Saturday, February 7, 2015

Redesigning a circular timeline

Yesterday a student of mine asked how to make an infographic like the one on the right in Adobe Illustrator (source). Click on the image to expand it.

I immediately tweeted that I would indeed show her some software tricks, but that I'd also explain why this may not be a good idea.

See, timelines (or bar charts, urgh...) shaped as circles are usually very hard to interpret, and not only because they force you to tilt your head to read the labels. It's true that they look pretty, and this one isn't an exception. It is very pretty. However, once you start trying to extract meaning from it, it becomes a bit frustrating.

Jer Thorp replied to my tweet:

I'm a fan of Jer's work —he's going to be one of the interviewees in my 2016 book— and, as I'm not in favor of strict rules in visualization,* I first conceded the circle might indeed let you cram more information into a smaller space. As for the idea that the original lets you clearly compare Dec. 2010 to Dec. 2011, I wasn't sure at all, and I said so in our conversation.

Anyway, I'm traveling alone today. I woke up really early and got bored during my long breakfast, so I decided to test my thoughts —and Jer's. In visualization you usually won't know if a particular shape works until you actually use it and compare it to as many alternatives as possible. I redesigned the timeline, using similar colors and font sizes as the original. Version 2 is much easier to read. I also put the new version next to the first one, to show that the circle turns out to be less space-efficient.

As a reminder, I believe that there's a difference, no matter how fuzzy it is, between information visualization (graphics to amplify cognition) and data decoration and data art which are both fine areas.

*I do believe that many flexible rules do exist, and need to be respected, though.


  1. Why the colors in the map for the involved countries is the same for 50,000 victims?

  2. Good question. I really don't know. I tried to stick to the color of the original, as I believe that it's a nice one, but it is true that this may be a confusing factor

  3. I'll have to side with Jer here. He's definitely right about the added length of the circle segments helping see the sequence of events in January. Also, the segments at the end of 2011 look like little blips in your version, but feel much more substantial in the original.

    Your redesign may be a bit easier to read, but I don't find the original that hard. They even labeled disconnected segments, so you don't have to trace them back. Titling my head isn't so hard (or even necessary) to read the text. This could be improved somewhat by adding shading to every other circle to make them easier to tell apart. It's even reasonably clear that the circle doesn't connect at the top, since there's clearly a gap.

    I think that this is an attractive graphic that works. Your version may be a tad easier to read, but it's also boring. Especially on a printed page, the original makes a lot more sense to me.

  4. I agree the original looks great, but answering some questions about relative duration of events is certainly harder on the circular layout.
    Since the countries are almost aligned horizontally (except Syria/Jordan), would it make sense (or could it be readable) to overlay the time-lines vertically over them on the map, as cartographers do for e.g. time-zones?

  5. It is not only harder but some comparisons are misleading. The distortion caused by the circular layout makes the timeline of Tunisia looks shorter than that of Yemen, when in fact the period depicted for Yemen is much shorter (about 15 days, I'd say, compared to approx. a month for Tunisia). That's why the segments at the end of 2011 don't look like little blips on the original version: they're exaggerated. However, the original looks way more attractive.

  6. Radial plots can be excellent, as this neat example by fellow student ( shows. But IMO Alberto's right that this one isn't. The fleeting wow moment disappears almost immediately into a puff of frustration as soon as you try and read it given all the real information has effectively been put through a beauty-mangle. If you think beauty trumps meaning you're doing a disservice to your audience (and in this case also to many unfortunates of the Arab Spring).

    A more specific gripe I have about this particular graphic is the glib reduction of complex political events to 3 simplistic events, with no further referencing or explanation whatsoever. If you seek to reduce complexity to this degree you really have a responsibility to explain your methods.

  7. I'm with the straight line version, if the purpose is clarity. To read the circles I need to perform, like, a zillion eye movements as a I move along the scale. In the horizontal version I can comprehend the X-axis immediately so I do not need to continuously refer back to it to know what part of the year I am in.

    I totally don't buy Jer's "radial layout helps in areas of visual density such as January". That's utter rubbish. IMHO :-)

  8. i used circles to build a timeline, a couple of years ago. it works (in my opinion), because it's about a cyclical phenomenon ( but i believe it's one of the few situations where circles would work.

  9. like Alex said, radial charts only work if the data has a circularity, and even then, it's not the easiest to read. In this particular case, because the data is spread over more than one year, the lines overlap, and the solution here is to force them apart by plotting one December next to the previous December. Imagine another month passes by. What is this design going to look like? I have some other comment on this, which I'll blog about soon.

  10. If we analyse them just under of the readability aspect appears clear that the linear one it works better than the radial one. However I'm thinking about the powerful of "eye-catching" that the first have instead of the second. Usually when something catch my attention I spend more time to analyse and go into deep of it and the circular one require just a little bit of effort by the be understood.