One Death is a Tragedy; a Million is a Statistic

Photograph by *Your Guide

I posted a comic from xkcd last week that implied graphs and data lead to a decline in love. I didn’t really think much of it, but Jim commented that an episode from This American Life (episode 88: Numbers), was very much related to the topic of personal data and what we often miss out on as a result. The lead-in to the show reads:

Numbers lie. Numbers cover over complicated feelings and ambiguous situations. In this week’s show, stories of people trying to use numbers to describe things that should not be quantified.

This reminded me of Joseph Stalin’s well known quote, “One death is a tragedy; a million is a statistic.” It’s a horrible thing to say, but when it comes to data visualization and analysis, it’s true a lot of the time. We have a huge dataset and we have to extract information from it. In the process though, we forget that every one of those numbers has real non-numeric value to it. There are emotions and feelings. Life is complex. Data represents life, and therein lies the purpose and meaning of FlowingData.

People often mistake my use of “flowing data” as purely meaning data that frequently updates – that it moves in some way like a river, and that’s certainly part of it, but there’s more to it than that. There’s life in that river. There are undercurrents, erosion, growth, curves and bends, temperature changes, rapids, waterfalls, and water that is practically at a standstill. Where does the river begin, and where does it pour out into? Where did that river come from? How old is it? How young is it?

Data visualization should portray this “life” in data that it represents. This calls for visualization that goes beyond standard graphs, that seem to capture the factual part of the data and disregard all else a lot of the time. The numeric value of a data point is only part of the story. Don’t forget the who, what, when, where and why.

This is not to say that there is no value in personal data collection as told in Act 3: When Days Are Numbered and sort of the running motif through the entire episode. There is plenty of meaning and purpose behind the numbers themselves; however, when we collect data – this self-surveillance – we’re not really after the numbers themselves. We are interested in what is behind the numbers. The life. The emotion. How we feel. People track their weight, because they feel fat when they are overweight. When they feel fat, they feel insecure. Without the emotion, the number doesn’t mean much, does it? It’s best to keep that in mind when we work with data.

Stalin was wrong. One death is a tragedy, yes, but one million deaths? That’s one million tragedies.


  • Brilliantly said. I gave up sociology because I felt I couldn’t see the people through the sociological windown and now feel pretty much the same about psychology.

    Now how can I save this article so I can find it again?

  • Of course, those same folks throwing up their hands about the sterilization of tragedies through statistics are the same people whose rage and sadness is linked the visceral representation of tragedies, rather than a more holistic, objective analysis of the many wrongs that take place.

    It’s no wonder, human’s ability to conceptualize numbers past the teens is woefully inadequate. We really can’t “feel the pain” beyond that many people — 1,000 die or 10,000 die, we can’t feel or conceptualize the difference.

    In fact, research has shown that we are in fact more likely to donate money in responses to pictures with LESS people, presumably because seeing large numbers of people makes it harder for us to connect with them viscerally.

    Is this a good thing?

    I say emphatically no. To link our moral outrage to substandard emotional reaction is worse than cold, calculating objectivity because it is a disservice to the very values and people we are claiming to help.

    Personally, I refuse to be sidetracked in trying to objectively assess the greatest injustices and opportunities to help in the world, simply because some folks, riding on the emotional high of a personal experience or gut-wrenching image, have found a new cause-de-jour that I’m apparently too cold and objective to truly appreciate. There are millions of others, dying of much less sexy causes in much greater numbers, who would disagree with our emotional friend’s assessment.

  • In studies of the human ability to relate to extreme tragedy ie. death the capacity of empathy and shock for death is actually 1 individual. this is a surprising factor for such a social species, but perhaps a coping mechanism. Unfortunately it leads to a lack of appropriate response to crimes against humanity.

    Interesting the normal capacity for falling in love is 2-3 depending on the individual.

  • Lorenzo Morales January 9, 2009 at 11:31 am

    I really enjoyed what you have written today and how you went about defining “Flowing Data” here.

    You should consider adding this to the intro or the About section of your site.

  • I think some are missing the main point here. I’m not saying to feel like your mom has died a million times over when you deal with data nor do I want a novel for every single data point. I’m saying that there are stories behind the numbers. Quantify doesn’t have to mean sterilize – which just might be the most important thing I’ve learned as a stat grad student.

  • Nathan – you should check out our recent take on Stalin’s quote, and an experiment we did on how quantity affects how people think about morality:

    I thought the recent This American Life episode Numbers (or at least it recently came up in my podcast) was excellent. Although Ira called it a show about people trying to quantify things that shouldn’t be quantified my first thought was those people generated some fantastic data sets. The uncle who has recorded if he was happy in everything he did every day of his life? Wouldn’t that be a fascinating data set to play with?

  • @Lukas – ah, so that’s where that stalin quote leaked into my subconscious :). I was thinking the same thing. The uncle’s journals would be fun to play with

  • This post seems quite timely in light NYTimes’ latest interactive:

  • @William – nice one. thanks. The darfur wall also comes to mind now for some reason

  • I wonder if this isn’t a function of how our brains ‘zoom’, if you will, to interpret information from different situations.

    When we’re in close, we see an enormous amount of detail, and immediately register how any changes here will effect us personally (e.g. a rattlesnake poised to strike actually striking).

    When we zoom out, we gain a global perspective specifically because we’re not concerned with our selves (safety, social demeanor, immediate goals, etc.)

    Edward Tufte talks constantly about the ethics of visualization information. Beyond the integrity of any one graph or chart, it seems the larger concern is ensuring that we don’t damage the ability of people to shift from macro to micro as the situation requires.

    Relying on propaganda, indoctrination, and brutality all seem like attempts to keep people from looking at the world through one lens when you’d prefer them to look at it through the other, so as to limit their natural reaction to whatever truth the appropriate view reveals.

  • I posted an entry on my blog making reference to this blog post:

    <a href=””the map is not the territory

  • fixed link:

    I posted an entry on my blog making reference to this blog post:

    the map is not the territory

  • third time lucky :-P

    I posted an entry on my blog making reference to this blog post:

    the map is not the territory

  • well said, Nathan. though it can still be useful to treat one million tragedies as a statistic

  • @emily – definitely

  • Nathan—

    I’m looking forward to your thoughts on this article I’ve just written; I think it’s coming out in the next issue of HOW Design (possibly the one after). In it, I talk about this very thing: that how we present data directly impacts the meaning of that data and, in an age of increased visual literacy and dependency, info designers have a significant responsibility to consider how they use their visual toolkit to do this.

    The article will likely tread familiar ground for many of your readers, but hopefully it will bring home the point that info designers have more control over audience’s reaction to data than they may realize.