Here at BuzzData we’re still having stimulating debates about UX, visualizations and everything in between. After batting ideas around with other team members, I lobbed a few questions over to data-viz wizard Noah Iliinsky, technical editor of Beautiful Visualization. His views, transcribed here from a phone interview, were as open-minded as they were acute:

As data visualizations become more popular, it seems the line between data visualizations, infographics and visual art are becoming less clear. Where do you draw the line?

N: If you get to the point where the data is no longer meant to be consumable as data — if you cannot take meaning from it, you can say “that’s a pretty picture” but you can’t see a trend or a pattern or anything in it — I think then you’ve transitioned over to art. Which is fine, that’s acceptable, but I think that’s an important distinction to make.

You can do anything you want with your data, and of course it’s a “data visualization,” but you wouldn’t classify it necessarily with data visualizations that are meant to actually convey knowledge.

[Iliinsky then deconstructed what he believes is the difference between an infographic and a data visualization. Surprisingly, it’s not necessarily about purpose, it’s more about process:]

N: Infographics are the ones that are usually illustrated by a graphic designer; they’re probably done in Illustrator, there’s some data in them, but they’re not necessarily data-rich. They tend to be manually authored, manually constructed — obviously on a computer — but somebody sat down and said, “we’re going to put the big windmill here for ‘more windpower’ and more sunshine for ‘more solar power’ and a smaller oil barrel here” or whatever. That’s an infographic.

A data visualization tends to be generated automatically or algorithmically. Someone sets those (algorithms) up, but from there you can point to the data source and, as the data changes over time, you can regenerate the graphic trivially. It’s not something somebody had to draw.

That’s the fundamental, definitional divide.

[According to Iliinsky, however, even this divide isn’t all that significant if you what you really care about is effective communication: ]

N: I came into visualization academically, through a masters program that focused on the needs of your users and satisfying their informational needs.

So that was the headspace I was in when I started drawing diagrams: “Let’s not just represent the structure or the data; let’s really do this in a way that’s conductive to it being useful, that really addresses context of use.”

That middle ground, that sweet spot of “who is my audience, and what do they need?” — I find that that is lacking in other camps. You have the “make it pretty” camp, or “make it a sound bite” camp (from the infographic side ) and then you have the “put every possible data point in the universe on one page” camp from the big-data end of things. Neither of them are really coming from a user experience (UX) tradition.

That sweet spot of “how do you really make this visualization a solution to somebody else’s informational problem?” — that’s when you get a real success; that’s when you get something actually useful and good and fun.

A while back I witnessed in a passionate discussion of whether or not a PivotViewer visualization was “good” or not. We couldn’t really come to a consensus because we couldn’t agree on whether putting data out there for the user to play around with is in itself valuable. Or, does it fail because you’re just handing over a bunch of data and and not pulling apart the story for them?

N: I’m gonna quote Bob Dole here and say: “It depends.” The answer is context of use. One of the fundamental considerations — and this is my UX training coming in — is: what’s it for? What problem are you trying to solve with this? Once you answer that, then you can talk about whether something is in the right format. But you first have to understand what it’s for.

There’s nothing wrong with data exploration, there are some really great tools for that. Some of the PivotTable tools are fun for exploration because you can slice and dice and sort in different ways, but those typically don’t have a single message. They don’t usually have a single story they’re trying to tell. They say: here’s some data, explore it and enjoy it.

Context of use, baby, it’s everything!

I’ve gone back and forth with others at work about the value of Planetary — some of us consider it remarkable, some of us don’t. Would you call Planetary a data visualization?

N: Sure. I mean, it’s a visualization, right? It’s based on data; it’s generated, they’re not hand-drawing each planet for every track. So sure, it’s a data visualization. It’s a highly stylized data visualization.

(Planetary) is heavily aesthetic which can make it more appealing, and that’s not a bad thing. If you define the purpose as a cool way to browse your music, then there’s nothing wrong with it. Or maybe it’s a cool way to view your music.

Is it successful? Well, it’s pretty, it’s awesome, people talk about it, so it might be successful by that measure. Is it a successful data visualization? It’s probably not the first thing I would use to visualize my data.  

Richard McManus of ReadWriteWeb recently wrote about how Planetary might represent a shift to using data visualizations as UIs. Considering how hard it is to create an effective data-viz on its own, adding a UI purpose on top seems like a challenging ideal.

N: It’s a tricky thing — but it can be done well, there certainly are visual interfaces that make life much better and easier. For example, go search for a flight on Hipmunk (like Chipmunk without the C):


http://www.hipmunk.com/#!Toronto_Berlin,Jun27_Jul02

Isn’t that gorgeous? It’s amazing, right? See the “agony” filter? “Agony” is a combination of price, time of day, number of stopovers. That’s the one you want! That’s really smart.

And it’s got the time bar across the top will have both time zones listed — just spectacular. You’re not doing a lot of manipulating of this data, but if you mouse over a flight, it gives you specific times, and if you click on one, and there’s your details. That is sweet. That is really excellent use of visualization, not just for representing what’s there, but giving you really clean access to what you need.

[I also asked Iliinsky if he could cite an example of a poorly conceived data visualization and explain its shortcomings. He offered the Periodic Table of Controllers, which he blogged about back in May:]


http://complexdiagrams.com/tag/periodic-table/

N: This just takes technology and pours it into a periodic table-shaped box. This is a rant, okay? [laughs] So there’s the word “periodic,” right? And the periodic table of the elements is periodic because the elements have properties that are periodic, so when you put them in the table, the periods line up and you get a periodic table.

Unless your data is periodic, don’t put your data in a damn periodic table! It’s that simple! 

There’s the periodic table of dog breeds and desserts and Google APIs! But the one that makes me cry, the one that strikes me to my very core, is the Periodic Table of Visualization Methods. What people want mostly from these tables is a categorization, but there are great categorical groupings for visualization methods and for Google APIs — do it categorically, do a chronology. Timelines are great — it’s a really powerful axis, that time axis, because you can see where there are clumps and trends. Pour it into a box like [the periodic table] and you get none of that.

[Iliinsky then offered an alternative visualization of the same data that uses chronology to much richer effect:]


http://popchartlab.com/collections/prints/products/the-evolution-of-video-game-controllers

This is a family tree; it shows the interesting influence of the different controllers. Which is awesome and fun, and totally valid and useful — like, “Oh, that’s information that makes my life easier” — that’s the one you want. Because it actually puts them in a meaningful context, not just a chronology.

And that changes everything.

-Momoko Price