Home   About Kurtosis   Courses   Course Calendar   Booking Information   Ideas   Technique   Clients   Contact

The Magic of the Grid

Grids can be catwalks for data to parade upon

One of my favourite moments in Gary Hustwit’s 2007 documentary film Helvetica happens when Mike Parker is being interviewed and he starts to get crazy passionate about grids. Or, as he prefers to call them, matrices:

“When you talk about the design of Helvetica, what it’s all about is the interrelationship of the negative shape, the figure-ground relationship, the shapes between characters and within the characters, with the black, if you like, with the inked surface…. so the counters and the space between the characters just hold the letters. It’s not a letter that is bent to shape; it’s a letter that lives in a powerful matrix of surrounding space. It’s …oh it’s brilliant when it’s done well!”

Yes, I know. “A letter that lives in a powerful matrix of surrounding space.” He’s talking about a typeface for God’s sake. But I don’t care. Because right now I'm a massive, massive fan of anyone who can find a grid (or a matrix, although I personally much prefer the English word “grid” to the Latin word “matrix”) as exciting as that. Right now I’m developing a theory that involves grids being the answer to one of life’s big questions.

I recently wrote a piece on what I’ve started calling “Erin Brockovich” numbers, and why the sheer ease and painlessness of what Daniel Kahneman calls “System 1 thinking” will always act to prevent us from (a) stretching ourselves intellectually when we look at data and (b) moving beyond mere “soundbite” indicators of dysfunction into the more difficult (but ultimately more rewarding) territory of process indicators and an understanding of cause-and-effect relationships that can only be gained if we invoke System 2 thinking. In that piece I hinted at my descent into professional despair in response to this grim realisation. If human beings’ resistance to System 2 thinking really is as deep-rooted as Kahneman suggests, how exactly are we to get proper, serious numbers into the decision-making fray?

What I didn’t write about in the Erin Brockovich piece was that my despair was mercifully short-lived. And that it was another movie — Finding Nemo, as it happens — that dragged me out of my despondency. More specifically, it was the storytelling lesson that Finding Nemo’s writer and director Andrew Stanton learned in the making of that movie that gave me renewed hope.

Stanton called this storytelling rule: “Don't give them four; give them two plus two.” A year ago I wrote a piece about Stanton’s stunning TED talk, but without really arriving at a satisfactory solution about how we might successfully implement that rule when we’re trying to get data embedded into decision-making processes.

He explains this storytelling rule as follows:

“The audience actually wants to work for their meal. They just don’t want to know that they’re doing that. That’s your job as a storyteller, to hide the fact that you’re making them work for their meal. We’re born problem solvers. We’re compelled to deduce and to deduct, because that’s what we do in real life. It’s this well-organized absence of information that draws us in.”

I believe Andrew Stanton is dishing out a profound piece of advice here. People want to solve problems; they just don’t really want to know that they’re doing it.

But hang on a minute! Stanton is talking about storytelling techniques. Which may be fine when we’re constructing fictional narratives. But how on earth are we supposed to make his guidance relevant to non-fiction? To data analysis and data reporting?

Here’s how. (And those of you who were wondering where I was going with all that “held together in a powerful matrix” stuff at the start of this article should wonder no more because I believe it’s the grid that’s the secret sauce here.)

About two years ago I was doing some consultancy work with a team of NHS data analysts whose main problem seemed to be that their customers (senior managers and clinicians) were complaining of being given too much unscheduled care information. They felt deluged by data. The superfluity of data was impeding their attempts to make sense of it.

Accordingly, the first part of our solution was to reduce the volume of data being thrown at managers. But another part of the solution was that we decided to “colour-code” the data. We decided that if we clearly labelled data as belonging to one topic or another (or — more accurately — as belonging to one “staging post” or another), we’d help managers solve the Battle of Camperdown problem.

And in fact we got a good deal of feedback to the effect that yes, managers preferred it when data was fed to them “pre-sorted”, as it were. Managers liked that data was no longer being spat at them in random fashion, that there was an order to it. It was a filing system, almost. And we were pleased with our handiwork. Here’s an example—an incomplete example, as it happens, but that doesn’t matter—of one of our unscheduled care grids:

(Please don’t be distracted by the single-number-ness of the cells in the grid. The grid isn’t a standalone end in itself. No. It’s a starting-point. A springboard. A homepage. The idea is that you can drill down from each of the individual cells and get as many trends, as many distributions, as many different and detailed visualizations as you like. But you’ll always find it easy to get back home to the grid.)

What we found, however, was that the grid wasn’t just a filing system. It turned out we were simultaneously — and serendipitously — doing something else with our grid.

It turned out were implementing Andrew Stanton’s Don’t give them four; give them two plus two advice.

It turned out that whenever we presented these grids to managers — particularly when we presented them using PowerPoint and populated the grids one cell at a time — managers couldn't help but try to work out what the next number would be. And why.

We had almost literally stumbled upon a visualization that enabled cause-and-effect reasoning. It was as if we’d designed a solution whose objective was to facilitate System 2 thinking. Except we hadn’t. It had happened by accident.

So what was it about this grid that enabled it to do this “magical” thing?

Our first thought was that it might’ve been the colour-coding (although the colour-coding was also accidental, and forced upon us because of the dearth of same-colour post-it notes in the first meeting room we tested it in). We wondered if the grid’s resemblance to an ice cream parlour or a battenberg cake somehow made the data less, well, “intimidating” (because let’s face it, confectionery isn’t scary).

Then we wondered if the grid’s magic might be to do with the fact that we had — again, unwittingly — put the first of Edward Tufte’s grand principles of information design (Enforce wise visual comparisons) into action. The numbers we wanted managers to compare were positioned in immediate proximity. Moreover, the grid itself — and this is where the “single-number-ness” of each cell worked in its favour — allowed you to see the whole system in one eye-sweep, letting you easily compare any single number with any other single number.

But my own personal favourite theory about how the grid weaves its magic is this. The grid reminds us of a solvable puzzle. Unscheduled care thought about in the abstract all too often feels like a massive, wicked, labyrinth of a problem. An unsolvable problem, in fact. But when we describe this problem using process indicators placed in a multi-coloured 4 x 3 grid, we are — at least temporarily — disguising this System 2 complexity. Whilst we look at single numbers in adjacent cells, we can fool ourselves we are looking at a wordsearch puzzle. Or a crossword (quick, as opposed to cryptic) puzzle. Or a sudoku.

Or noughts and crosses.

The point is, we are making something complex appear solvable and we are appealing to our human instincts to solve problems through deduction.

Whichever explanation is the strongest, it actually doesn’t matter. I’m a pragmatist and frankly I don’t care which of these candidate explanations has the most bite. I just care that the grid seems to work. It seems to do that Andrew Stanton thing. It prompts data-driven cause-and-effect reasoning and it achieves it without the readers or viewers realising that they’ve started doing it.

I began this piece by quoting one graphic designer. I want to end it by quoting another. If there’s one designer who is the most associated with grids, it’s Massimo Vignelli (who, like Mike Parker, also sadly died last year). Pretty much everything Vignelli designed was constrained by the grid, but he insisted that the grid should never be visible to the viewer:

“The grid is an integral part of book design. But it’s not something you see. It’s just like underwear: you wear it, but it’s not to be exposed. The grid is the underwear of the book.”

Massimo Vignelli may well have thought that about his grids. But my grids are different. My grids are aren’t underwear; they’re more like the visible outer garments whose purpose is to make the wearer look as attractive and tempting and alluring as possible.

My grids are catwalks for data to parade upon.

[23 January 2015]

© Kurtosis 2015. All Rights Reserved.