Rosling—Where’s the data?

Hans Rosling of GapMinder is one of my heroes. He has become an engaging and powerful teller of quantitative stories. He’s making a difference in the world. Even the most talented among us, however, sometimes slip up. Rosling’s recent video, produced by BBC Four, takes advantage of technology to place him behind a transparent bubble chart, making it possible for him to direct our attention to particular items with greater ease and clarity, without blocking our view-a worthy goal for statistical narrative. This approach suffers, however, from one significant flaw: in addition to Rosling, an entire room with bright lights, beams, and windows appears in the background as well, resulting in a great deal of distraction.

The production crew could have easily used a clean backdrop for the video, which would have removed all distractions and made it easy to focus on the data and Rosling’s narration. This problem perhaps never occurred to the technicians (although it should have), and I suspect that Rosling had no idea that all those windows and lamps with glaring lights would show up in the finished video. Attention to these details, however, makes the difference between fun and engaging data visualizations that tell stories effectively and those that feature novelty and entertainment over substance. To focus attention on the story, all distractions must be removed. As we venture into new opportunities that technology makes possible, we dare not forget the important lessons of the past. In the years since Edward Tufte began promoting the reduction of non-data ink in visual displays, research has confirmed the importance of this practice due to limitations in human perception, cognition, and memory. We can only focus on a small portion of the visual field at one time, we can only consciously attend to one task at a time, and we can only hold about three chunks of visual information at a time. There is no room for distraction of any kind. Anything that isn’t data must have a good reason to be there or it should be eliminated. As Antoine de Saint-Exupery once wrote, “Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” This is especially true when telling quantitative stories.

Take care,

23 Comments on “Rosling—Where’s the data?”


By Marty Gierke. December 16th, 2010 at 9:28 am

Agree Steve, it’s a bit over the top. I did not have an issue with the way the story was introduced, and the setting is kind of cool, but when he got into the data it would have helped to tune out the background – especially the strong back-lights & windows as you mentioned. I found myself thinking about Powers of Ten, and the way Eames used more discreet markers to orient the viewer to specific points in the animation. It’s almost as if Rosling needed to exaggerate the year markers to compensate for the background, and that in turn became another distraction.

By Andrew Battye. December 24th, 2010 at 8:29 am

I stumbled across the TV show and was fascinated by the use of stats and visualisation. I must admit, I did not get distracted by the windows and the harsh lighting during Rosling’s presentation, but, I guess I’m used to it by now, as every BBC show seems to be filmed in a stupid warehouse or somewhere equally inappropriate.

I did wonder, however, where he got all the stats from – life-expectancy and average incomes for practically every country on earth, going back almost 200 years – from some countries that were notoriously secretive and from some where no reliable collection of such information would even have been attempted.

By Chris W. January 4th, 2011 at 5:49 pm

I have to disagree this time (though it’s one of the few times). His goal in this case was to appeal to a larger demographic. Your assertion that Hans point would have been better suited had he toned down the background noise only really appeals to those that truly care for the details of the information. In this case his audience was via television, so most likely they didn’t care for the in depth analysis, or crystal clear graphics. They wanted a split television show / data analysis. Which I felt was hit perfectly.

Hans’ great effects grabbed my attention and I listened more closely to the story told than I would have if it were a flat graph on the screen. In this way – he was a success. However, when I need to know mortality rates vs income in various countries I will not be referencing this video. That was not what this was for. I believe you may have misinterpreted Hans’ purpose with the live graph.

By Stephen Few. January 5th, 2011 at 2:44 pm

Chris,

If this presentation was meant to engage a TV audience in the visual effects without really informing them, it would have perhaps succeeded, but I don’t believe this was Rosling’s intention. He cares deeply about the data. He wants to deliver a clear message to his audience. He could have appealed to a TV audience just as effectively without the distracting background. The background wasn’t engaging; it was merely distracting. Had a solid background been used, the presentation would have been even more engaging, because what was happening on the screen could have been easily seen. I suspect that Rosling didn’t know that the finished product would look as it does. Had he known, he probably would have suggested a background that was designed to provide maximum visibility for the data.

By Phillip R.. January 6th, 2011 at 11:53 am

Of course windows and lights are not the ideal background, but I was so enthralled (only a slight exaggeration) with his presentation I didn’t even notice. Should he have used a proper solid (boring?) background? Maybe so, but somehow I think I’d have found it a bit less compelling.

