April 6th, 2013

Look ma, no story!

Storytelling has been one of the big buzzwords in data visualization the last year. By now, there are even whole conferences about the topic and I heard even some podcasts carry the word in their name :D

So, one could be tempted to think that storytelling is the magical ingredient to turn boring charts into killer visuals, make the blind see and save the world at large.

But, as so often, the pure and simple truth is rarely pure and never simple.

In fact, some of my favorite visualizations have no story to them.

map-market

Look at the legendary map of the market. A fantastic tool to understand the state of the stock market at one glance.

koblin-flight-patterns

Consider Aaron Koblin’s Flight Patterns.

xkcd-tic

The brilliant map of optimal tic-tac-toe moves by xkcd.

And so on.

Tools have no stories to them. Tools can reveal stories, help us tell stories, but they are neither the story itself nor the storyteller.

Portraits have no story to them either. Like a photo portrait of a person, a visualization portrait of a data set can allow you to capture many facets of a bigger whole, but there is not a single story there, either.

Let us not forget about these important genres. There is more to information visualization than punchlines.

Update: After a few twitter discussions, here is a clarification: I argue against the often heard claim “every good visualization tells a story”. I would agree that, in a loose reading of the word, you could say that some of the above visualizations “reveal stories” or might “trigger stories” in the viewer’s mind.

21 Responses to 'Look ma, no story!'

Subscribe to comments with RSS or TrackBack to 'Look ma, no story!'.

  1. Data Viz News [1] | Visual Loop
    April 6th, 2013 at 4:14 pm

    […] Look ma, no story! | Well-Formed Data […]

  2. Dino Citraro
    April 6th, 2013 at 5:45 pm

    There are stories found in the examples you provide, but they are made in the viewer’s mind. Perhaps you are suggesting that there is no narrative?

    The essence of data visualization is pattern recognition, and patterns are stories. For instance, related to the portrait example, a smile tells a story. As does a wrinkle or a scar.

  3. Moritz Stefaner
    April 6th, 2013 at 7:23 pm

    Thanks Dino, good point, added a clarification above.

  4. Marian Dörk
    April 6th, 2013 at 8:14 pm

    Interesting argument, Moritz. I also agree with Dino that the stories are being created in the viewer’s mind, but the ingredients are chosen by the vis designer. The characters (i.e., type of entities visualized), their costume (i.e., visual encoding) and plot (I.e., the plane and maybe interaction?) are to some degree pre-determined or at least shaped. So while your examples might not have the linear stepper that many journalistic visualizations have, i would argue that there is a narrative– not linear and fairly open. Nice thought provoker :)

  5. Moritz Stefaner
    April 6th, 2013 at 8:45 pm

    Agreed, but why do we need to call these constellations, insights, associations “stories”? Not everything that is “interesting” or “thought-provoking” or “surprising” is necessarily a “story”, right?

  6. Nikhil Sonnad
    April 7th, 2013 at 10:12 am

    Hi Moritz. We had a bit of an exchange about this topic on Twitter, I’ll add some more detail here.

    As I alluded to, I think there is a difference of opinion over how to use the word ‘story’. Essentially, I see a ‘story’ as any kind of narrative structure that a person applies to disparate observations, in order to make sense of them. So if you see a billiard ball hit a second, which moves, you would tell yourself the story: ‘the first ball caused the second to move’.

    However, even with this broad definition, I too disagree that ‘every good visualization tells a story’. But I do think it should be the primary goal of a visualization designer to get viewers to come away with something beyond ‘it was about X’. Many of the larger-scale visualizations today pride themselves on technical achievements — like the millions of words or kilobytes processed — but users unaccustomed to data analysis are easily confused by ‘here is a bunch of data’.

    So the goal of a visualization should either be 1) to tell an explicit story, 2) to facilitate easy creation of narratives by viewers, but never 3) to ‘put the data out there’ for users to deal with themselves. Number 2 is where dataviz really shines, but most ‘here is some data’ visualizations are nothing more than beautified spreadsheets.

  7. Stephan Thiel
    April 7th, 2013 at 10:36 am

    Thanks for bringing this up, Moritz!

    Reading also over the exchange on Twitter, it’s probably best to highlight one important issue in this discussion: the definition of a story & the meaning of storytelling heavily depends on the cultural context.

    While, sadly, I’m, not an expert in the field of intercultural communication, I have the feeling that I can at least identify some differences between the British, US-american and German view on the importance of ‘stories’ in society and culture.

    US-american and British communication culture, with many differences between themselves of course, is heavily based on the ability to ‘tell a story’ at all times. How you’ve met you wife, what happened at lunch and so on. This is the glue of every day interaction between people and highlights the importance of storytelling in these cultures. Along with that comes also a definition of a story which feels more ‘loose’ or general to me. That’s at least the conclusion I made for myself.

