What Is the Best Response to Bad Practices?

During the last two days, I spent a great deal of time corresponding with my friend Alberto Cairo after he informed me that he was hosting a public lecture by David McCandless at the University of Miami. Alberto and I are both critical of McCandless’ infographics. I am more passionate in my criticism, however, perhaps because I frequently and directly encounter the ill effects of McCandless’ influence. More than anyone else working in data visualization today, McCandless has influenced people to design data visualizations in ways that are eye-catching but difficult to read and often inaccurate. Also more than anyone else, when my readers and students talk about the challenges that they face in the workplace because their bosses and clients expect eye-candy rather than useful information effectively displayed, they identify McCandless as the source of this problem.

You can imagine my dismay when Alberto told me about the lecture. I argued that he shouldn’t provide McCandless with a forum for promoting his work unless he also provided a critique of that work during the event. Alberto’s position was that, as an academic and a journalist, he should provide a platform for anyone whose work in the field of data visualization is known, regardless of quality or the harm that it does. Further, he argued that his students and those who have read his books already know that he finds much of McCandless’ work lacking. My response to this was, “What about those who attend the event but are not your students or readers?” After the discussion, I found myself wanting to ask one more question: “What do I say to someone who tells me that his boss attended the lecture, and this exposure to McCandless’ work set his efforts to promote effective practices back by several years?” Even worse, what if he also says, “Steve, I encouraged my boss to attend the lecture because it was hosted by Alberto Cairo, whose work you’ve praised.”

To no avail, I pleaded with Alberto to provide a counterpoint to the presentation to make it clear to attendees that McCandless often promotes practices that are ineffective. I argued that without providing this counterpoint, he was abdicating his responsibility as a teacher and a journalist. He saw it differently. He replied that his indirect approach to combating ineffective practices is perhaps more effective than my direct approach.

Is Alberto right? Was it appropriate for him to host a public lecture by McCandless without offering a counterpoint? Should I become less direct in my criticism of harmful practices? Will they cease to plague our work faster if I do? What does your experience tell you?

Take care,


75 Comments on “What Is the Best Response to Bad Practices?”

By Kim Rees, Periscopic. October 28th, 2015 at 2:23 pm

I hesitate to respond as I know I will be sucked down this rabbit hole for days. Notwithstanding, I hope I can make a succinct point.

Critique has soft edges. Critique cannot be black and white. It allows for dialog and glimpses of alternate realities at the periphery. I once had a passionate debate with an esteemed colleague about why circles sized by radius were wrong. He convinced me that it’s not necessarily wrong, depending on how the design frames it. (I know, you probably just jumped out a window.)

From what I’ve seen of David’s work, I can’t say I’ve seen anything downright “wrong.” I may not always agree with his simplification of things or his methods, but who am I to say we should throw his work out wholesale? I think David’s work has been some of the most inspirational for our field. It’s welcoming and engaging and he makes it look easy. His work is a conduit for newcomers. And for that, I thank him.

You can take critique as far as you’d like, but it can’t shut down ideas. People will produce what’s in them to produce, and if it falls on craving minds, it will have a life. For instance, I think the work of Frank Gehry is loathsome. Many do. He widely criticized; yet he’s continually commissioned. His work, no matter what my thoughts of it, are still awe-inspiring. Will he kill the field of architecture with his little niche? No.

Should we try to eradicate visualization practitioners who don’t conform to what we believe to be “right?” In my opinion, that would do far more harm than good. We’d be killing off far more than poor practices. We’d be killing off new ideas, beautiful tangents, and avenues of exploration.

By Jay Lewis. October 28th, 2015 at 2:28 pm

I’d give Alberto the benefit of the doubt on it.

Inviting someone to speak at a university isn’t a direct endorsement of that speaker’s work or beliefs. The Q&A at the end of the presentation is an opportunity to challenge McCandless with questions about style, substance, and truthfulness in data visualization, and I have little doubt Alberto’s students will jump at the chance. Pitching this event to McCandless as a point-counterpoint debate would diminish the chances of constructive dialog because he might view it as a hostile environment and would be less likely to accept the invitation in the first place.

Here’s another way to think about it –

McCandless has reasonably high visibility, so people who take an interest in data visualization are likely to come across his work in one way or another. Assuming that’s the case, is it better for students to hear from McCandless in a forum hosted by Alberto, or would you prefer for them to learn about McCandless in a forum like a TED talk where there are no questions from the audience at all?

By Stephen Few. October 28th, 2015 at 3:09 pm


When we critique the effectiveness of a particular data visualization that was created to communicate a particular message, and there is clear evidence that the visualization either works or doesn’t, then the critique can definitely be black and white and should be. In what sense does a black and white critique eliminate dialog. It doesn’t. I can say that a particular data visualization doesn’t work for a particular set of reasons based on empirical evidence. The person whose work I’ve critiqued can then disagree with me and defend his design.

You apparently haven’t seen much of McCandless’ work if you haven’t seen anything of his that is downright wrong. Read any of my critiques of his work and you’ll see plenty. I’m not saying that we should throw out McCandless’ work wholesale. I’m saying that he should stop promoting ineffective practices, and until he does, we should caution people about his work. McCandless’ work indeed has served as a conduit for newcomers, and that is precisely the problem. People who do not know better emulate his work and, as a result, visualize data ineffectively. This results in enormous losses to organizations.

Critiques do not shut down ideas; they evaluate ideas. If ideas are shown to be erroneous, they are exposed as such. This is the essence of science.

A critique of Frank Gehry’s work is not an appropriate analogy for this discussion. To the degree that his buildings are dysfunctional in some way, they have little effect on our lives and the world at large. People of influence in disciplines related to data sensemaking, however, can have an impact on decisions that affect us all. Decisions that are based on misunderstanding or misinformation cause harm. I don’t know how closely you work with data visualization practitioners. I do work closely with them, and I see firsthand the harm that is done when people follow many of the practices that McCandless has popularized.

We should definitely strive to eradicate data visualization practices (not practitioners) that are ineffective and therefore harmful. I’m not talking about practices that don’t conform to my style or preferences, but practices that research has shown to be ineffective. McCandless’ work is rife with such example. Eradicating such practices would directly eliminate a great deal of harm that is being done today in the name of data visualization. I’m intimately familiar with this harm because I work closely with data visualization practitioners who struggle to combat it in their organizations. Perhaps the work that I do provides me with a perspective on this matter that you lack.

If you can identify any specific “new ideas, beautiful tangents, and avenues of exploration” that McCandless has introduced, please share them with us. I’m aware of none. If you reveal something of his that is indeed worthwhile, I will embrace it and thank him for it. I will not, however, stop criticizing him for the many bad practices that he has continues to promote that cause harm.

By Robert Monfera. October 28th, 2015 at 3:15 pm

I was surpised about that, in a similar way when I saw this otherwise nicely produced video, with a time series style visualization that has curves evoking time travel here: https://youtu.be/AdSZJzb-aX8?t=226 and also posterized, ‘flat design’ views here: https://youtu.be/AdSZJzb-aX8?t=215 -> in contrast to the points made by Tufte at the beginning and end of this video.

So much of journalism is, due to the nature of those segments, about attracting eyeballs and catching the attention of the general public. A good fraction of which, on a long commute after an exhaustive workday, ‘consumes’ media exclusively or primarily as entertainment rather than understanding economy or public policy. I’m not suggesting that engaging journalism and clear information conveyance can’t be combined, but not all outlets try to serve both of these purposes. (imo)

By Stephen Few. October 28th, 2015 at 3:17 pm


It is true that giving someone a platform to speak is not necessarily an endorsement of his work. The problem is that nothing was done during the presentation to challenge McCandless’ work, which means that people were influenced by McCandless without having an opportunity to hear that his work is often flawed and harmful. Essentially, hearing from McCandless at the University of Miami was no different from hearing his TED Talk. That’s the problem. Had the event been set up to provide a counterpoint to McCandless’ work, I would have no problem with it at all.

By Stephen Few. October 28th, 2015 at 3:27 pm


The video that you mentioned, which I commented about when it first came out, is deeply disturbing. It contains contradictory information without even being aware of it. The goals of infographics as defined by some of the infographic designers who appear in the video are not even met by their own work. For instance, they talk about their purpose being to make something that is complex, which might take days to understand, simple and clear. As they are making these statements, however, examples of their graphics appear in the video that are incomprehensible.

