clearly works incredibly well. In fact, I wish there were more of these even in this book.
Some of the examples also have a nicely structured list of meta-data, some of which is connected to the visualization with arrows. It sounds simple, and it is, but it’s very effective. Unfortunately, these are mostly done in the beginning of the book and trail off later. But the format is great, and I hope others will
steal borrow it.
There is an assumption that because you can see the visualization, there is no need for a lot of explanation, and in particular not for detailed or diagrammatic explanations. But I don’t get why that would be the case. Sure, a good visualization should be fairly self-explanatory, but there is also a lot of subtlety. And many visualizations take a bit of work to read. Meirelles shows how this can be done in a way that helps the reader understand without getting in the way or trying to explain the obvious.
In sidebars throughout the book, Meirelles explains the deeper theory, like encodings, perception, etc. The focus is clearly on the more general explanations and the case studies. But because the sidebars are there, the reader can decide whether and when to read the deeper information.
An interesting little detail is the place where data types (continuous, nominal, ordinal) are explained: in a two-page appendix. To be fair, ordinal data is explained once earlier, but surprisingly this actually works. It's not even clear why that appendix is needed.
Each of the six chapters (trees, networks, time, maps, spatio-temporal, text) consists of an introduction and case studies. The case studies explain the examples, both historic and current.
The distinction between the examples in the introduction and the case studies is not always clear to me. It seems that the introduction is more based on historical examples, but in some cases those are also used in the case studies. Either way, the general structure makes a lot of sense to me because it allows for a more general introduction and then a deeper dive into details when discussing the specific examples.
If there is something to criticize, it’s the focus on fairly simple visualizations (and ones that have historic examples). There is no real multi-dimensional data visualization in this book, and if you’re looking to learn about parallel coordinates and scatterplots, this is not the book for you. There is also limited analytical depth: only a few examples rely on interaction, and the data shown is mostly fairly simple. Where interaction is important, it is explained well, though.
For somebody looking to write a book about more complex techniques, and perhaps with more analytics thrown in, this is a fantastic book to draw inspiration from, however.
On the other hand, the book covers some visualization techniques that have so far been largely overlooked in the literature. In particular, the text visualization techniques developed by the Many Eyes team, like phrase nets and word trees, are really underappreciated.
There is a certain obsession in visualization with historical examples. Those are often good and interesting, but you can also get stuck in beautiful hand-drawn maps and diagrams that don’t exactly reflect what we can do today on computers. Many of them are intricate and beautiful, so showing them off is nice, but it’s also a trap that allows authors to take shortcuts and cheat their readers.
Meirelles strikes a good balance between historical and modern examples. She also does not throw them around just because they are famous or look good, but always ties them into a chapter.
She also nicely covers a broad range of current examples, including many from news media (in particular The New York Times), Twitter visualizations, etc.
Finally, A Good Visualization Book
Meirelles’ book is a great overview of visualization. It’s a beautiful coffee-table book. It covers a lot of ground. It provides enough depth to let the reader dig deeper. It has references to papers and books throughout for further reading. It explains. It explains!
A book that explains visualization. What a concept.