By Stephen Few. January 6th, 2011 at 12:55 pm

Phillip,

The action and objects of interest are in the foreground, not the background. The purpose of the background is not to draw attention to itself and therefore away from the story, but to serve as a neutral canvas on which the story can be easily seen in the foreground.

I’m confident that you would have actually found this more compelling without the distracting background, which did nothing to support the story. This opinion is based on what I know about our brains and how they work. Great storytellers of all types (orators, comic book artists, infographic artists, data visualizers, etc.) engage their audiences–pull them into the story–by focusing their attention as fully as possible on the story. A critical method for doing this involves the removal of everything that distracts attention from the story.

Rather than defending the use of meaningless effects, if you care about the story that Rosling told in his presentation and other important stories about the world, it would be more helpful to promote best practices.

By Phillip R. January 9th, 2011 at 9:54 am

I wanted to express my reaction to his presentation after seeing it for the first time. It was simply that I did not recall being distracted by the background. I was so taken with his words and visuals all else seemed to “disappear.” Perhaps this observation was not worth making. My intention was not to defend the use of meaningless effects.

By Stephen Few. January 9th, 2011 at 11:12 am

Phillip,

You don’t have to be consciously aware of the distraction for it to affect you. You found Rosling’s story fascinating, so you were trying to focus on that, but the distractions in the background were undermining your effort, whether you were aware of it or not. Most of the silly stuff that finds its way into data visualization are not recognized by viewers as distractions, but they obscure the data nonetheless. By raising people’s awareness of these problems and how much more they could be getting from data visualization, I and others of my ilk hope to move us to a tipping point: a point when most people understand the potential of data visualization and the best practices that support it well enough to demand better. Once we reach this tipping point, the glaring lights in the background of Rosling’s presentation will be turned off.

By Janett. January 25th, 2011 at 11:47 am

I did not see this tv bit – mostly because I watch tv rarely – if at all. However, the first thought that came into my mind when Stephen mentions a clean backdrop is the movie The Matrix. Ok, not exactly a movie of immense poetry – but think about the scene when Lawrence Fishburn is showing Keanu Reeves the complete arsenal available in the virtual world: they are standing in a stark white room. Would the rush of weapons to their side have been as dramatic had they already zipped into a location? I don’t think so. The focus is on quantity of the weapons – or whatever popped up in the “white room” – not the room itself. I know, kind of a cheesy comparison – but just giving my thoughts. I side with Stephen. The focus is the data, therefore for the same reason you don’t put a colored chart on lavender paper…you wouldn’t put a man behind a see-through “chart” in the center of the warehouse.

By Bob. February 1st, 2011 at 8:27 am

“You found Rosling’s story fascinating, so you were trying to focus on that, but the distractions in the background were undermining your effort, whether you were aware of it or not”

– justify this claim with data or retract it. Don’t think that the burden of proof is on you? Look up stochastic resonance for how noise can improve signal perception. Look up the literature on how general arousal (of the sort arrising from a visually stimulating scene) can improve attention. Your empirical claims are at best unsubstantiated and at worst entirely wrong.

By Stephen Few. February 3rd, 2011 at 5:27 pm

Bob,

Rather than citing decades of research regarding distractions to visual perception and attention, let’s examine my claim directly, with our own eyes. Compare the chart that appears in Rosling’s BBC4 presentation below to the second version of the same chart with a simple, clean background. In which can you more easily see the information?

Rosling's Bubble Chart along with Recreation

Rosling’s BBC4 video presentation is not a case when noise (the features of the room) enhances our perception of the signal (the information in the chart). The features of the room do not stimulate us in a way that enhances our perception or understanding. Even if you find the room with its shadows and glaring lights engaging (seriously?), do you really believe that it engages you in a way that draws you into the story? The effectiveness of a presentation is not measured by the degree to which the audience’s eyes dance around the scene, but by the degree to which people understand the message and are moved by it.

The stories that Rosling tells are important. They deserve to be heard, understood, and acted upon. Had he stood in front and to the side of a chart similar to the second version above and pointed to items as he referred to them while narrating the story, his passion and engaging personality would have directed our attention more powerfully to the content that matters.

Steve

By DR. April 21st, 2011 at 8:40 am

(I recognize this is an older post/discussion, but I feel compelled nonetheless..)

I find this to be a very strange argument…

Let me ask this: If Rosling did “interact” with the graphic as your suggesting he should have (no background), do you think this video would be as popular?