    Germans on the other hand often find it difficult to accept such ‘loose’ definitions. Again, I’m aware of all stereotypes and that there is no such thing as ‘the [insert nation here]’. But the German translation of the word story alone, Geschichte, makes me think we simply expect more ‘solid’ narration to make story a ‘real’ story (again, ‘real’ from a German culture centric point of view).

    For visualization and storytelling from an anglo-cultural point of view that probably means that the principle of closure (as in McCloud’s Understanding Comics) already fits in the definition of ‘storytelling’. And I think it is perfectly correct for comics, because the author specifically uses this method to narrate . But it is also often argued, that transitions between multiple displayed data points create a story in the viewers mind in a similar way. And there you have it: storytelling.

    From a German culture-centric view I think this is not true yet, since data visualizations simply are not at the level of narrative control you can find in comics in order to make it a ‘real’ story. And we still have to invent the narrative mechanics to create ‘real’ stories being told through data visualization. This is currently underway, I think, with many great recent works that better guide the viewer into a visualization (»U.S. Gun Deaths« by Periscopic and »Out of Sight, Out of Mind« by Pitch Interactive come to mind). But it is also not the goal of every visualization to ‘tell a (real) story’ as Moritz highlights. Many visualizations are simply tools that tell no story themselves, but enable the viewer to discover a story for himself/herself.

    That’s how I see this one part in this discussion: When is a story a story? And: When is a story being told? Mind the cultural context! If you have found these definitions for yourself, it will probably also influence your answer on the question if every visualization should tell a story.

  8. Moritz Stefaner
    April 7th, 2013 at 1:20 pm

    Wow, great comments! Thanks!

    Nihkil: thanks, very helpful and thoughtful clarification. How would you see the xkcd piece from above? Isn’t it worthwhile to do visualizations like these? But I think it does neither fit 1) or 2)

    Stephan: excellent point on the role of stories in different cultures. There is definitely something to it. I am fine and familiar with a loose reading of stories/narratives, and more so, even think that half of the time we talk about “story” in the “news story” sense (a la “what’s the story?”, “is there a story here?”), not in a narrative sense. Yet, still.. I would stick with my original argument.

    Don’t get me wrong: I am all for interesting stuff, and relevant stories, and digging for truths in datasets and finding interesting narrative techniques to present them.

    But: not every data set contains sensational new insights that fits in a single headline (and that actually check out). Much data is mostly confirming existing knowledge, is a bit complicated, messy, and is often ambiguous.

    To me, good visualizations are well-structured mosaics of facts that allow multiple readings. This is, in my experience, not really compatible with a (maybe naive, but common!) reading that there should always be “a story” and if that story is not super-obviously clear on first sight, then the visualization is a failure.

    Again, I have huge respect for all the tradition we have in information graphics and design with how to explain complex states of affairs, and am really interested in all the great new work that is emerging in the “narrative data vis” area. I just wanted to point out that there are many ways data visualization can shine.

  9. Jan Willem Tulp
    April 7th, 2013 at 1:56 pm

    Great discussion indeed!

    I have recently been thinking about what stories mean in data visualization, and how it is different from a collection of insights. I must admit that I don’t have a final answer yet, but I’ve bought some books to help me shape my mind on this. Both books are on the relation between storytelling, and the fact that we seem to be ‘wired for stories’.

    The stories that are discussed are mostly about character-based, but I will transcribe a short piece on the definition of stories and narratives that is mentioned in one of the books:

    The general term, narratives, may be plot-based event descriptions, stories (character-based) or information-based articles, reports, data sets and other similar documents. Information-based narratives provide just the new essential information and assume the reader has adequate banks of relevant topical prior knowledge to create context and meaning and sufficient related personal experience to create relevance.”

    And: “All three types of writing are narratives. Only stories are structured around the character-based informational elements receivers need in order to trigger and successfully drive the mental process that lead to understanding; to the creating of meaning, context and relevance; and to active memory.”.

    So, this could suggest that, making a liberal translation to data driven story telling, that some visualizations can have a different type of narrative, like an event description (or portrait, like the Flight Patterns one for instance), or an information-based narrative, which would perhaps mean a more or less be the “collection of insights”.

    And I don’t know, but I would say that if you would want to stick closer to the character-based narrative, you probably would want to have some linearity (or multiple-linearity) in your visualization, with a stepper, or some other time based elements.

