Too sexy for its own good? Ebb and Flow At the Box Office

You can click the graphic to see it in higher resolution at the New York Times' website

The graphic shown alongside this article has just won the Malofiej-award as the best graphic produced in the year 2008. Do you agree with the judges on this one?

I won’t be too surprised if you’ll nod your head and say yes! - the graphic has been widely discussed already in the blogosphere, and most of the comments I have read, is very enthusiastic about the visualization of box office receipts.

Myself - I’m not too sure about this kind of ‘visual journalism’. My main gripe with a graphic such as this is - Where is the story? The headline sorts of point to this problem itself: ‘Ebb and Flow At the Box Office’ - sounds a bit like a working title for a dataset, where the journalistic angle is yet to be decided.

Fun With Numbers

At any time I’ll acknowledge the technical execution, the subtleness, the elegance, the massive dataset. But I still have questions for the real usefullness of this graphic. Would it be too rude to call this category ‘Fun With Numbers’? - Or a more polite version: ‘Art With an Infographic Function’.

Reading the comments from some of the readers has been fun and a bit disturbing too. Someone named Codepink at Gawker gives us this observation: ‘It’s like looking at a vagina in 3-D movie but when you don’t wear the 3-D glasses’ and a few comments later you get MisterHippity chiming in with: ‘To me, the colors suggest a flowing stream of diarrhea. The anal leakage of Hollywood, if you will.’

(Note to self: Readers can really get anything out of your graphics … )

A closer look at the visualization technique

Well, back here at VisualJournalism, I’ll concentrate a bit on the visualization technique rather than what the shape and color might resemble. (And actually I do find both the shape and the colors very pretty).

Let me point to some of the details:
1. The smooth lines are certainly a pleasing aesthetic choice - but they are incorrect …
1a. The Streamgraph is made up from weekly numbers - and it should be obvious that a movie opens with full impact in the first week. But this important feature is not shown correct in the graph, where it takes a while before the movie is at full height.

1b. When you look even closer to the graph, and compare it with a traditional barchart you’ll notice that the areas for each week simply aren’t correct. There is too much ‘friction’ in this method to show the numbers correct. In the example with Spiderman 3 you’ll notice how a lot of the opening week spills over to the second week.
Perhaps these problems could be solved if we give up a bit of the sexy smoothness, and instead start each curve with the full datavalue rather than starting out from zero every time?

In the paper about Streamgraphs written by Lee Byron & Martin Wattenberg you’ll read this:
“A key theme is the role of aesthetics in visualization design, and the process and trade-offs necessary to create engaging information graphics”. You can read the entire 10-page paper here (.pdf). It’s very interesting and very thorough. I’ll highly recommend it.

2. When you choose to show something symmetrically you give your readers a hard time deciphering the actual happenings.
Look at the example where you can see how there is a definite ebb between Spiderman 3 and Pirates of the Caribbean. But then on the other side of the axis you see Shrek 3 shooting up. You get a lot of sexy smoothness and curves again, but what is really happening? If you want to see the total box office earnings, you’re left in the dark.

3. Specifically for the online version - which is tilted 90 degrees and run horizontally instead of vertically as the printed version: It is really confusing that going below the middle is just the same as going upwards. Everyone always assumes that up is good and down is negative, so unless your users are really into the thing, you’ll risk confusing them to an extent, where they’ll give up on the interactive before they understand it.

And finally - when I see a cool service of ‘Click for details’ for every movie in the interactive, it’s not the NYT Movie overview I’m looking for. I want to be able to see the total box office earnings for a movie or being able to zoom in on the weeks and see number of theatres or the ebb and flows of the daily earnings.

All that said - I can certainly see why the judges made their decision to give this entry a gold medal. And it’s not too often you see an infographic piece like this discussed in a broader community.

/Gert K Nielsen

CATEGORY: Editors Picks, Graphics, Malofiej2009
Read more about:, Box Office, Streamgraph, Visual Journalism.
Published: Sunday, March 29th, 2009

7 Responses to “Too sexy for its own good? Ebb and Flow At the Box Office”

1. Sean McNaughton says:

   April 22, 2009 at 13:53

   I’m sad that I missed this graphic when it was in print, and I’m really enjoying the insightful discussion — but I wonder if some of the criticism isn’t founded on assumptions that bear a bit of examination themselves…

   I’d like to pause with the “Where’s the story?” comment. Clearly, the vast majority of journalistic infographics must make a clear and incisive point; having worked in daily journalism as a writer and artist for 18 years, this makes perfect sense, and our readers are often best served when we start by focusing on our “nut graph.”

   But is that the only use for infographics? Clearly, the explosion of encylopediac infographic newspaper pages starting in the late 1980s would say otherwise: What is the point of the two-page graphics for every new sports stadium built? To show people the facility (and we’ll leave aside for now any discussion of the value and proliferation of 3d stadium models). But is there “a journalistic angle” beyond that? How would we summarize the stories of these graphics? They’re not as straightforward as most journalistic graphics, but the value of the approach is to give readers an explanatory visualization.