I would argue that nobody would watch this thing if you removed the background. Let’s continue your line of reasoning: Rosling himself is a distraction – no? Why not simply do a voice-over and highlight/indicate visually the points he is making? That would really focus on the data. Yet, by doing so, you’ve removed anything compelling about this presentation.

For better or for worse, by putting the whole thing in an environment that is comfortable and “alive” (not sterile and institutional – which is the effect you’re after) the entire presentation becomes more approachable. Yes, if I want to examine the data in detail, there are better approaches, but if I want to draw in an audience (Rosling’s stated goal in much of his work), this approach is spot-on.

By Stephen Few. April 21st, 2011 at 8:49 am

DR,

Except for this one example, Rosling has always interacted with a display on a normal screen, which lacks the distracting background, and people find his presentations extremely compelling. When he spoke at the TED conference for the first time in 2006, the audience was astounded, moved, and enlightened. Go to http://www.GapMinder.org and see for yourself.

By DR. April 21st, 2011 at 3:05 pm

I’m quite familiar with his work, quite familiar with gapMinder, have watched many of his TED talks, and am generally a fan. I therefore recognize that he was doing something different in the BBC program.

The BBC program was about making data sexy. I’m agreeing with you in terms of data-interpretation, but I’m simply arguing that data-interpretation was not the primary goal of this presentation. And when released from that restraint he did something new, interesting, and compelling. Just refreshing I thought.

By Stephen Few. April 21st, 2011 at 3:19 pm

DR,

So you didn’t really mean it when you said: “I would argue that nobody would watch this thing if you removed the background.” If the goal of the BBC4 presentation was not to reveal the story in the data, what was the goal?

By DR. April 21st, 2011 at 3:43 pm

Fair point. Let me clarify:

I hold that if you removed the background, the video would not have 4 million (4 million!) views on youTube.

It would instead hold the attention of academics and data-wonks (I count myself among one or both of these groups..) which explains its success at TED and other similar venues.

(Do note that Rosling has other talks on youTube, with the same graphic, on the same topic, that do not reach the same order of magnitude of viewers. Why do you suppose that is?)

As I stated before, the goal of the BBC presentation was definitely not to reveal the story in the data. The goal was to make statistics sexy. It’s in the title; the name of the program was Joy of Stats. And Rosling is quoted as saying: “I kid you not, statistics is now the sexiest subject on the planet”. The tag line for the show goes something like “…takes viewers on a rollercoaster ride through the wonderful world of statistics…”. The program was clearly not aired so that the people of the U.K. would understand the historical relationship of income and life expectancy. Rather the program was aired to convince people that statistics is cool.

And with that stated goal, the data takes a back seat, and the background becomes a sign of comfort and a point of grounding for the layperson. In that respect, I think it’s a smart and effective choice.

By Stephen Few. April 21st, 2011 at 4:19 pm

DR,

“Smart and effective choice” for what purpose? Certainly not for any purpose related to the data, which is the whole point of statistics. The presentation was sexy, not the statistics. The popularity of this particular version of Rosling’s presentation stems from its novelty and sense of science fiction (“Minority Report”), combined with the reach of BBC4, not in my opinion from anything intrinsically comfortable about the setting (cold, empty, large brick room, made even less comfortable through the use of unfamiliar visual effects).

My point when I wrote my initial comments above was that this is not a useful direction for data visualization, but a potentially harmful detour. Data visualization doesn’t need another detour into ineffectiveness.

By DR. April 22nd, 2011 at 9:50 am

I believe I’ve answered your direct questions above – so I won’t repeat my argument. The whole point was to make the presentation of the data sexy – not the data itself.

But just food for thought since I missed this originally:
http://www.nytimes.com/2011/04/03/business/03stream.html?_r=1

I guess I think that the popularization of this discipline is important. And perhaps I’m more forgiving of perceived missteps along the way than you are. You often use the analogy of the early-days of the web. You tend to say “we were lousy at first, but we got better”. I look at it differently – the “lousy” early days had the incredibly important side-effect of very rapid user adoption. I’m more than happy to see massive adoption of the discipline, then work out the kinks later. To argue that we don’t need to make the mistakes along the way because people like Tufte have written the “manual”, is to miss the scope of what is going on in the field, and what is possible.