By Kim Rees, Periscopic. October 28th, 2015 at 3:50 pm

I don’t think anyone should judge a visualization based on a still (or few frames) used in a film. All of those images are taken out of their framework — even the Mersenne diagram of 12-tone harmonies — which is an elegant and useful visualization, but in the film is mere garbage because you don’t know what it is.

By Stephen Few. October 28th, 2015 at 4:11 pm


Knowing Robert, I suspect that he is familiar with the Mersenne visualization of 12-tone harmonies apart from the video, but perhaps not. We’ll let him tell us if his comments about that visualization in particular are misplaced due to insufficient understanding of the context. I have seen this swirly line graph in context and found it pretty but incomprehensible. It is a timeline, but the lines suggest that some things go backwards in time, which makes no sense. Many of the visualizations in the video can be shown to be ineffective, even when the contexts are fully understood. This, however, is tangential to our topic.

By David Mendoza. October 28th, 2015 at 7:04 pm

Note that Stephen never said that McCandless shouldn’t be allowed to give his lecture, just that the audience should be made aware of the contentious nature of his work. If done in a tactful and objective way, this should neither offend McCandless nor prejudice the audience against his work.

But I think Stephen makes an even more salient point that there is too much relativism among people who make data visualizations. Many design choices are not simply a matter of personal preference. There are right and wrong design decisions and when we see errors that affect the accuracy of a visualization, we should point it out. Or as Alberto recently and more succinctly put it, “If you see BS, say ‘BS’!”

And some of McCandless’ work rises to the level of BS. See this example: https://politicalmath.wordpress.com/2009/12/03/bad-visualizations-david-mccandless-lets-politics-get-in-the-way/

Or this: http://www.informationisbeautiful.net/visualizations/colours-in-cultures/

Acknowledging that McCandless’s work is replete with mistakes is not eradicating him. It’s edifying him and hopefully everyone else.

By clancy. October 28th, 2015 at 7:28 pm

A better approach would have been to have 2 presenters 1 after another representing the 2 different camps and allow the attendees to make up their own minds.
OR a follow up lecture with a critical review of both methods to elaborate pro’s and cons.

By Fernando Cucchietti. October 29th, 2015 at 1:30 am

I believe that if Alberto thinks badly of McCandless’ work hosting and allowing him to present unbothered is a grand gesture. I’ve seen people walk in at the QA just to shout BS and leave…
But the question is wether there should have been a counterpoint session. This depends on the speaker, who might feel ambushed or mistreated no matter how tactfully you do it. I love being called out in my mistakes, sometimes I learn and other times I get to defend the decisions. But I don’t have a clue about what McCandless would do.
After an open discussion about the design choices of a particular visualization, I think the next most beneficial thing for the students is a lecture by Alberto and an internal workshop redesigning those offending cases. Which he probably did, either before or after.

By Andrew Vande Moere / infosthetics. October 29th, 2015 at 4:21 am

In my simplistic view, you interpret the claims and visuals of McCandless literally, whereas they are meant, and should be understood and evaluated, in a different frame. Whereas this frame might conflict with some parts of the data visualization dogma, it still has a strong and worthwhile purpose. Although McCandless himself does not necessarily articulate this purpose, it is still there and is recognized by others.

“What do I say to someone who tells me that his boss attended the lecture, and this exposure to McCandless’ work set his efforts to promote effective practices back by several years?”

I give the following suggestion: “Your boss misinterpreted McCandless’ talk as data visualization dogma, although McCandless never presents his work as dogma. I guess you have now have the perfect opportunity to talk with your boss, and discuss the difference between form and function. You’ll find plenty of literature to support both views.”

By Anthony. October 29th, 2015 at 5:02 am


I think we need people to continue calling out bad practices and using Design by Redesign (makeover) techniques to show where gains can be made. This provides practitioners something to point to and to learn from. I think it is essential.

But I don’t belive that through education/calling BS alone is sufficient — or ever will be.

I think the greatest gains will be made by making good practices the defaults and making it difficult or impossible for people to choose bad ones. We have to make the right choice easier than choosing the wrong. I think the best thing we can do is encourage software libraries, vis tools, and process-makers to help us make easy, good choices. To keep us on the road and safe. Vis tools like Tableau and software libraries like Seaborn are on that route to providing smart defaults — we just need more of it!

Constraints are a good thing. We cannot continue to blame human error when it’s the majority of the population. That is design error (source: Don Norman). Imo, this is the real problem with PowerPoint– it has no safety .. it is too easy to shoot yourself in the foot without knowing it. Its defaults basically point the gun at you.

Food for thought: here at IEEE Vis 2015 I learned that there are still many VIS papers this year that still use the rainbow color map — a practice that has been panned for … How long? A decade? More? And I wonder why? How is it that we can’t win here amongst the visualization geeks — the scientists? If we can’t get rid of this bad practice here what hope does that leave us for the rest of the world? Have we not tried hard enough?

By Stephen Few. October 29th, 2015 at 8:51 am


I’m having a hard time making sense of what you’ve written. Regarding data visualization, I’m dogmatic about one thing: a data visualization should inform and do so effectively. By effectively, I mean that it should inform in a way that is as easy as possible to understand and in a way that is truthful. What is the “different frame” that we should use to interpret McCandless’ work? The title that he bears — “journalist” — provides a frame. His work is supposed to be read as journalism. A journalist strives to express the news in a way that is accessible to his readers and is truthful. He fails to fulfill this role. He is not a good journalist by any definition that I know.

Your answer to the boss’ question suggests that you don’t work closely with people who struggle to communicate data within their organizations. They put up with nonsense constantly as they deal with the naïve demands of people to present data in ineffective ways. “Make it more colorful! Give me bubbles, lots of bubbles!” They are not looking for more reasons to explain to the boss why that set of five values shouldn’t be shown in a treemap. They do not deserve to have their efforts set back by the dysfunctional desires and expectations that McCandless creates.

By Stephen Few. October 29th, 2015 at 9:01 am


I want to respond to your frustration about the IEEE VisWeek conference. Ineffective data visualization practices are still rife in the papers that are presented there because few of the professors and students who write the papers have been trained in data visualization. Shocking, but true. Also, few of them approach data visualization as a science. Instead, most of them approach it as computer programmers. That’s their training. They write code. They might be brilliant coders, but they are applying their skills to a field of study that they don’t understand. I’ve been fighting this for years and have made little progress. Why? Because too many of IEEE VisWeek’s leaders fit my description above. The few data visualization researchers who understand the field don’t have enough influence to turn the ship.

Regarding better design defaults in the tools that we use to visualize data, it doesn’t help when the few tools that have actually made an effort to create effective defaults give into the influence of people like McCandless and introduce senseless features to appease them. Tableau did this when they added “Packed Bubbles” and “Word Clouds” to their chart library. When Tableau introduced these features in version 8, I felt much as I do now in response to the fact that Alberto hosted McCandless at the University of Miami without providing a counterpoint. I felt my efforts and those of data visualization practitioners throughout the world being undermined. What we do is too important to take these matters lightly.

By JBJNR. October 29th, 2015 at 9:38 am

To suggest McCandless is the source of all our data visualization woes amounts to a gross exaggeration. There are lots of proponents of infographics, and there have been lots of bad charts and useless eye candy since before PowerPoint was even invented. You are tilting at windmills by going after McCandless. He has his first amendment rights, and I see more benefit than harm by letting him speak at a conference. I think you need to let this one go.

By Anthony. October 29th, 2015 at 9:40 am

Stephen. I’m not frustrated about VIS and I’m sorry but I don’t agree with your assessment of the participants. I thought it a good example to make my point, but maybe it wasn’t the best choice.

Your post is entitled how to end bad practices and that’s what I tried to address. I have found your posts that call out bad practices helpful. So thank you for that. I think we should continue to call out bad practices, but that calling it out alone will not turn the ship.

Lastly I find this recent piece inspiring and fits well into this discussion : https://medium.com/@hint_fm/design-and-redesign-4ab77206cf9

By Stephen Few. October 29th, 2015 at 9:46 am


Don’t misquote me. I am not blaming McCandless for “all of our data visualization woes.” Stick to what I’ve said. McCandless is an influential proponent of bad data visualization practices. He gets a great deal of exposure, which gives him a great deal of responsibility. I’m saying that he shouldn’t be given a platform without providing a counterpoint.

By Dale. October 29th, 2015 at 9:47 am

There are many reasons to invite someone to give a lecture at a university, including belief that their work is excellent, important, influential, etc. possibly wrong and misguided even. I don’t see anything wrong with the invitation – EXCEPT that there such better options. A panel format, paper + discussant, interactive discussion session, etc. I do think it is fair to question why a better format was not chosen. I would stop short of urging that the invitation not be made. I would only advocate for a more meaningful environment where differing viewpoints could be heard.