    Anyway… still contemplating… ;-)

    P.S. in this book the dictionary definition is being critiqued as being focused on the plot-based narrative style, so don’t always accept the dictionary definition! :-)

  10. Amanda Hobbs
    April 7th, 2013 at 2:56 pm

    Wow — what an incredible discussion — thanks for posting. As a writer, editor, and info graphic research/content developer, I am fascinated to hear other’s thoughts on the role of data visualizations. As a journalist, my first instinct is to look for the story or meaning in a data viz. I want to find something fun, interesting, or inspiring. Yet, at the same time, I have a fine art background and certainly value the intrinsic beauty of visual patterns and display. For me, I think I can appreciate data viz that takes either approach, but I often think its a shame when data viz doesn’t “tell a story.” Visualizations can be such powerful tools for greater comprehension that it seems a waste not to take advantage of it whenever possible.

  11. Moritz Stefaner
    April 7th, 2013 at 3:13 pm

    Thanks Jan, good pointers. It would be great if you could post the links to the books..

    Thanks Amanda! To make this more concrete, what would this mean for the examples from the post? Do they lack a story? Would you say they represent missed opportunities?

  12. Jan Willem Tulp
    April 7th, 2013 at 4:18 pm

    sure:

    http://www.amazon.com/Story-Proof-Science-Behind-Startling/dp/1591585465/ref=sr11?ie=UTF8&qid=1365347892&sr=8–1&keywords=story+proof

    http://www.amazon.com/Wired-Story-Writers-Science-Sentence/dp/1607742454/ref=sr11?ie=UTF8&qid=1365347918&sr=8–1&keywords=wired+for+story

  13. Amanda Hobbs
    April 7th, 2013 at 4:20 pm

    I tend to think of “story telling” in the broader sense, as others have mentioned. Sometimes the best approach (though not always) can be to allow your audience to discover a/the story themselves. Regarding the examples included in your post — I think the first two fit into this category. To me, they tell a story, or at least invite you to discover the story. They spark your interest and awaken a sense of curiosity; they encourage you to ask “why?”

  14. Stephan Thiel
    April 7th, 2013 at 4:20 pm

    Jup, book links would be lovely! Thanks for the input. I have to read more in depth, to really figure out where the author draws a distinction between a story and a narrative. Not sure about the focus on ‘character-based information’ here, but anyways :)

    @Amanda: I think we all agree there should be a point in doing a data visualization besides the visual experimentation. If visual is the only focus, one at least should not call it data visualization but rather information aesthetics.

    But none of the above examples belongs in that category, I believe. Just thinking out loud, I would agree that they have a story, because they relate to something which a viewer can in turn relate to. This means, potentially, a richer story could unfold depending on the experience of the recipient. But that doesn’t necessarily mean they also tell a story because this would require narrative structures which the above examples do not have, or have they? (well except from the time dimension in Flight Patterns maybe)

    Would that distinction make sense?

    And I absolutely agree, Moritz, that visualization cannot and should not solely be a medium of crystal clear insights and stories to be told from data, since no just things exists as you said. Data visualization should always enable people to also ask questions about the data by observing and interacting with a created system.

    In my previous comment, I was more or less focused on what kind of stories are being told and how. And if that can be considered ‘storytelling’ in the first place.

    Thanks all for the great discussion!

  15. Erik Jacobsen
    April 8th, 2013 at 8:21 pm

    Thanks for the thought-provoking discussion.

    One reason to argue against the idea that “every good visualization tells a story” is that it’s not a very helpful statement — partly because the definition of “story” can be broad (as seen in this discussion), but also because it doesn’t give us much to go on if we are creating or critiquing a visualization.

    Rather than asking “does it tell a story?”, better to pose questions that get at the qualities Amanda alludes to: Does it spark interest? Does it awaken curiosity? Does it encourage additional questions and research? Does it lead to insight and understanding?

    Does a visualization have to tell a story to achieve those aims? I think that is what Moritz is pointing out here. Each of the examples presented does those things without telling an explicit story.

    On another level, telling an explicit story — with an annotation layer or other narrative device — can be fraught with peril. It’s not hard for the narrative to be colored, especially when the creator (or the client) starts out with a specific agenda or a point to prove. These biases are easiest to detect when they go against our own world view, but they can be as dangerous to us as the emperor’s new clothes if they stoke our pre-existing opinions.

    Is it possible that some of the best visualizations are tainted by a story? Or at least by an unreliable narrator?

  16. Bruno Carneiro
    April 9th, 2013 at 2:18 am

    I like to think of “stories” as being the connections between different blocks of information. Explicit or implicit in any given visualization piece it’s always through the reader’s eye and brain that the connections that create meaning take place. The reader is always part of the “story” and there’s always one. In fact he can’t help but creating one in order to make sense of what he sees.