By Stephen Few. April 22nd, 2011 at 10:34 am

DR,

The difference between data visualization and the Web is that when we began using the Web, it was brand new — ways of doing it well were not yet known. We’ve been doing data visualization for a long time, however, and have already discovered many of the principles for doing it well. There’s no reason to go through a long period of rediscovering what we already know. If everyone who claims expertise in data visualization built on the foundation of what’s already known, we would progress much faster.

By DR. April 22nd, 2011 at 10:55 am

A fair point – but it doesn’t carry into the present, much less the future…

For example, I couldn’t find the chapter in “The Visual Display of Quantitative Information” that discusses how best to zoom-in on a sparkline with a multi-touch enabled device who’s display changes relative to it’s position in space.
(sorry for the sarcasm but I’m working on this problem now)

My point is, the world is changing very quickly, and saying “a bar chart is better than a pie” is just not sufficient. It’s a reasonable start, but I’d align it with “animated under construction .gifs aren’t cool anymore”. We’re in a space that is absolutely unexplored, and it’s not clear to me that the old “fundamentals” necessarily apply.

To that end Rosling did something nobody has done before. I think he was right to question whether existing rules for paper and screen apply. I happen to think his experiment was successful and you disagree. But the only way to know what works is to try it.

By Stephen Few. April 22nd, 2011 at 11:22 am

DR,

You love hyperbole, don’t you? We are not “in a space that is absolutely unexplored.” There is much about data visualization that is unexplored, and that’s where we should focus our efforts, not on what is already well established. In the beginning, Rosling created a useful visualization technique by extending an established means — a bubble plot — with a new way to show how values change through time using animation. He didn’t recreate the bubble plot, nor did ignore what’s already known about the design of bubble plots, he extended it into a new realm. Following his innovation, others such as George Robertson of Microsoft Research examined the effectiveness of animated bubble plots for various purposes and found that they don’t work for data sensemaking because we can’t follow all those moving objects, but that they work primarily for visual storytelling, because the narrator tells the audience where to look. Rosling’s original invention and Robertson’s subsequent research are both examples of worthwhile effort. What the BBC4 did with Rosling’s presentation, however, undermined its purpose — to tell a story using data — because they ignored what we already know about visual perception and the limits of attention. The video crew wasn’t engaged in thoughtful exploration, they were playing with technology to do something that they thought was cool. All of the old fundamentals about graphical perception still apply whenever graphics are used. Nothing about the BBC4’s video changed any of this. There is no doubt a whole world of new opportunities for visualizing data using cutting-edge video technology, but attempts to innovate in this realm that ignore what we already know are wasteful. They’re worse than wasteful, in fact, when they’re held up to people as examples of how data visualization should be done.

Here’s the bottom line: there is never a legitimate reason to obscure the data. Innovation and applications of new technology to data visualization should always seek to present information clearly, accurately, and meaningfully in an attempt to support understanding.

By DR. April 22nd, 2011 at 12:46 pm

I’ve enjoyed this conversation but I think it’s time to bring it to an end because I’m afraid you won’t take these notions seriously.

I’m am absolutely not being hyperbolic. Have you designed for an iPad? An iPhone? For 3D TV? A 3.5″ screen? A 164″ screen? Those of us that are doing so are absolutely finding ourselves in unexplored territory. The interaction, the perception, the way the mind digests information is different. Meaning: not the same. A well designed excel dashboard is instructional. But it’s value disappears at an alarming rate when you move into newer media. Particularly when you’re competing with the remarkable amount of noise being generated in these platforms.

I take your points seriously as I respect your work – so I appreciate the conversation, but I fear that if we ignore the differences, and discourage experimentation from known norms, we’re doing ourselves a serious injustice, and will never fully realize the potential of new delivery mechanisms.

By Stephen Few. April 22nd, 2011 at 2:29 pm

DR,

I have not encouraged you to “ignore the differences.” Rather, I’ve encouraged you to bear in mind those things that are not different. Although the technologies that you’re working with (yes, I’ve worked with them as well) are new in some respects (not all), human perception and cognition is the same as it’s been for ages. You dare not lose sight of this. No matter how different the technology, if data is being presented to people’s eyes, you must take into account the abilities and limitations of visual perception and cognition. Discover the opportunities of these new technologies to interact with visual perception and cognition in ways that take better advantage of their strengths and work around their limitations. You won’t do this by ignoring the lessons that we’ve already learned, but by using this knowledge as the platform on you can build new ways of making data understandable that surpass what we’ve done in the past. Don’t waste your time using new technologies to make old mistakes. You’re smarter than that.