Tufte Talk: Chapter 4

Chapter 4 is Data-Ink and Graphical Redesign.

So here’s where we really start getting into the question of design. The Aesthetics, if you will. Chapter four beings a new section of Tufte, “Part II, Theory of Data Graphics.”

The look of a data graphic is a unique subject. It’s a blend of factual and emotional, mathematical and ephemeral, truth and philosophy. To me this is broken into two questions: Is it clear? And is is cool?

Tufte doesn’t waste much time talking about aesthetics. He instead offers a very rigid and rational approach to revising your data-graphics. By showing the revision of foundational data artists (siting Playfair a lot), Tufte points out that there is tipping point between “ink used to convey information” and “Ink used to convey redundant information.” Nothing hits this home quite like showing off an early Playfair to a later Playfair. A LOT of the lines are reduced, which makes focusing on the actual information - the insight you’re looking for - easier, faster, and bolder.

Most of the chapter is dedicated to further illumination by repeating examples of this. In some cases Tufte even takes a single charge and shows you the whole of it, then just the elements that convey the data, and then the parts that were ‘erased’ or cut away from it to help focus on the data.

But what’s really interesting here is the principles for this chapter:

Above all else show the data.

Maximize the data-ink ratio.

Erase non-data-ink.

Erase redundant data-ink.

Revise and edit.

While this is how they’re stated at the end of the chapter, within the text Tufte says them differently:

Above all else show the data.

Maximize the data-ink ratio, within reason.

Erase non-data-ink, within reason.

Erase redundant data-ink, within reason.

Revise and edit.

Here you can see that we’ve crossed the bridge into aesthetics. Because it’s no longer a solid formula - everything, when actually engaged as a topic - gets appended with “within reason.”

So, naturally, Tufte spends the last bit of the chapter explaining when to not do the things he just said to do.

Don’t get me wrong: I don’t see this as a contradiction. I think a lot of aesthetics, and a lot of analytics, are so dependent on context that rules have to be seen as principles, at best, and even those principles have to be stated more like questions rather than tenets.

Think of it this way:

Are you showing the data, or are you showing a lot of noise-to-signal?

Can you reduce the non-data-ink?

Should you show redundant sequences, or can you par it down to one stretch?

This is the process of revising and editing your data graphic. You put down everything you need to show your work, then you take away the parts that distract, over-complicate, or strain the understanding of your insights.

I wonder what Tufte would think of the modern options, too? Where do we fit questions like, “when do you hide specific details in a tool tip window rather than placing the information as a label on the graphic?” Or “can we show this data better with a dynamic pivot table, or does the viewer need to see it all in one space?”

I’m especially intrigued by these thoughts as I dive deeper into the art of Infographics. I’ve been working on one for a week now, and the process is so incredibly different from the means to making a data viz through Tableau or PowerBI it’s worthy of a post to itself. One is data exploration, the other is data storytelling. One is clearly graded on clarity, the other on impact (at least that’s how it feels right now, ask me again in a month, maybe I’ll have learned more).

My biggest take-away from this chapter is to pay attention to that tipping point between signal and noise. Too much detail and your data graphic takes work to interpret. Too little detail and your data graphic is confusing by lack of context.

Above all else show the data indeed.