By Stephen Few. October 29th, 2015 at 9:51 am


What part of my assessment of IEEE VisWeek is inaccurate?

By Anthony. October 29th, 2015 at 9:58 am

Stephen, I don’t want to go there. I can see you don’t like my VIS or Tableau examples. That’s ok.

I’d rather focus on the macro goal: If the goal is to curb bad vis practices I think it would help our community to look at how other industries prevent accidents and encourage good practices. Let’s make it so easy (and,,, fun!) to get it right.

By Anders. October 29th, 2015 at 9:59 am


Although I don’t think every speaker who isn’t following best practices or what research has demonstrated requires a public critique, with someone who’s as contentious a figure as McCandless I think it’s an excellent pedagogical technique. Having two very different points of view clashing on stage can be an extremely effective way of educating folks — both people who’re attending and those who watch it online. It’s particularly useful for novices such as myself. Wrapping your head around the rules of data viz can be a bit daunting at first, especially when a lot of practitioners ignore them. Getting to see a systematic critique followed by a vigorous defense is far more illuminating than responses to questions from the audience (especially if there’s a body of research that shows that the speaker is dead wrong). It also can make for great theatre, which increases the odds the lessons will stick.

By Stephen Few. October 29th, 2015 at 10:15 am


You are mistaken. I like your VisWeek and Tableau. They are appropriate. What I don’t understand is your objection to my comments about VisWeek. You pointed out a problem with research that is presented at VisWeek and questioned why it still exists. I provided an explanation for why that problem exists and you objected to it. Why? If you can’t explain your opposition, that’s fine, but say so. If you’d rather not express it in this forum, then express it directly to me. I’m curious.

I think your suggestion that we focus on how other industries are combating bad practices successfully is a good one. Do you have examples?

By Stephen Few. October 29th, 2015 at 10:21 am


I appreciate your response. The purpose of my critiques is always pedagogical. I regret that Alberto, who is a teacher, did not make the event at his university an opportunity for people like you to learn best practices. Unfortunately, what Alberto did was give McCandless a platform for promoting his work, even though Alberto opposes much that McCandless’ work exhibits.

By Stephen Few. October 29th, 2015 at 1:27 pm

I’ve observed that opinions on matters like this are strongly correlated with the nature of people’s involvement in data visualization. For example, those whose involvement is academic tend to have a laissez faire attitude and can’t understand why this matters so much to me. Those whose work involves data sensemaking and communication to support decisions in organizations tend to share my concern. This latter group—my primary audience—feels more strongly about this because they are faced with the harm that someone like McCandless does when he promotes ineffective practices. If you work for the WHO or the CDC, two of my clients, the failure of a data visualization to inform effectively can result in lost lives. Even if you work for a corporation that sells a product that doesn’t have a life or death effect on people, you still care when the failure of a data visualization results in lost revenues or a lost business opportunity. This isn’t an academic discussion. Whether you use the area or the diameter of bubbles to encode their values, or better yet, use bars to encode them in a way that people’s brains can easily decode and compare, is a choice that matters. It has consequences.

By Alberto Cairo. October 29th, 2015 at 1:35 pm

Yes, but Steve, you’re assuming that McCandless’s work is done to support decision-making. It isn’t, and he doesn’t present it that way. I am trying to put my thoughts together, and will start sending responses soon.

By Stephen Few. October 29th, 2015 at 1:50 pm


No, I am not assuming that McCandless’ work is done to support decision making. I am arguing that his work has an influence on people who rely on data visualization to support the decision-making process.

I assume that as a journalist it is McCandless’ job to inform. If so, he does his job poorly. If the purpose of his work is not journalism (i.e., to inform us about news events), shouldn’t he clearly state what his purpose is so that people won’t rely on his visualizations for information? His work definitely does not effectively support decision making, yet he speaks at analytics conferences and doesn’t bother to point out that his work should not be emulated if you’re visualizing data in support of decision making. What the hell is the purpose of his work? I’ve relied on his own statement that he is a journalist. If he’s defining journalism in a new way, perhaps he should tell us.

By Alberto Cairo. October 29th, 2015 at 1:56 pm


Even if I have to finish proofreading ‘The Truthful Art’ before the end of November, I am going to devote some time to respond to this. I will try to address these topics:

(1) David’s visit to UM
(2) Your approach to criticism vs. mine
(3) My opinion about David’s work

Let me address (1) in this comment.

I didn’t really brought him to Miami. He was going to be in town for other reasons, we communicated over e-mail and he generously offered himself to spend some time with students. I seized the opportunity. I like my students to be exposed to —to plagiarize William James— the “varieties of the visualization experience”, no matter how far those varieties are from my own view.

David gave a talk in the morning. The audience was 90% students from my school, who HAVE been exposed, ARE being exposed, or WILL be exposed to elementary visualization principles, as my infographics and visualization class is part of the core of our programs. In my clases we constantly critique graphics, and that includes some similar to McCandless’s controversial ones. We even did a very detailed critique of the Billion Pound-O-Gram once or twice: students analyzed the data, and the way to make the display better and clearer.

The other 10% of the audience were professionals I know; most have read visualization books like mine or yours. They also are capable of seeing a presentation like David’s and taking his words (or anybody else’s, for that matter) with a grain of salt. Therefore, your fears are baseless.

I’d also say that that even if this weren’t the case, it wouldn’t be a problem at all for me. I am fine with people being exposed to ideas I consider dubious (and I didn’t see many in David’s presentation, to tell the truth), to later reason why they may not be so good. That’s part of anybody’s education. By the way, I’ve made much worse graphics than David has in my career, and I learned from them. That includes 3D exploded pie charts with shadows underneath.

Finally, and just to finish point (1), I’d like to add that I invited you and David to come to the University of Miami, sit together for a couple of hours, and have a conversation moderated by me. David has already accepted, depending on data availability. I hope you’ll be up to it. I think that it’d be very informative for the community.

(2) Your approach to criticism vs. mine

During our e-mail interchange, we talked about your approach to visualization criticism versus mine. I am going to borrow a bit from my own messages.

We do have different ways of addressing things we disagree with. Yours is direct, sharp and even harsh sometimes, mine is oblique, based on either:

1) …meeting people who do things in ways I consider puzzling, learn about the reasons why they do them, and then, if I haven’t changed my mind, try to convince them to do otherwise,

2) …or trying to redesign their work in a constructive manner. On this, I will refer to Viégas’s and Wattenberg’s fantastic article, which summarized my view well: https://medium.com/@hint_fm/design-and-redesign-4ab77206cf9#.4b09uzdkp

3) …or writing about the general problems of their work in my books and articles, often (but not always) by abstracting those problems, rather than referring to specific graphics.

It is still to be seen which strategy is better in the long term. Based on what we know about the human mind, though, I have the hunch that mine will be much more effective than yours, as people are put off when dealing with articles like the ones you wrote on McCandless. I know this first-hand, and it is a shame. It has happened even with my own students, even if I always explain the context of your articles. You preach to the choir, and may not be convincing many. The evidence I have for this is anecdotal, of course so, needless to say, I may be wrong.

Niceness isn’t a pose or a desire to be liked. It’s a persuasion strategy. It is often said that it is wrong to focus on the tone of a criticism when you should be focusing on its content instead. I disagree. Tone matters, and it matters a lot if you really want to change people’s minds, like you do.

(3) My opinion about David’s work

I am torn on this one. I still don’t like much of what appears on his first book, and part of what appears in his second one, and that’s why I use part of it for constructive critiques in my classes. I think that a noticeable portion of his work is either too simplistic or misleading because the data behind it has been mishandled. Some other readers of yours have pointed out some instances of this. However, I don’t think that this rules out his work. I have made worse mistakes in the past, and I learned from them, little by little.

I agree with Andrew Van De Moere’s first paragraph above. Also, perhaps we should try to find a way to understand why David’s work is so appealing to so many people. Really, I am curious about it. This includes people I know who work in quantitive and scientific fields, people who know a lot about Math, and stats, and science. I find people’s tastes fascinating and worth respecting. If so many think that these graphics are attractive, if those graphics make them stop, and read, and feel curious about the topic covered, and prompted to learn more, there’s something to learn about them, whether you or I like or not.

Steve, I hope that you and your readers will understand that I’ll refrain from replying to any further messages in this thread. I have tons of projects to grade and a book to finish proofreading before the end of November. Thanks for a great discussion.

