There's new competition for Swivel and Many Eyes: iCharts. A good name, to be sure, but will they live up to their promise of being "YouTube for Charts" (a claim Swivel also made in the beginning)? A first look at their website suggests that they likely will not.

iCharts is more similar to Swivel than to Many Eyes, both in their limited choice of charts and because they are an independent start-up (whereas Many Eyes is run by IBM). In comparison to Swivel, their vision is quite limited though ("to bring charts online"), and Swivel also had a clear idea how they would eventually make money from the very beginning. I don't see anything resembling that on iCharts (correction: iCharts wants to offer certain features, like embedding of charts in PDF files, to premium users. That's a start, but I'm not convinced that that will be enough.).

Seymour Duncker, one of the co-founders, talks about "lousy-looking charts" in a TechCrucnch50 presentation that they have embedded on their about page, but I don't see how their charts look any better than Excel did many years ago. In fact, the current Excel's charts are a lot prettier, and so are Swivel's. He also claims that there is no good way to create charts online and embed them in web pages, and that is simply not true.

iCharts offera the usual chart types: line charts, bar charts (including stacked), and pie charts (including a way to do concentric donut charts around a pie chart). There is some interaction with mouse-over labels and interactive filtering of axes. The use of the latter is kind of pointless because you can really only zoom in on one axis, there is no way to gain more insight into the data this way. It is possible to add annotations to the chart for explanations and to point to particular elements. This would be cool if it was possible for viewers to add those to the charts (like sense.us did), but currently they can only leave old-fashioned comments. Charts can of course be embedded in webpages, just like with Swivel and Many Eyes.

iCharts is clearly a very low-budget operation at the moment. Their whole website looks like very basic and unfinished, including their staff pictures and the horrific video introduction on their about page. Given the artistic credentials of Tyron Montgomery, another co-founder, one would expect quite a bit more. 

Another hint is their domain name: icharts.net (they also own the .org). The .com domain of the same name is owned by a domain squatter. It's not a good sign that they obviously don't have the money to buy that domain. It's not their fault that the domain is taken, or what it is used for, but it reflects badly on them (e.g., when people look for them and find the spam site). And if they become successful, the price for the .com domain will only go up.

They have just opened the site for a public beta, and it is obviously a bit early to tell what they will be able to do. But not only is what I can see right now not exciting, they also fail to present any kind of compelling vision. Bringing charts to the web was a good idea two years ago, but it's been done, and done better than what iCharts currently offers. And if it's only about the same three chart types all over again, I really don't see why the world needs another way to do this.

(Thanks for the link, Jorge.)

See also: Swivel vs. Many Eyes

One of the most common mistakes in chart design is to scale an area by two sides at the same time, producing a quadratic effect for a linear change. That overstates the larger numbers and produces a badly skewed chart. A little care and some basic high-school math can help avoid the problem.

The following detail from a information graphic produced by Princeton's International Networks Archive illustrates the problem (the numbers are presumably from 2002):

Comparing Starbucks ($4.1bn) and KFC ($8.2bn), the problem becomes clear: the difference is a factor of two, but the KFC logo has four times the area of the Starbucks logo (even more because one is square and the other round). This can be seen in a number of the graphics on that website, though they also have some where they scale correctly.

The reason for the problem here is the use of logos (or of images, more generally) to make charts look better. Scaling a logo in only one dimension (which would be done in a bar chart) does not work because the image would look stretched and ugly. So instead, the image is scaled in two dimensions, leading to a perceived difference that is the square of the actual difference.

In more general terms, a linear change (I will use a factor of two to illustrate this)

becomes a quadratic change.

Or, to employ a bit of high-school math: The area of a square A = a², with a being the length of side of the square. If we double the side length, we get A' = (2a)² = 4a²: four times the area. This is the case for many other shapes as well, including circles and circle segments.

That last part was actually the subject of a discussion I had a while ago with a rather senior visualization person. He did not believe that changing a circle segment's radius would lead to a quadratic increase in its area. It's easy to show, though: a circle's area is r²π (r being the circle's radius), the area of a circle segment that covers an angle θ is r²π·sin(θ). It is no more difficult to show that doubling the radius will quadruple the area than with the square above.