  17. […] Moritz Stefaner […]

  18. Francis Gagnon
    April 15th, 2013 at 2:06 am

    Most discussions about data visualization assume that the data set is available before the visualization is created. This makes sense in the media, but it is not the case in the business world. Dashboards and reports are created before sales, revenues and expenses figures are known. It is not possible to choose a data visualization that shows a particular story found in the data. Yes, these graphs need to support understanding. I think it adds to the point of view that not every visualization needs to tell a story, whatever meaning we give to the word.

  19. christian schaefer
    April 15th, 2013 at 4:02 pm

    i am very pleased about what is beeing discussed right here; moritz, to me, this is a good and up-to-date question.

    a story, besides differing cultural contexts, has always that notion of maeandering chronology of facts, which are entertaining and can be told. and this would be the point: a story is something that can be spoken of in a manner that is entertaining. because this is what a story is about. there is no fundamental cultural difference between that attribute, hence, stories are prototypical. this has been like this for centuries. the bible is a huge piece of a story – the grade of informativeness may differ from person to person.

    the paradigm of digital storytelling by means of data visualization techiques may not be the right one at all. hence, data vis work does not need to tell a story at all; it may be better of if there is no story at all. by definition, it would assume discussion, at least online, literally verbal. hence, the grade of ‘storyness’ may not be an attribute of just the design itself, but more about a specific form of beeing re-told.

    having a look at how young citizens are communicating right now: they are ping-ponging data. it is associative, at high frequencies; content is constantly re-whatevered. on the contrary, to tell a story, you need time, a story implies linearity to be memorized. here, i would claim that the re-whatevering of content fundamentally differs from the way of retelling of a story.

    a story is not a data visualization, and vice versa. they are just not identical. a vice versa example would be the appreciated book of stephen silverman ‘quicksilver’. it aims at beeing a kind of a data visualization by means of telling the story – here, stephenson tries to illuminate the development of stock broking, beeing a quite impressive informative history book. in the end, it cannot be read as a story anymore (you could assume that i am too stupid to read it as such, but i am still self confident enough to respond that it is more a matter of statistics than personal capacity ;)).

    just because you imitate structural properties digitally, it does not necessarily mean you archieve your goals the same or even similar ways. i prefer a perspective where you just combine certain data to enable a user to see some difference. this may be planned and structured, but not necessarily at best by means of book or story metaphors. data visualization itself is not about affirmation, but about instrinsic categorical changes. the entertaining factor implies marketing purposes, and is not inherent to data visualization itself.

    there is this writing technique of cut-up, where chapters can be read in any order. or, to put it another way: there may be linearity in time, and there may be graphing and hierarchies in space. these are topological and time-related dimensions you are dealing with, visually. those self-referential expositions, commonly refered to as data visualizations literally gain their denomination – they are nothing more than just that. the foreground-background problem of congnition presumes a context, not just some graph structure, chosen out of some arbitrary set of visualization methods, as provided by, lets say, this guy: http://d3js.org

    both, a story and a data visualization share the fact that they are biased. there is no such thing as an unbiased comparison of two pieces of data. if i chose to compare china to japan, it may fosters some insights A, and if i compare germany to japan, i may get some insights B, they are not identical. hence, just by choosing two elements out of the same class of objects, comparing. anything that has to do with statistics is biased, as are stories. hence, to claim something to be unbiased always means that the author hides from responsibilty, or this dimension is simply irrelevant to him.

    it all comes down to your goals – what do you have to say, or show? do you have something particular to say, or do you want to arouse associations, or discussion, or just to be instructive?

    and now here is the synthesizing part: it all comes down to information emergence. what do you want? you want your things to be seen and understood, and you also want memo values. do you?

    hence, the question of whether a data viz has to tell a story, or not, is, as moritz pointed out, arbitrary. another arbitrary question would be: how can positive aspects of memo value, categorization, informative potential, in relation to (analogue) stories, be actually used in (digital) data visualizations?

    what properties do we have? we have pages, chapters, creature relations, history, linearity, character studies, spacial localizations in form of scenes, we have language (in contrast to images, which are quite frequently used to show context and to evoke meaning, which, in itself, is quite problematic), we have paths of difference and repetitions, associative priming, redundancy, and we also have periodicity, and we also have a thing which could be described as ‘enough space for intrinsic or personal imaginations’ (and no interaction btw;)) so even looking at things quite formally, many data visualizations are anything but about story telling.

    the term story is as diffuse as the term information. hence, they may be equally interesting.

  20. Fredric
    September 19th, 2013 at 11:15 pm

    When I originally commented I clicked the “Notify me when new comments are added” checkbox and now each time a comment is added I get four e-mails with the same comment.

    Is there any way you can remove people from that service? Cheers!

  21. […] Fazit: Eine ansprechende Visualisierung allein erzählt noch keine Geschichte. Das ist eines der großen Missverständnisse beim Thema Datenvisualisierung. Vielmehr geht es […]

Leave a Reply