By Stephen Few. October 29th, 2015 at 2:05 pm


To set the record straight, you have not invited me to sit with David to have a conversation moderated by you. Yesterday, I suggested that I would be have been willing to engage in a public debate with McCandless. You responded that this was a great idea and that you would gladly serve as the moderator. That’s the last that was said. Your statement above makes it appear that you came up with the idea of a debate, that McCandless readily accepted, and that I did not. That’s not how it went.

By Andy Cotgreave. October 29th, 2015 at 3:13 pm

Hi Steve! You ask: “Should I become less direct in my criticism of harmful practices?”

My answer is “Yes” because what is “harmful” to you is not “harmful” to all.

I first challenge your assumption that McCandless is harmful. It comes down to your reply to Andrew Vande Moore. You said: “I’m dogmatic about one thing: a data visualisation should inform and do so effectively.”

Who says “data visualization” should inform effectively? Which of the two words “data” and “visualization” proscribe effective communication? I look at the work of Stefanie Posavec and I see data visualization. I see art. I see inspiration. It’s data, visualized. I look at McCandless and I see data, visualized. It doesn’t inform. but it’s beautiful, and fun to look at. There’s no harm in that, per se.

“Harm” comes into it if we assume that McCandless’ approach is the right way to make effective, informative visualizations. We should, and do, challenge that assumption.

The thing is, I’m not sure McCandless makes that claim about his work. I just rewatched his TED talk. He describes curiosity and exploration. He doesn’t claim it’s the best way to do things. He talks more about his personal drive to explore the data for himself. Sharing it is almost a secondary concern.

Most data visualization is undertaken to inform effectively. Alberto’s course, presumably, focusses on this approach. A course focussed only on functionality would be a sad thing indeed: students need to learn about engagement and beauty, too. My assumption would be that Alberto’s students are well enough educated, and presented with enough opinions that they make their own judgements. I’d also hope that Alberto also encourages his students to explore online resources. I very much hope he points them all to this blog and this fascinating debate.

Summary: yes – be critical of harmful practices when they pretend to be something they aren’t. Does McCandless claim his work is anything more than fun, beautiful stuff made from data and built to satisfy his curiosity? I’m not sure he does.

By Stephen Few. October 29th, 2015 at 3:29 pm


Yes, McCandless claims that his work is something more than fun. He claims to be a journalist.

If you reread what I’ve written, I’ve made it clear that I am not concerned about Alberto’s students. They have the benefit of his instruction. McCandless’ lecture was a public event, open to anyone. I’m concerned about the people who don’t have the benefit of Alberto’s instruction. Alberto had an opportunity to provide a counterpoint to McCandless’ work for them but chose not to. According to his statement above, he chose not to because those folks can figure it out on their own.

By Stephen Redmond. October 29th, 2015 at 4:12 pm

Considering the dross of “debates” that are airing at the current time, I think that a discussion between Steve and David, chaired by Alberto would be pretty awesome.

Not sure if it would get as many hits on YouTube as Adele, but it would an interesting bar chart ;-)

By Stephen Few. October 29th, 2015 at 4:22 pm


To be clear, I would welcome an opportunity to have a true public debate with David McCandless. Not a debate like the ones that we’ve been seeing among presidential candidates in recent weeks, but one in which each party must respond to questions and are subject to immediate fact checking. I’ve been having a one-sided debate with him for years.

By the way, McCandless was involved in a debate once that, if I’m not mistaken, aired on one of the BBC’s channels. Unfortunately, what was shown was highly edited, but in the portion that was aired McCandless was challenged by his opponent in much the same way that I would challenge him. Spoiler Alert: McCandless found it difficult to respond.

By Stephen Few. October 29th, 2015 at 4:33 pm


In my previous post, I failed to respond to the first part of your comments, which I’ll do now. That McCandless does harm is not an assumption. I am personally acquainted with the harm that he does. I’ve already made this point clear. The harm that he does is not to me. If I were concerned only for myself, I would merely ignore him. The harm that he does is to people that I care about. These people are my readers and students who struggle to correct his negative influence and to countless others who follow the example that he sets.

If you are seriously questioning whether we can say that data visualization should inform, then there’s no need to discuss this further. Let’s bear in mind that the term “information” contains the term “inform.” In the introduction to his first book, McCandless writes: “Every day, every hour, maybe even every minute, we’re looking and absorbing information via the web. We’re steeped in it. May even lost in it. So perhaps what we need are well-designed, colourful and — hopefully — useful charts to help us navigate.” This suggest to me that his intention is to inform. The fact that he calls himself a journalists suggest that his intention is to information us about the news. If McCandless defines data visualization is some way that means something other than a visual representation of data that is used to help us understand data (i.e., to be informed), then he should clarify what he means by the term. In my books and courses, I define the term quite explicitly, and then I help people learn how to do it well based on that definition. I find it interesting that so many people suggest that McCandless means something different by data visualization than I do but no one seems to know what that is. Wouldn’t it be nice if he told us.

By Stephen Few. October 29th, 2015 at 5:01 pm


I’ll respond to your three points in turn.

1) McCandless’ lecture at your university was advertised to the general public. As such, you had no way to restrict it to your students and readers. Apparently this doesn’t matter, however. You wrote, “I am fine with people being exposed to ideas I consider dubious…to later reason why they may not be so good.” I disagree. I believe that it is my responsibility and yours to provide guidance. You provide guidance to your students because they need it. By failing to provide a counterpoint to McCandless’ talk, you missed an opportunity to provide guidance to those who aren’t your students.

2) As one of the earlier commenters pointed out, you once wrote, “If you see BS, say ‘BS’!” That is what I do. My critiques of the work of people like McCandless are perhaps more damning than yours because I am frequently confronted with the harm that is caused by his work. You work at a university. You don’t see what I see. I make no apologies for expressing anger towards those who undermine data visualization. I don’t go out of my way to offend anyone, nor do I go out of my way to play nice with someone who is doing harm. Any of us who claim expertise and put our work out there for the world to see invite critique. We should welcome every critique that is well reasoned and accurate. My critiques are always thoughtful, well reasoned, and accurate. I take the time to do more than criticize. I explain the problems in great detail and offer solutions. Like you, I redesign visualizations to teach people how they can be made more effective. I put an enormous amount of time into these efforts and do so without charge. I don’t do this lightly.

You said, “Niceness isn’t a pose or a desire to be liked.” Actually, it often is. Too often, I have looked upon a smile that was there to conceal a dagger. I loathe disingenuous behavior. As you know, I am a very nice guy. I am very generous with my time, as you know given the many days worth of effort that I put into reviewing the draft of your new book. When I’m angry with someone, however, I don’t conceal the fact with niceness. I express my anger honestly and directly. My style doesn’t work for everyone, nor does yours. I’m fine with that.

3) I think I know why McCandless’ work is so appealing. It’s candy for the brain. It provides a sugar high. It’s fun. It doesn’t ask us to think very hard. It is superficial. In other words, it is cognitively undemanding. Brains that are undisciplined — actually all of our brains at times — love candy. I unwind with television versions of it most evenings when I become too tired to read. When McCandless begins calling his work candy rather than journalism, I’ll no longer be concerned with it.

By Elijah Meeks. October 29th, 2015 at 8:15 pm


Throughout your responses you return to the claim that anyone who disagrees with you is not a professional data visualization practitioner–that they must be an academic or a dilettante. I disagree. I think there to be a significant number of practitioners who look at McCandless’ work not as libertine regression, but as innovation, and who conversely look at your own work not as the pinnacle of data visualization theory, but rather as stolid advice for simple and effective charts. Are his data visualization products successful? Perhaps, perhaps not (No, it’s not that I’m demonstrating academic nonchalance, it’s that the question of success in visual rhetoric depends on the definition of the audience–we are not all building BI tools for busy executives). But regardless of how we might feel about the individual things a McCandless produces, I think it’s clear that the advancement of the field, by which I mean the establishment of new techniques that prove effective in a wide variety of data visualization uses, will come from someone like McCandless, who experiments and courts failure.

The subset of the field that you focus on in your professional capacity is not concerned with advancing the sophistication of data visualization methods. That’s fine, it’s a big subset, and more importantly it’s a group that needs the effective and clear advice that you provide. But I think it’s equivocating when you present that group as a stand in for the field. I think data visualization needs effective radial charting, network diagrams, abstract geospatial visualization, and yes, even effective word clouds. By labeling the work of someone like McCandless as degenerate, and dangerous to the field such that it could set it back for years, or that it might infect a decision maker and make the life of some analyst somewhere more difficult, I think that’s ignoring the need for innovation, risk-taking and advancement in data visualization, a field that really is very young and desperately needs such activity.