Florence Nightingale already knew this in 1858 when she developed her coxcomb chart (a predecessor to the pie chart): she represented the numbers of soldiers using the area, not the radius, of the circle segments.

A similar effect can be seen in the petal chart or star glyph, which connects points on a number of axes that radiate from a common point. Whether they are filled in or not, the impression is that of an enclosed area, and that changes in a quadratic way similar to the circle segments above.

The solution? Either use a better visualization (star glyphs in particular are very difficult to read) or scale your circles, squares, and other lengths by the square root of the value you want to represent. That makes the area scale linearly with the value that is to be represented.

SPSS recently released their new Viz Designer, a visualization engine built on Leland Wilkinson's work (The Grammar of Graphics and nViZn). The comparison with Tableau is unavoidable since both are based on the same underlying ideas. Right now, Viz Designer does not look good in that comparison. 

In fact, it would be easy to slam the program. The user interface has a certain engineering look to it that reminds me of early GUI programs. There is also very little exploration and basically no interaction with the created graphics. There is also very little intelligence about the data, like detection of categorical vs. continuous dimensions. But then, the program is built on a very powerful basis and there is a large enough organization behind it to carry it over the first few, rough, releases. And there is also room for a different approach than Tableau, even though I think they're doing a lot of things right.

One feature the folks who demoed the program for me stressed quite a bit is how the look and feel of a graph can be stored as a stylesheet and applied to other graphs to provide a unified look across a corporation. In a similar vein, visualization specifications can be stored as templates and applied to different data (with some restrictions that can be defined in the program). Both are integrated with a server, so that central maintenance of stylesheets and templates is possible, and users always access the latest versions.

Templates become a part of the library of visualizations that the program offers when the user opens a dataset and selects dimensions to be shown. This is an advantage and a disadvantage at the same time. While it makes good use of the underlying engine and provides the means for users to seamlessly extend the library of visualization templates, it also means that Viz Designer's elementary unit is a single chart. This is quite unlike Tableau, where it is easy to switch between visualizations, add new ones on additional pages, and combine them into dashboards. Viz Designer graphs can be saved as images or be embedded in SPSS documents and edited directly from there (by double-clicking the image).

The most obvious feature in Viz Designer is that the bottom half of the window is filled with code that describes the visualization in what SPSS calls VizML (Tableau calls their definition language VizQL) or GPL (the language defined in The Grammar of Graphics). The idea is that power users would be able to create or modify graphs directly in that language. While SPSS undoubtedly has users that are comfortable with programming and know its language very well, I have my doubts if those same users would be able to use a complex, declarative language to define visualizations. And even if, how visually effective will those be? Tableau steers the user in a way that makes it difficult to produce bad visualizations, and its user interface is powerful enough to allow a lot of flexibility without the need for tinkering with VizQL (though that would undoubtedly be a cool feature).

Viz Designer was made for producing graphics as the result of statistical analysis. As such, it does not currently have any exploratory tools. While this might be enough for people used to working with static graphs, it is a striking omission to somebody who is used to more interactive work. The ease with which the visualization can be changed, different dimensions added and removed, and especially the hierarchical layout of categorical dimensions (and the hierarchical treatment of time) make exploring a dataset in Tableau fun and interesting. None of that is currently possible in Viz Designer, and even something as basic as stacked dimensions on an axis requires some tricks or VizML hacking.

One area were Viz Designer is clearly ahead is its integration with SPSS. While that might seem obvious, it offers a vast array of data analysis capabilities and of course also allows them to tap into their existing user base. SPSS is also planning on hosting a repository of user-contributed templates on its website, so they will be able to learn from others and presumably discuss their creations. That is undoubtedly a good idea, and one that even Tableau could learn from (though to be fair, they have their own user conference).

Exposing the underlying engine and definition language is also interesting and might lead to some really interesting things. Involving their users is also a good thing and will certainly get people interested. In terms of features and interaction they still have a long way to go, though, and time will tell if Viz Designer can be more than a static back end to the actual (non-visual) analysis.

