#3: The Limitations of Displaying Detailed Data

There are a few stages to data journalism. First you have to get the raw data, then you need to pull insights from it, and finally you need to tell a story. And it’s the telling of the story that things get really complicated. Text is easy — write some words into a system like Medium and they’ll take care of the rest. But what about actually presenting the data?

You could just export a graphic and embed it on the screen, of course. This approach works well for simple insights and datasets such as this one:

If you’re asserting that we spend more money dining out than on groceries, this graph succeeds in a supporting role. But what about insights that are more textured and multi-contextual? What if you wanted to compare across different states? What if you wanted to see how much of this data is driven by alcohol? You could export more graphics, but that doesn’t work.

First, graphics don’t adjust across different screen sizes. So when you see a graphic that’s a bit hard to read on your phone, you have to tap it, zoom in yourself, and pan around. That makes sense for some media, but it seems crazy that we do it for media. We have the technology to adjust tables of data based on the screen size of the consumer’s device. Put another way, we can do responsive graphs, so we should.

Second, graphics have clumsy and inconsistent visual design. When I read blogs that are showing a lot of data, each graph uses different colors, different type, different everything. It makes the publication feel slapdash and amateur. The graphics don’t feel like they’re supporting the text, they feel like found art and bumper stickers pinned on a high schooler’s wall.

Third, graphics aren’t accessible. Google can’t index them. Screen readers can’t make sense of them. You can’t copy and paste text out of them. They’re lifeless blobs of bits rather than being dynamic and flexible text.

But the biggest problem is they’re not software. Good data software is like a vehicle on safari. It lets you explore and learn at your own pace, in your own way. Good data software helps you find the big insights on your own, but lets you discover your own insights too. But graphics can’t do this. Data graphics don’t feel like a tour of the outback, they feel like a picture of a kangaroo that comes with ten quick facts cribbed from wikipedia.

So I’ve been trying to figure out how to deal with this. One of the first insights I had is that I can’t use Medium as much as I was planning to. There is no “Medium for infographics essays,” but maybe there should be.

Until then, I’ve been putting together a style guide and strategy for how I’m going to publish this data. I want my stories to hang together in a consistent visual language. I want the data to be a part of the story, not giant graphics set to the side of the story. I want the data to be malleable and full of life, so I have to build it out with code, not screenshots.

So I’m working on that now.