   Similarly, I think the value of graphics like the one we’re talking about here is that they allow people to explore information that would otherwise be utterly opaque and boring to all but the most dedicated spreadsheet addicts. Your point that the streamgraph is less correct for each individual movie’s data is entirely valid — but is that the point of this visualization?

   I think not. Each mode of visualization foregrounds specific aspects of the data, even as it masks others. The streamgraph approach masks the granularity of individual movies’ box office receipts, but gives us an elegant and engaging overview of the rhythms and patterns evident only in an “Apollo’s eye” view of EVERYthing that’s happening.

   Looking at this, I am able to see and understand things about (1) the Hollywood movie industry, (2) the moviegoing public, (3) the economics of the film industry, (4) the social and economic value of Oscars, (5) the seasonal rhythms and how they’re reflected in movie content — and this is impressive. I can’t think of another approach that would make available to me this kind of depth, richness, and complexity in a way that is half as engaging.

   But what I love most about this approach (and yes this graphic) is that it refuses to compromise: Rather than cutting information to make this simpler or “easier,” Amanda and Lee have organized, correlated, structured and presented complex data in a way that honors both the data integrity and that of the reader.

   Nice work.
2. Jane Nielsen says:

   April 9, 2009 at 13:48

   The colours in this graphic are quite awful and the image has absolutely nothing to do with films. I´m sure Peter Sullivan would be turning in his grave if he saw this graphic!
3. Lars Pryds says:

   April 3, 2009 at 22:47

   Hi Gert - good work. I’ll grant you (and Lee) that this is a tremendous accumulation of data, be it accurate or not so accurate. This discussion aside, I doubt many people will actually delve into the numbers of this graphic, cleansed as it is from all visual impact. Where are all the goodies in the shape of the odd movie still here and there? It’s all smooth lines and text, which – sorry to say – leaves me cold. And I haven’t even mentioned the colours…

   /Pryds
4. Gert K Nielsen says:

   March 31, 2009 at 06:48

   I agree that interpolation can be a good thing, but every time I get my computer to do work for me, I’m still trying to check it up against reality.

   (You know how some old folks use their calculator to add together a whole list of numbers, and when they get the result, they calculate it an extra time by hand, just to be sure?

   Movies are perhaps such a crazy market, that it ‘breaks’ interpolation. Or at least need some extra experimentation before we get the picture we’re after.

   If we have a look at the numbers from the Spider-Man 3 opening week, there’s not much up a buildup - rather a headstart explosion, which is hard to get interpolated when starting from zero:

   Day 1: 59.841.919
   Day 2: 51.336.732
   Day 3: 39.937.865
   Day 4: 10.285.268
   Day 5: 8.042.682
   Day 6: 6.717.488
   Day 7: 5.908.618
   adding up to 182 million dollars
5. Lee Byron says:

   March 30, 2009 at 23:20

   Gert - I certainly think this specific pitfall could very potentially be minimized by starting layers at full height rather than interpolating up from 0.

   Interpolation is a tricky subject which is always the subject of hot debate. I must say that interpolation is more than just a purely aesthetic decision, and can often help people better understand a data set when the available resolution is low or contains unrelated signals (such as weekly undulations). That said, filling in the gaps in the story has pitfalls as well, and those pitfalls are different based on each story, data set and visualization technique.

   The problem which you have clearly (and rightfully) criticized in this case, is the misleading interpolation specifically between the pre-release week (which is always 0), and the first release week (which will always be high). You can assume some lead up of course, but likely for movies, that lead up is very fast - perhaps much of the revenue in the first week is made in the first days - and so this interpolation which moves from 0 to peak over the course of a week is likely not accurate.

   I think this is a great observation, and could be an interesting modification to the streamgraph technique. I intend to investigate. Thanks again for the thoughtful critique.
6. Gert K Nielsen says:

   March 30, 2009 at 20:16

   Hi Lee -

   I’m happy that you’ve read the critique of this graphic. I do like the look of your work, but as you can tell, I personally think the trade-offs are too big, when you give readers the impression that movies start out slow and takes a couple of days to reach full sale.

   I have tested this claim - everyone seeing this graphic reads it like this - and even starts to make up theories about why this is so:
   #1 Moviegoers hearing from friends it’s a movie to go watch.
   #2 Movies starting out in a limited amount of theatres.

   When they hear it’s just a ‘calculated fault’ with the visualization technique they feel very disappointed.

   If I show them the Spider-Man example above, where it’s obvious that weekly sales aren’t visualized truthfully with a streamgraph, they are ready to give up on the graphic.

   Do you think the problems could be solved by giving up the (purely aesthetic) idea of building the area from zero?
7. Lee Byron says:

   March 30, 2009 at 03:22

   This is a fantastic critique and really highlights the trade offs made designing this graphic. Density vs detail is a common theme in printed graphics. Interpolation vs discreet units is another.