Colin Ware is a well-known researcher in Information Visualization (InfoVis), and I consider him the one with the most scientific approach in the field. His book Information Visualization: Perception for Design is widely read and used in visualization courses, and is without doubt the most thorough treatment of the perceptual and cognitive psychology foundations of visualization (and design, for that matter). He was also kind enough to contribute a list of influences to this site, where he mentioned the book he was working on at the time.

(Why) Is This An InfoVis Book?

My interest in this book comes from several terms mentioned in the title and subtitle. I think that visual thinking is marginalized in visualization, where we often present particular data rather than provide the visual means for solving problems. Design and design skills are also still underappreciated in InfoVis. Visual gist, narrative, etc. are things people think and talk about, but very little of that translates into the work being done in InfoVis.

The Book

This is a well-designed book by somebody who knows exactly what he wants. In the preface, Ware talks about how he placed the images in the text so they would appear where they are needed, without the need for "See Figure x" to send the reader hunting for the right image. He uses that to great effect to set up little experiments where the reader has to read the instructions at the bottom of one page and only sees the image when turning the page. Most figures really are were they belong, but some are not, and that is a lot more apparent when so much emphasis is put on figure placement.

The writing is vivid and very readable. This is a book for an audience with a vast range of backgrounds, and Ware does not assume much previous knowledge or a great tolerance for jargon.

However, the book also feels superficial in places. A lot of the usual basics are only skimmed over, and that helps make the book manageable and get to the important parts, but some readers will want to know more and will not be provided with many pointers where to look.

Chapter By Chapter

Here is a brief summary of every chapter, with some thoughts on each of them. I am only mentioning topics that I found particularly interesting and/or relevant, more things are certainly covered in the book.

Chapter 1 dives right into cognition and change blindness to discuss visual memory and introduces the concept of the world as its own memory, which we constantly query as needed instead of keeping a model in our heads. Ware compares the eye to a digital camera, which is generally a bad idea, but he makes it work by introducing the concept of the brain pixel. The mixed bottom-up and top-down aspects of visual processing are also discussed and used to present a first overall model of vision.

Chapter 2 describes the parallel processing that takes place in the brain, the different pathways involved and how eye movement planning works. This is the basis for a discussion of features that "pop out," i.e., pre-attentive features and some of the mechanisms behind them (Ware points out that pre-attentive is really a misnomer, but it is probably too late to change the term now). A thorough model of visual search is constructed from the underlying mechanisms described in the chapter. This is also the first chapter that has a section on concrete examples of applying the presented information to design questions in InfoVis and visual design more generally.

Chapter 3 starts out discussing spatial organization of information, then drills down into a lowest-level description of edge detection on the neuron level to quickly bounce back to high-level tasks like texture detection, pattern learning, and all the way to visual metaphors. That is a lot of ground to cover, and it feels a little superficial. This chapter in particular seems to call for more details or at least more pointers to further reading, which are missing (see below).

Chapter 4 deals with some of the perceptual effects of color and explains, among other things, why we can see more detail in greyscale than color images. There are a lot of useful figures that illustrate the described phenomena. The interesting connection between names of colors in most languages and the number of colors that can be easily recognized and distinguished is worth pointing out, and provides a good rationale for a limit on the number of colors used.

Chapter 5 goes into depth perception by discussing depth cues and the role of motion in depth perception. The fact that we really only perceive a 2.5D world (or even less, Ware argues it's closer to 2.05D) is discussed and why 2.5D design makes sense. This chapter also talks about how depth perception is only really necessary if we make use of it for movements like grasping, which Ware argues is the reason for many 3D technologies like CAVEs and 3D movies failing (or at least not being the revolution everybody thought they would be).

Chapter 6 is about 3D objects, geons, as well as short-term (working) and long-term visual memory. Here like in most other chapters, Ware offers design tips related to the content, which are not just tacked on the discussion of the underlying mechanisms. One thing that confused me about this chapter was the assertion that most of what we see is already in our heads - how does that work with what was said in chapter 1, that the world is its own memory? A bit more discussion would have been useful here.

