Purpose of Design #

~5 minute read

What is the point of ‘good design?’

Why do some data visualizations get ignored, while others are shared around the world? Why are some tools a joy to use, while others are a frustrating mess?

It’s not magic; it’s design. Good design isn’t just about making things look nice. It’s a functional discipline built on three core goals.

A well-designed object:

Makes the task easy.
Feels natural to use.
Is beautiful.

Let’s look at each goal in turn.

1. Good Design Makes the Task Easy #

Bad design can render objects hard to use or confusing. From forks to signs, even simple objects are not immune to bad design.

Two images illustrating bad design. The image on the left contains a fork, but the head of the fork is joined to the stem via a metal chain—making stabbing impossible. The image on the right has a sign which reads “People are eating children in this area.” The lack of an “and” between eating and children creates humorous misinterpretations (we hope).

Bad design gets in the way. The chain fork (left, by Katerina Kamprani) fails its primary task, while the sign (right) fails at being clear. Both are examples of design that prevents a user from accomplishing their goal.

In the above examples, the quality of the designs is judged against the task they are trying to enable: the fork is supposed to allow someone to stab food with the tines, while the sign is supposed to convey information with clarity. Thus, when you designing an object, keeps its purpose in mind; always remember what you are trying to enable the user to do.

Let’s assume that you are making a data visualization for a publication or presentation. Presumably you are doing so because you have found something in your data that you wish to communicate. Therefore, good design in a data visualizations seeks to enable the reader to gain this same understanding in as simple a manner as possible.

Above, we saw that even simple objects can be ruined by bad design. The same is true in the world of data visualizations: one can make something as simple as a line chart nearly unusable. Below is a figure that I created some years ago. The intent behind the figure was to allow comparison between the infra spectrum of an inorganic compound in different solvents. However, this plot attempts to show too much data at once and does not make good use of consistency and contrast, rendering the desired task more difficult than it should be (for example, try comparing bands in ethanol and acetonitrile). The deign could be improved, for instance using the technique of small multiples, each with high consistency and contrast. This change makes it simpler to accomplish the desired task, and so makes the design better overall.

A line plot containing 5 different, strongly overlapping, spectra. The only parameter that distinguishes the lines is color, and this makes it hard to read the chart.

Before (Top): An image from one of my publications. The goal was to compare spectra, but by overlaying five colors, the plot is hard to read and fails to make the task easy. (Try comparing the blue and green lines.)

After (Bottom): An improved design using “small multiples.” Each plot uses consistency and contrast to highlight one spectrum against the rest. Now, the task of comparing spectra is much simpler.

The same spectroscopic data shown as above, but now they are represented using 5 small multiples. Each small multiple shows all 5 spectra, but in each 1 spectra is colored and the rest are grey.

2. Good Design Feels Natural #

Good design does not exist in a vacuum. You are making data visualizations for people who are used to seeing things presented in a certain way. If you deviate from that expectation, you make the object feel less familiar, and make the viewer work harder to understand it.

In the real world, we might find a green stop sign confusing, while in the world of data visualization we might find an ordered bar chart of stock market returns confusing. Both of the underlying objects are familiar, but their implementation is not.

Two images showing a lack of consideration of convention. The image of the left is a stop sign that is colored green instead of red. The right image is stock returns over time, represented as a bar chart, rather than a line chart.

There is nothing inherently wrong with a green stop sign. However, we find it confusing, because it does not meet our expectations. Similarly, there is no reason not to represent stock market performance using a ranked bar chart (which highlights what returns are most likely); however, since we are more used to seeing this represented as a histogram, we find this visualization confusing.

This is not to say that you shouldn’t try new things. Indeed every current ‘standard’ was once a new idea. However, when you are trying something new, you need to acknowledge that you are increasing the burden on the user, and ask yourself if the perceived gains you get from the new approach outweigh this drawback.

3. Good Design is Beautiful #

People like beautiful things. We enjoy interacting with them and looking at them. Thus, if you make a beautifully designed object, you will create something that people will choose to interact with. That is powerful.

Some time ago, dedicated portable digital music players were extremely popular. Apple created the iPod. Archos created the Jukebox. While the Jukebox was arguably a technically better device by nearly every metric, only one of them is recognizable by most people today, and it is the one with the simpler, more natural, and better looking design.

Two images of MP3 players from the early 2000s. The one on the left is the iconic ipod, the one of the right is the much lesser-known Jukebox. The ipod is all white with 6 input buttons and a large screen. The overall shape is a simple rectangle. The jukebox has many different colors, a small display, 10 buttons, and a lumpy shape.

The music player on the left eventually became synonymous with portable music devices, while the one on the right is largely forgotten. The one on the right was much more capable. The one on the left, however, was much more beautiful.

A common complaint of scientists is that it can be hard to get people to pay attention to your work. However, if you can make the iPod equivalent of a data visualization, you will have people sharing your work for you. Consider the two maps of election results shown below.

An image showing two political maps of elections in the USA. The one on the left is a choropleth, with two colors (red and blue) and no labeling in the states. The one on the right is a choropleth with 8 different colors and significant text labeling.

The map on the left (from FiveThirtyEight) is beautiful, simple, and answers the one question viewers care about most: “Who won that state?”

The map on the right contains more granular data, but its complexity and unconventional colors make it harder to read and less aesthetically pleasing. Unsurprisingly, the clear, beautiful design on the left is the style you see shared most widely._

The one on the left is simple to read, reflects how many people in the USA view the politics of the United States, and is attractive. The one on the right, while containing more information, is harder to immediately understand, represents things by colors not typically used in US politics, and is less attractive. Is there any wonder that maps like the one on the left are the most shared on election night?

Your goal is to create objects like the map on the left; objects that allow a reader to understand a complex topic as easily as possible, to frame the data in a context they understand, and to create an object that they enjoy looking at and wish to share with others.

As for how to accomplish all of this? Well, that is what the rest of the wiki is all about!