By Kelly. October 29th, 2015 at 11:04 pm

Kathy Sierra refers to this brain candy as ‘Easter Eggs’ in design. She did a good write-up on this a few years ago: http://headrush.typepad.com/creating_passionate_users/2005/05/the_case_for_ea.html

I think this is my favorite point she makes: “Until you’ve nailed the fundamentals–the things users want, need, and expect–don’t bother trying to “surprise and delight” users. That just pisses ’em off.”

By Alberto Cairo. October 30th, 2015 at 6:15 am


Remember that I am both an academic and a practitioner. Like you, I consult with companies and have worked in media organizations. I also design infographics a freelancer. Please don’t think that I live in an Ivory tower. We don’t have those in Miami.

I have found David’s influence many times. Often, I use it as a teaching opportunity, or as the topic for a discussion. In other cases, the influence is good, as David’s graphics are sometimes engaging and fun. I actually compare his most humorous work to Nigel Holmes’s in that regard.

I have the hunch that if people don’t get inspiration from David when designing posters or large superficial graphic summaries, they’ll find it elsewhere. I actually prefer people to be influenced by him than by those horrid marketing “infographics” that have become sadly so common in the past decade, and that taint that noble term.

I also believe that cognitively undemanding graphics may have their role, often if they are complemented with deeper coverage afterward or with links to sources that readers whose interest has been raised by the graphic can follow.

I feel tempted to use a term coined by Stephanie Posavec –I believe– to refer to David’s style of graphics: data illustrations. Based on what I heard in his presentation, I think he’d find it appropriate.

Regarding your being nice. Indeed you are. And your input on my new book has been invaluable.

By Daniel Zvinca. October 30th, 2015 at 6:38 am


I personally don’t see the need of a counterpoint for a public lecture of McCandless.

I doubt he can influence people to design data in the way he does. His work cannot be a standard or a reference while it does not show any principles or rules. I think that newcomers in data visualization field are safe. Sometimes his eye-candy work gives me a motivation to redesign or a subject for a small debate. Nothing more.

By Mike Hobson. October 30th, 2015 at 7:46 am

I’m not trying to be a keyboard warrior when I type this, but I will give my ten pence on the discussion…

From my view, and I own one of his books, he’s a great ambassador for the extreme lengths you can go to in showcasing different ways to look at data. They are specialist, one of a kind visualisations that highlight different ways you can make sense of a large amount of data. He’s not trying to say ‘this is how you do it’ and he’s not trying to say ‘line graphs and pie charts are not good enough’. However, if he put the book together using just those graphical representations, then it wouldn’t sell very many copies. What’s he’s created is a couple of books that do a very good job of showing people you can make some beautiful, fun designs with a little bit of thinking outside the box and a lot of coding. He deserves the high level of respect he’s given for this work. And yes, I know what we really strive for is repeatable, attractive and reliable solutions. That’s why tools like Qlik, Tableau, SAP, Board, Birst, SAS, Cognos and so on are around. It’s not about one-off coding solutions, it’s about ease of use and delivering multiple solutions. Though I will say some of those solutions are sometimes sold forgetting a key principle – single version of the truth (it’s almost vanished in the race to sell into departments with the ability to upload your own spreadsheet).

In the real world i.e. away from published media and actually delivering visualisations, it’s completely different. And yes, I’ve had to deal with people saying, ‘but can’t it look a but sexier’ in the past (more on this in a moment – this appears to be turning into a big piece!). The main point about data visualisation, for me, is that they should always be delivered through a lifecycle project process and a team effort. Yes, you need to get the data right, stage it properly and ensure there is 100% trust in it. But you also need to ensure it’s not just you delivering it as a quick one-off. It takes a team of people to view ideas objectively and constructively. And it takes a defined process where you are willing to revisit the solution to tweak or change it to get it right. There is no point one person delivering a visualisation and walking into the sunset without getting that correct feedback. Yes, the feedback can seem harsh or uncaring, but it’s also key to your work being accepted, trusted and used. I for one would not want it any other way.

Going back to those people who say ‘but can’t it look a bit sexier?’. You know what? They’re right, it can. It’s all down to the pinnacle of what we’re trying to achieve – adoption (as in your solution being used constantly, not obtaining extra children). However, we have a duty of care to those people we are delivering a solution for. Yes, sometimes expectations have to be managed, but you know what, where’s the harm in including them in the design process? That way you are ensuring the end user can’t get away with just saying ‘I want it like that’ whilst pointing at a graph of kittens and having no further input. It gives them some responsibility it, keeps them engaged, makes them think about the end result more and enables you to manage their expectations and perceptions along the way.
So when customers say to me ‘can I have that in red, that in green, that a bit bigger, and that visualisation is a little underwhelming/over the top’, we should always strive to say ‘OK’, roll our sleeves up and think about it a little more. And if a manager does say they really want pictures of kittens or something equally as terrible in their board reports, steer them, advise them, guide them – because as well as design requiring a team, these big solutions are never built for one person.

What I would say is ‘good on you, David’ for bringing concepts such as data visualisation to the masses, as too many people spend too long looking at tables of data when they can spend a moment to get the same understanding of where the trend is in a chart or graph. He’s not trying to say all visualisations should be like this, he’s trying to say look what you can do if you think about it a little. Context, that’s what he’s selling you and I keep referring to his book to remind myself of that.


By Kim Rees. October 30th, 2015 at 9:47 am

I have a full-time job and family, so I haven’t read every comment here, but one thing I feel I must address is the gross underestimation of people in this thread.

If someone came to me and wanted “more bubbles,” I would either hang up the phone or more likely, I would dig deeper to see what that actually translates to. Most buyers of visualization don’t claim to be experts on viz, don’t have time to even inform the design, and deep down, really crave for an expert to find out their business/org needs and create an informative, useful, and engaging visualization.

Much of this thread views practitioners as passive producers of whatever is requested. If someone cites McCandless and asks for colorful bubbles, that’s a great opportunity to find out what is compelling to that person. Is it color? Is it playfulness? Is it mood/emotion? Is it subject matter? Something else entirely?

I’m sensing this is really about unease of data visualizers who have no design background. Anyone can make a bar chart. It takes a good visualizer to dig into the nuance of data, find the compelling aspects, find the appropriate visual shape for it, make it engaging to the mind, and make it compelling to the eye. It’s time-consuming, thoughtful work.

When someone asks for “more bubbles” (or similar), and we throw up our hands and find someone else to blame for creating this problem — we’re really not embracing the work.

I know your next comment will be that some people aren’t receptive to this approach. If that’s actually true, and the practitioner feels this strongly about their craft, then they should find a new job/boss.

It’s our responsibility to uphold our ideals of the craft, no matter where on the spectrum it lies. There is no single truth in visualization or data for that matter.

By Stephen Few. October 30th, 2015 at 10:22 am


You are a skilled and thoughtful data visualization professional. When you work for a client, you take the time to understand their needs and to provide a solution that addresses their needs effectively. I applaud that. You are rare in the world of data visualization, however. Keep in mind, however, that your work and the work of others like you constitutes a tiny fraction of the visualizations that are created and used in the world. Most of the people who visualize data lack your training and the time that’s needed to build elaborate infographics or graphical applications. The vast majority of data visualizations are created by people like those who, if they’re lucky, attend workshops like mine and read books like mine that are designed to teach simple and practical skills. These are the folks who struggle to make sense of data and present what they’ve found to others in their organizations. These are the folks who are asked to produce bubbles because their boss watched McCandless’ TED Talk and decided that bubbles are cool. This is what concerns me. There is a reason that people ask for silly bubbles. This desire didn’t spring from their imagination unbidden. It was implanted in them by examples that dominate the web. These examples should be stamped with a warning: “This visualization may be harmful to your health.”

By Kim Rees. October 30th, 2015 at 10:24 am


Your dismissal of architecture is a bit stunning to me. You claim that Gehry’s buildings “have little effect on our lives and the world at large.” I was using a direct analogy – so for this discussion, the comparison should be Gehry’s influence on other architects. Meaning, can Gehry, being a superstar architect, have such an influence as to be detrimental to architecture as a whole. My argument would be no. Although I hate his work, there’s not shortage of “architects” and builders pumping out absolutely abysmal garbage into the world.