Chapter 7 tackles visual and verbal narrative. It starts out by questioning the old adage that a picture is worth a thousand words. Of course, a thousand words (or even just a handful) can express many things a picture simply cannot, or not easily. So deciding when which should be used is an important question - at least in design, but perhaps also in visualization. But I felt that this chapter didn't quite live up to my expectations because it just stays too superficial. It talks about film directing and work done on assembly instructions for furniture by Tversky and others, but leaves the connection with more abstract kinds of representation open. I also found the treatment of sign language too superficial. While Ware talks about how it uses visual abstractions, it seems a strange choice for this chapter (it uses the same key parts of the brain as spoken language, despite its different modality), and there are many fascinating things about it (like the use of location for pronouns) that might inspire some new ideas in design.

Chapter 8 calls creative visual thinking "meta-seeing," and essentially switches the book's point of view from the recipient to the designer. Random scribbling is discussed as a way to find starting points, as well as recognizing figures by adding little "props," like beaks and eyes. It also talks about some fairly high-level design topics like spiral design and critique. This chapter in particular provides good insights into the design world for non-designers like InfoVis researchers.

Chapter 9, "The Dance of Meaning," is a review of the previous chapters, which I did not find necessary for a book of this length, even though it adds new examples and other information. There is an interesting example of using static representation vs. animation in a study of the behavior of humpback whales, which I felt could have been discussed in more detail in the main part of the book. Ware is a bit too modest in talking about his own work and misses some opportunities to illustrate and develop further some of the topics in the book because of that.

The book ends with an argument for the importance of design in the designed world we live in as well as touching on the fact that some of the basic human perceptual limitations and skills are changing because of exposure to new kinds of stimuli, like video games.

Further Thoughts

There are quite a few typos and other small errors in the book. Ware consistently misspells Anne Treisman's name, which is especially problematic because he uses it to showcase how a difference in orientation pops out (interestingly, he also misspelled her name in his previous book's text, but got it right in the references). I also found almost two dozen errata on my first reading.

There is also a lack of references, especially given the introductory nature of the book. While I like the sparse use of marginal notes and lack of footnotes (for the same reason figures should be where they are referenced), additional references in the back would not have hurt. Image credits could also be improved. For one of the images, the visible man is mentioned, but no explanation is given what that is or where it can be found. The same is true for TubeGuru, a planning system for the London Underground that does not live at an obvious URL (and there are many other sites of that name that do other things, like let users share videos). For an image sequence from the Powers of Ten movie, Ware mentions the names of the people behind the images, but not its title.

What is great are the connections to design throughout book. They really connect with the content of each chapter, and appear quite useful (to this non-designer). This is probably not a typical design book, but one every serious designer should read carefully to understand his/her profession in a much more profound way.

This is also an inspiring book. I consider it a good sign when I catch myself thinking the things described further rather than reading on, and that happened in every chapter of this book. I almost feel that every single chapter could be the synopsis for an entire book, with more details and ideas. That is all the more reason why there should be a lot more references in the book.

The book is listed as having 256 pages on Amazon, but it's really only about 185 (not counting the preface, index, etc.). I consider this a good thing, there are too many half-read books on my bookshelf that I will likely never finish. A manageable size makes a book much more practical and useful, though again based on the assumption that there will be plenty of pointers.

Is This A Visualization Book?

As an introductory text that covers a lot of ground on perception and cognition, and that draws connections with practical design issues, this book is hard to beat. If you have not read Ware's previous book, I would strongly recommend this one as a starter to whet your appetite. Perception for Design goes into a lot more detail on many of the topics covered (but does not cover all of them), but is also a lot more technical.

In addition to Ware's background, what makes this book relevant for InfoVis are the examples and the fact that most (if not all) the covered topics are directly relevant and applicable to visualization. In addition to the perceptual and cognitive topics, that is also true for the design issues that are mentioned. This book will provide a lot of thought-provoking and useful material, and many starting points for future research.

A paper on visualization criticism just appeared in the Visualization Viewpoints section of this month's Computer Graphics and Applications (CG&A). Authors are yours truly, Fritz Drury, Lars Erik Holmquist, and David Laidlaw.

The idea behind this article was to discuss the background of vis criticism a bit, talk about practical experiences, and offer a glimpse into a possible future. David and Fritz share their experiences teaching a class of computer science and design students, where critiquing played a big role. Lars responds to my criticism with a critique of critiquing. And I claim that criticism is something we're already doing (paper reviews) and need to do more to further develop the theory in our field.