If you really are dismissing the effects of architecture on humans, I’m shocked. To compare a craft that influences our daily lives, shapes the way we live, enables or disables connections to nature and others, and can completely crush or elevate our emotions — to compare that to someone making business bar charts is ludicrous.

By Kim Rees. October 30th, 2015 at 10:32 am


I hate to dispel the myth that I and my firm are so special. I have a computer science bachelor’s degree, but everything I learned about visualization was with far less than even you provide. I ate your books up as they appeared in the book store. We were working off academic papers, the influence of a few pioneers, and pure gut feeling.

It’s every practitioner, no matter what their field, to use the best practices that fit the job. This is a nascent field rife with growing pains, but that doesn’t dismiss us from our responsibility. We can’t simply blame someone else’s practices for our output.

By Stephen Few. October 30th, 2015 at 10:39 am


Here’s the difference that you’re failing to acknowledge. Architects are professionals who go through many years of training and are certified before they can design buildings. Most people who visualize data have had little training, if any. As such, they are easily influenced by bad practices. If everyone who visualized data had to first take Alberto’s courses or mine and then spend years working under the supervision of a skilled professional before they were allowed to visualize data, McCandless’ influence could be compared to Gehry’s influence. Unfortunately, this isn’t the situation. I’ve spent the last 11 years writing books and teaching workshops in an attempt to make a dent on this problem. I’m telling you that my efforts are often undermined by the influence of people such as McCandless. Most of the people who work with data in organizations have never been trained to do this work. How do we address this problem? We don’t do it by exposing them to McCandless’ work.

By jlbriggs. October 30th, 2015 at 10:40 am

All I can say here, is once I read this line:

“Who says “data visualization” should inform effectively?”

From Andy Cotgreave, I had to throw in the towel.

If that’s really the kind of argument we’re up against, what’s the point?

By Stephen Few. October 30th, 2015 at 10:49 am


Your are very rare. In saying this, I’m not saying that you possess a latent talent that is rare, but that you have developed a set of skills that is rare. Like you, my formal education had nothing to do with data visualization. Over many years I worked hard to develop the skills that I have today. What we do, many others could do as well, but few have taken the time to learn. Unfortunately, the data sensemaking that organizations must do requires a large number of skilled practitioners. I’m trying to address this need with my books and courses. I’m working hard to teach people best practices. In the course of my work, however, I’m having to combat bad practices that are promoted by people like McCandless’. This is a real problem that most of the people who read blogs like mine and Alberto’s aren’t aware of. Can we blame McCandless for the negative influence that he has on people who are struggling to develop data visualization skills on the job? We can, and I do, because he knows that he is exerting this influence and he is doing nothing to address the problem.

By Stephen Few. October 30th, 2015 at 10:50 am

Two objections to my position seem to dominate recent comments: 1) I excessively emphasize the importance of day-to-day data visualization done by people in the workplace, and 2) I fail to appreciate McCandless contributions as an innovator. I’ll try to address them now.

It is absolutely true that I approach data visualization primarily from an awareness of and concern for the needs of typical practitioners. By typical, I mean those who represent the vast majority of information visualizers. These are people who rarely spend their time reading data visualization blogs—even this one. Their jobs don’t allow them the luxury of spare time to do anything but produce work. I’m talking about people whose day-to-day jobs require them to make sense of data that concerns their organizations and to then present what they’ve found to others who use this information to do their jobs, much of which involves decisions. After all, as Daniel Kahneman pointed out in Thinking, Fast and Slow, organizations of all types are, more than anything else, “decision factories.” Although we don’t see their work on the Web, these folks create well over 90% of the data visualizations that are produced today. These folks are my primary audience. These folks are not well served by the data visualati. In fact, most people who write about and/or teach data visualization are not familiar with the work of these folks and the challenges that they face. These are the folks that encounter in my teaching and consulting work, but rarely hear from in forums such as my blog.

The other audience that I try to keep in mind in my work, but can only reach indirectly, are the consumers of data visualization. These are the folks whose expectations are often set by examples that they see on the web. These are the folks who might watch the new commercial for Microsoft Office 2016, which features McCandless, and from that assume that they should be seeing more treemaps and sunburst diagrams in their daily reports.

Most of you who have made comments, including those of you who appreciate McCandless’ work, have admitted that his work should not be emulated by this vast audience that I serve. The problem, however, is that McCandless’ work reaches into their world and has an influence. This is due, in part, to the fact that McCandless accepts invitations to address these folks, even though they shouldn’t be emulating his work. When he does, he doesn’t point out that they shouldn’t follow his lead. He talks about the beauty of information and demonstrates how much fun it is to play with data without any particular end in mind beyond the pure joy of the experience. The people whose interests that I serve find this disheartening and frustrating. When their bosses ask for silly charts that look like those that McCandless’ displayed, they beat their heads against the wall. This could all be resolved if McCandless would clearly state what he does and then caution people who visualize data for other purposes from following his lead. Unfortunately, he doesn’t.

Do I fail to appreciate McCandless’ innovations? The answer to this is that I am not aware of any useful innovations that have been produced by McCandless. He does not engage in experimentation in the scientific sense. He does not put hypotheses to the test. He plays with data. Mostly, he draws from a limited bag of tricks, many of which have been shown to be ineffective.

Although it is not my primary activity, I engage in data visualization innovation. I’ve introduced several new ways of visualizing data to the world. As a good scientist, however, I don’t waste my time doing things that I know don’t work. That’s not innovation. I identify real problems—things that current data visualization techniques don’t handle or don’t handle well—and then I try to use my knowledge and imagination to find a way to solve that problem in a new way.

If any of you are aware of any useful data visualization innovations that McCandless has produced, please share them with us. If not, then please don’t encourage us to appreciate McCandless’ work because he contributes useful ideas that advance the field of data visualization.

By Harun Osmanovic. October 30th, 2015 at 11:18 am

Hi Stephen, Everyone,

I believe people need to take a few minutes to read your post about GE and McCandless. I thought your points were quite irrefutable there – http://www.perceptualedge.com/blog/?p=995

It seems to me that data is irrelevant in McCandless work. What I mean by that is you could replace the data with any other data set and reach the exact same outcome both in terms of informing readers about the topic and eye candy effect.
For instance when they ask the question “how does natural gas compare to other fossil fuels”. Needless to say McCandless’ visualisation can’t help us compare!

I’m more of the opinion McCandless deals in abstract art – data are his prime matter, computer softwares his tools. It is rather bothering to see him presented as a journalist (though he might do journalism otherwise, his work on data has very little journalistic value).

I’ve had similar debates about relativism in other fields and my feeling has always been that although “perpatrators of ‘wrong’ practices” should not be forbidden to speak ever, people who are exposed to them should have the opportunity to hear the other side.

Thanks for the fascinating conversations on here.


By Harun Osmanovic. October 30th, 2015 at 11:20 am

PS: Must mention that McCandless blog is “Information is beautiful”.
If the graphs do not inform, then they are failures, aren’t they?

By Stephen Few. October 30th, 2015 at 7:01 pm

Benjamin Disraeli once wrote, “There is no wisdom like frankness.” Let me try to be wise. When we visualize data, we strive for simplicity, clarity, and truthfulness. Any visualization, no matter what its intentions, fails if it complicates the data beyond its inherent complexity, obscures it, or misrepresents it. David McCandless errs in all three ways. No reasonable and acceptable objectives excuse him from these errors. If you disagree, explain the objectives that excuse McCandless’ errors and provide evidence, not speculation, that they are his.

By Jeffrey Shaffer. October 30th, 2015 at 8:04 pm

I would encourage the readers on this thread to read Dave ‘Fab Macca’ McCandless’s response to the Q&A he held on Reddit a few weeks ago. There are responses that hit home on a few of Steve’s points and clarify David’s position.

David does not see himself as a data artist. Here’s one response.

“I still see myself as a journalist. I see a lot of similarities between information design and journalism. Both set about to condense and optimise information into tight, understandable, elegant form, telling stories and helping others to navigate.”

He describes his process:

“I usually consider elements from both the design and content sides. Things like.
Strength of idea / proposition / question, Information & data quality, usefulness, revelation, effectiveness of execution, ease of use, aesthetic beauty, originality, style, creative touch, originality.”

Both of these statements clearly indicate his intent and purpose is to inform people and to share his visualizations with others. He seems to churn out visualizations at incredible speeds.

“I’ve created a lot of data and information visualizations. Around 540 over six years (An eye-bleeding, marriage-crumpling average of 1.7 per week).”

Also, I have had similar experience as Kim has described. I don’t find executives or clients forcing bad design in every case. Certainly I’ve encountered situations where someone’s client or the VP of All Things Circular requests a pie chart or bubble chart, but in many cases, the client or executive is trusting the expert. One of Steve’s points is that they don’t often have an expert on hand to guide them. There is frequently someone who is not trained in best practices, hasn’t read the books or taken a data visualization class and they are finding work online by people like David McCandless and trying to apply it to their real-world scenarios. This can lead to the situations that Steve is describing and that he has encountered in real-world situations.

As I read the comments from Alberto, it sounds like both the students and the non-students were versed in the fundamentals of data visualization, but I’m sure that’s not the case in other places where he speaks. The unfortunate reality is that the online advertising announcing that he was a speaker at the University of Miami could lead others to draw conclusions and make assumptions that Alberto is endorsing his work or methodology.

I am pleased about one thing, that my students at the University of Cincinnati will be able to read this very interesting discussion.


By Jeffrey Shaffer. October 30th, 2015 at 8:05 pm

Link to the Reddit Q&A:


By Kim Rees. October 31st, 2015 at 5:53 am


I do understand distinctly the problem you outline. We don’t consult nearly to the level you do, but I regularly encounter and work with people who have to do the type of work you’re talking about. In fact, I’ve sung the praises of these practitioners in many of my talks because they are producing work of impact that will never reach more than an insular audience.

However, I still think singling out certain people is misplaced. The scenarios you describe exist in every field. For instance, I used to work in web development. Whenever there was some new crazy site all of our clients would want to copy it. It was annoying, sure. But it was our responsibility to translate what they were responding to and see how we could apply it to their sites, if at all.

It was the same scenario when I went to my stylist once and told her I wanted to look like Angelina Jolie. Unfortunately, she couldn’t stifle her laughter. Does my stylist curse Angelina for making her clients unrealistic? No. She welcomes the opportunity to make them look as good as they can. In fact, most stylists love celebrities who have unobtainable and sometimes unhealthy beauty. It brings them clients.

It’s our job to translate those desires into something suitable.

By Stephen Few. October 31st, 2015 at 8:56 am


McCandless is not the cause of all data visualization’s ills. Rather, he is a prime representative of a group of infographic designers who promote ineffective practices. I’ve critiqued the work of several, but none more than McCandless because he is the best known of them and he actively and consciously promotes his work to the very people who stand to be most harmed by it (i.e., those who visualize data to support decision making). In addition to the bad infographic designer camp, I warn people about two other camps that plague our work: software vendors that build ineffective tools and people in the information visualization research community who do poor work and promote poor practices.

The Angelina Jolie analogy doesn’t quite work. She is not actively promoting harmful practices. McCandless is. It is absolutely true that when people are influenced by bad data visualization practices, such as those that McCandless promotes, we must help them understand what will actually work for them. When the sources of influences are known, however, we have another way to combat them and thus prevent the harm from happening in the first place. We can expose those sources and do what we can to either shut them down or help them move in a productive direction. I’m taking advantage of that opportunity. To bring this back to the original point of this blog article, we can oppose those who promote bad practices by either denying them a platform or making sure that if they’re given a platform a counterpoint is also provided. Alberto missed this opportunity with McCandless, which has led to this discussion.

By Harun Osmanovic. October 31st, 2015 at 11:52 am

If, as Jeffrey pointed, David sees himself as
“a journalist. I see a lot of similarities between information design and journalism. Both set about to condense and optimise information into tight, understandable, elegant form, telling stories and helping others to navigate.”

Then, from my perspective, having lookes only a dozen of his visualisation, he fails to make them understandable. I certainly do not navigate easily through them.
If his visualisation are not understandable, like some graphic codes, then they are not really information, and it’s even more important to present counterpoints to his work.

By Stephen Few. October 31st, 2015 at 4:33 pm

I can’t help but think that one day McCandless will make a major announcement that begins with the word “Gotcha!” He’ll confess with a smile that he’s been pulling our legs all along to make the point that the world is gullible, that we’ll accept anything that’s wrapped in a pretty package. He’ll ask, “How could you have possibly believed that a journalist would exhibit this much disdain for information and communication?” Take a look at this example of my work:

No one could have made this many errors unintentionally. And would a journalist name a search engine as his source?

If my suspicion (and hope) is wrong, however, and he’s actually serious, then I’ll prepare for that debate that Alberto has offered to moderate. In fact, I’ll even help McCandless prepare for it by previewing a few of my questions now.

  1. When you represent values inaccurately in your charts, is this because:
    1. You made a mistake
    2. You don’t care, because you don’t think it matters
  2. If your answer to question #1 was “You made a mistake,” do you correct the mistake once it’s been pointed out? If so, can you provide an example?
  3. If your answer to question #1 was “You don’t care, because it doesn’t matter,” why do you feel it doesn’t matter?
  4. When you encode values in ways that are difficult for the human brain to interpret and compare, do you do this because
    1. You don’t understand human perception
    2. You are intentionally making it difficult for people to process the data
    3. You don’t care, because you don’t think it matters
  5. If your answer to question #2 was “You don’t understand human perception,” why haven’t you taken the little bit of time that would be needed to understand the basic of human perception as they apply to data visualization?
  6. If your answer to question #2 was “You are intentionally making it difficult for people to process the data,” why are you accomplishing by doing this?
  7. If your answer to question #2 was “You don’t care, because you don’t think it matters,” why do you feel it doesn’t matter?
  8. What are the primary objectives of your data visualizations? What are they designed to help people do or experience?
  9. Are you a journalist? If so, how do you define journalism as it applies to your work?
  10. Do you believe that people should emulate your data visualization practices if their goal is to inform and to inform well?
  11. If your answer to question #10 is “Yes,” provide examples of the specific data visualization practices that you think people should emulate.
  12. If your answer to question #10 is “No,” why then do you present your work to people whose goal is to inform without making it clear that they should not follow your practices if that is their goal?
  13. Do you consider yourself a data visualization innovator? If your answer is “Yes,” please provide an example or two of the innovations that you’ve produced.

Perhaps McCandless could do us all a favor and respond to these questions now.

By Marco Torchiano. November 2nd, 2015 at 1:58 am

Is data visualization a science? Then, are bad viz practices w.r.t. visualization similar to what homeopathy is to medicine?
If the answer is positive to both questions, then an academic should not give stage to an “homeopathic visualizer”!

The thing is I’m not sure whether I’d answer positively to the questions.

By Stephen Few. November 2nd, 2015 at 1:08 pm


Data visualization is not a science in its own right. Rather, it is a discipline that is informed by several branches of science. Its effectiveness can be assessed scientifically. Consequently, in my mind when Alberto hosted a lecture by McCandless, it was very much like a medical doctor hosting a lecture by a practitioner of homeopathy. My disappointment in Alberto’s failure to expose McCandless’ bad practices stems in part from the fact that, as both an expert in data visualization and a journalist, no one is better qualified to do this than him.

By Tom Shanley. November 2nd, 2015 at 8:39 pm

First of all, thank you Stephen for your great blog and passion for data visualisation – it is great to find this kind of debate, ignited by your concerns for how poor data visualisation can have negative effects.

The internal debate I have had about McCandless’ work is this: it was his books that first inspired me to look beyond Excel defaults. I now recognise, from books and blogs like yours, Alberto’s and others, that much of McCandless’ work is ineffective; but nevertheless, it was his work that kicked started me into learning and applying data viz to my work. and I’m sure that this applies to lots others who have read his books or seen his talks.

Hopefully some of them follow our path to learn more, or work with people who do, and go on to be better data vizers. Now I can’t reconcile if this is good? Does the good that comes from more people being inspired to use data viz by McCandless, offset the negative impacts of those who blindly follow McCandless’ designs?

and would Cairo challenging McCandless at his lecture deter McCandless from speaking (and inspiring) and the audience from using data viz? If McCandless was to stop, who could step into his place to be a more worthy and public advocate (Rosling perhaps)?

By Stephen Few. November 2nd, 2015 at 9:29 pm


I don’t want McCandless to stop visualizing data; I want him to stop doing it poorly. If his work demonstrated effective practices and the genuine respect for information that we expect from journalists, it would have still inspired your interest in data visualization and would have prevented you from wasting time pursuing ineffective practices as well.

By Tom Shanley. November 3rd, 2015 at 1:08 am

Cheers Stephen – I hope you get your answers from David M. I’ve forwarded this debate to many of my peers , primarily due to the time and thoughtfulness you’ve taken to respond to everyone’s questions \ opinions

By Sawsan Khuri. November 3rd, 2015 at 2:44 pm

Hello Stephen,
Thank you for thoughtful and thought provoking posts above. My take is simple, there’s room for everyone because each of you fulfills a different purpose. You are an exacting journalist working in business intelligence with a passion for clarity, accuracy and detail. David, on the other hand, is playing with colours and shapes, offering us a lighthearted sense of the data he is handling. I don’t know for sure, but I would suspect that the businesses that recruit you are probably not the same ones that would be interested to recruit David for their visualizations. For me, coming across David’s book at a bookshop in 2009 inspired my fascination with data visualization per se, which as you know continues to this day.

For the record, I’m not a data visualization expert, designer nor journalist. I am a scientist lost to academic administration, with a vested interest in the public understanding of science. In that domain, I see scope and a role for all sorts of different approaches to visualizing data. The approaches should be as diverse as your audiences. The strength comes in applying lessons learnt to one’s own data: maybe take an idea from one, add it to a certain perspective of another, change something here, customize it there, and we all stand on the backs of many giants.

If I may also speak to whether Alberto was “right” to have David present, yes of course he was. We need to be be exposed to all the data visualization ideas and methods that are out there, we can be trusted to take what we think makes sense and leave the rest. I can’t tell you the number of times I’ve sat in scientific seminars thinking not very positive thoughts about the science I happen to be listening to!

Thanks again, enjoyed this thread, and here concludes my 2-cents worth.
All the best,

By Stephen Few. November 3rd, 2015 at 10:15 pm


I’m not a journalist. I’m a teacher. McCandless claims to be a journalist. As such, it his job to inform people about the news. The problem is that he doesn’t do this well. His displays are often difficult to read and inaccurate. Assuming that “offering us a lighthearted sense of data” is a viable journalistic objective, McCandless would still need to present data in an accessible and truthful manner to achieve this objective. The problem is that he doesn’t.

As someone with a vested interest in the public understanding of science, you should be deeply committed to data displays that are accessible and truthful. McCandless’ work conflicts with this interest. When you attend seminars and listen to other scientists present their work, you evaluate its merits. You reject poorly designed and otherwise invalid research and appreciate science that is done well. I evaluate the work of data visualizers. I reject poorly designed data visualizations and appreciate data visualizations that are done well. Would you invite a pseudo-scientist (e.g., an advocate of intelligent design) to present his work in a public venue without providing a counterpoint? I doubt that you would.

By Andrew Craft. November 4th, 2015 at 4:22 pm

The best response to bad practices is to try to understand them and make them better, and if that can’t be done then abandon them or repurpose them. Which I guess covers both points of view, so now we can all be friends again..?

Of course the real question is whether McCandless’ stuff is bad practice. I guess maybe some people just don’t think it is. Which is unfortunate, because his work is actually quite awful for people who care about the data and need to see it in a meaningful way. Pretty, yes. But meaningful? No.

And that objection has nothing to do with McCandless not conforming, as some suspect. His graphics represent the data so poorly, mistakes hide in them. That’s right – some of his pieces contain mistakes (usually disproportionate lengths/areas/etc), and you can’t see them unless you go looking for them. That’s fine if you’re just looking at the graphic for your own amusement – you might be able to get the general idea despite some things being off a little. But it’s pretty much all the motivation a person needs to not ever use his innovations for serious analysis or decision-making.

By Terry Hayden. November 5th, 2015 at 8:47 am

Well said, Andrew! The folks who don’t think it is important that the data representation be accurate, first and foremost, must not use data to make important decisions. The accuracy of the data should be the number one concern of any report/visualization.

By jlbriggs. November 5th, 2015 at 11:08 am


“Pretty, yes. But meaningful? No.”

I guess the biggest problem I have with this whole debate is McCandless being framed as the guy who makes “pretty” graphics.

I don’t find that the vast majority of his work is aesthetically pleasing at all, which really makes this “pretty vs. useful” argument a dead issue regardless.

By Andrew Craft. November 5th, 2015 at 12:07 pm


You mean beauty is subjective? Who knew!

The way I figure, it’s just not worth arguing whether his work is aesthetically pleasing, lest arguments of personal preference plague the more important points of this debate.

I’m trying to argue that McCandless’ graphics are not conducive toward good decision-making. The last thing I want is people telling me that the only reason I say that is because I have a biased *opinion* that his work is “ugly”.

By Andy Cotgreave. November 5th, 2015 at 3:28 pm

You said: “The folks who don’t think it is important that the data representation be accurate, first and foremost, must not use data to make important decisions”

Why should all data be used to always make important decisions? Why not use data to make something playful? Or use data to make a sculpture? Or just use it to get a sense of the world? All of those are perfectly valid ways to use data.

IF you use data to make important decisions, THEN you should look at it in the most effective and efficient way. But if you use data for other reasons then go nuts and be creative (but don’t deceive!)

By Andrew Craft. November 6th, 2015 at 10:19 am

@Andy Cotgreave: “Why should all data be used to always make important decisions?”

Who made that assertion? I feel like you’re responding to a different discussion entirely.

“IF you use data to make important decisions, THEN you should look at it in the most effective and efficient way.”

Agreed – that’s the contrapositive (P->Q, thus !Q->!P) of what Terry suggested – if you don’t care about presenting the data in a way that informs effectively, then you must not be using the data to make important decisions.

By Robert Monfera. November 11th, 2015 at 1:39 am


I liked the Odds example above as illustrative of your point. Hate to cite out of context but an answer from Mike Bostock from a couple of months ago seems pertinent and self-contained enough:

“If there’s one thing I learned from NYT it’s the importance of labels, legends and annotations. A graphic has to explain itself. Minimalism can be good if it represents focus—cutting superfluous elements to emphasize the key points. But removing elements that are essential to understanding makes the graphic useless. So for each element, ask what it explains or what purpose it serves. Include precisely what is needed to convey the message.”


Mind you, I’m not claiming that work from McCandless is minimalistic – often it’s boldly decorated. Low data to ink isn’t even the issue here. Lack of proper citations of data sources, improper, misleading and unexplained scales, and often, sparsity of context is surprising about his work, which I shelved internally as ‘inspiration, drafts and exploration in attention-catching visualization’ rather than as journalism, or good journalistic practices. I have limited experience in journalism but if anybody, groups at the NYT and their ilk are definitely in the trenches to know better.

By kris erickson. November 13th, 2015 at 1:38 pm


First of all again thank you for your incredible dedication to your work. Secondly, thank you for your willingness to offer thoughtful criticism to the data community.

In my position I’ve had the opportunity to see a weekly multi-paged CEO dashboard for my employer (a major US retailer) and contribute a (very) small part to it. I can assure you that McCandless’ influence on the day-to-day operational reporting is zero. Everything consists of bars, lines, Gantt charts, or tables (with various color). Zero packed bubbles. Perhaps McCandless influences some of the publicly available annual reports, but not the operational charts. Something that’s “cute” on day one is aggravating to the user by day three (I believe you have said something similar in the chart-junk debate). In addition, many people are risk-averse and will not deviate from the perceived defaults.

There are however several examples of tool-defaults being propagated to the final CEO dashboard. Luckily this is limited to a few shadows and default color choices. My biggest fear is when the tool-makers get starry-eyed about McCandless’ work and go on to incorporate it within their tool because “it’s cool”, or “someone will use it (maybe for non-important data)”. The tool-makers have a unique position of passing along functional wisdom automatically in the design of their tools.

By Ladi Omole. November 16th, 2015 at 4:07 am


I work on different information technology projects. I found your site and books very resourceful when working on data visualization related projects and hoping to reference your work on a data visualization initiative I am working on. Thank you for making it public.

Will it be a fair statement to refer to you as “Data visualization expert” and David as “Information graphic artist expert”?

By Stephen Few. November 16th, 2015 at 9:57 am


I’m not sure what McCandless should be called. He is certainly not an expert in data visualization. According to a recent interview, he does not consider his work data art. The only professional title that I’ve seen him use is “journalist,” but what he does isn’t journalism–certainly not good journalism–by any common definition of the term.

By Chris Piechowicz. November 29th, 2015 at 9:27 pm

I wonder if these comments/responses would be any different if McCandles was teaching their own kids in math class that 2 + 2 = 22. Would they still be comfortable with their impressionable children learning falsehoods before they had a chance to learn the real answers?

Leave a Reply