Data Is Not Beautiful

Despite the rhetoric of recent years, data is neither beautiful nor ugly. Data is data; it merely describes what is and has no aesthetic dimension. The world that’s revealed in data can be breathtakingly beautiful or soul-crushingly ugly, but data itself is neither.

We can respond to data in ways that create beauty, justice, and wellbeing. We can do this, in part, both through data visualization and data art. Though data visualization and data art are constructed from the same raw materials (i.e., data), their methods differ. What does not differ, however, is their ultimate purpose to present or evoke meaning. When I visualize data, I do it to bring specific meanings to light or to make it possible for others to do that on their own. Similarly, when skilled data artists express data, they do it to evoke a meaningful experience. Even if the data artist’s meaning is less specific than mine as a data visualizer, the artist intends for the viewer to experience meaning and often emotion as well.

I appreciate good data art just as I appreciate good art of all types. What I cannot stomach is meaningless visual drivel that calls itself data art or, even worse, calls itself data visualization. I stridently object to the work of lazy, unskilled creators of meaningless, difficult to read, or misleading data displays. I’m referring to visualizations that fail to display data in ways that promote clear and true understanding. Many data visualizations that are labeled “beautiful” are anything but. Instead, they pander to the base interests of those who seek superficial, effortless pleasure rather than understanding, which always involves effort. There might be occasions when meaningless pleasure is useful, but not when data is being displayed. Data can potentially inform. We should never squander this potential.

Take care,


12 Comments on “Data Is Not Beautiful”

By Jeff. August 17th, 2017 at 10:20 am

There is indeed an irony that Stephen alludes to between claims of “beautiful data” on one hand and the “meaningless visual drivel” presented as visualizations produced from that data on the other hand. Why if the data are so beautiful do so many monstrosities emerge as visualizations based on that data??? Since data are not beautiful and data do not speak for themselves, it is up to us as thoughtful designers to deliver that beauty and meaning.

By Andrew. August 18th, 2017 at 8:32 am

“Data is data; it merely describes what is and has no aesthetic dimension.”

While this isn’t typically what’s argued by the usual “Data is Beautiful” crowd, I might suggest that the aesthetic dimension is how well data “describes what is”.

Data is therefore beautiful when it is precise, complete, relevant, honest, etc. Beautiful data lends itself to aimful exploration, detailed analysis, and effective visualization.

Data can also be ugly, such as when it lacks precision, has large gaps in relevant details, comes from unreliable sources, is previously manipulated beyond the point of usefulness, etc.

By Stephen Few. August 18th, 2017 at 8:47 am


I agree with what you’re saying except for one important clarification: when a display of data is precise, complete, relevant, honest, etc., it is the display that is beautiful, not the data.

By Andrew. August 18th, 2017 at 3:21 pm


Except I wasn’t speaking about a display of data, I was talking about the data itself. Of course when a data display is beautiful, it is the display that is beautiful.

What I’m suggesting is that even when not displayed, data might be appealing or unappealing, beautiful or ugly.

By Stephen Few. August 18th, 2017 at 4:10 pm


I stand corrected, except that I’ll challenge one of your points. Data is neither honest nor dishonest. Data has no intention. It can be true or false, but not honest or dishonest. Honesty only comes into play when people use data for their purposes.

By Andrew. August 21st, 2017 at 2:12 pm


The word “honest” isn’t limited in usage to subjects that can have intentions – it can mean that something is accurate or authentic or done in good faith. Data can be accurate, it can be authentic, it can be recorded in good faith. That is, it should be.

By Stephen Few. August 21st, 2017 at 2:47 pm


You’re stretching the definition of “honest” to a point that I can’t support other than in poetry.

By Andrew. August 21st, 2017 at 4:24 pm


I’m not stretching the definition at all, just using the word like anyone else.

“Honesty 1 a: free from fraud or deception : legitimate, truthful” – it’s literally the first entry from Merriam Webster (

Data can be recorded honestly or dishonestly. I’m sure you don’t disagree with that, so I’m not sure what point of mine you’re challenging.

By Stephen Few. August 21st, 2017 at 4:52 pm


“Free from fraud or deception” implies intention. Frauds and deceptions are committed by people, not by data. Regardless, we’re debating something that isn’t worthwhile. You and I share the same essential objective: to use and present data effectively. I’ll take the blame for this departure into trivial matters. I should have resisted the urge to point out your application of the term “honesty” to data, despite my perception that it is inappropriate.

By Daniel Smith. August 29th, 2017 at 2:28 pm

Thank you for fighting the good fight, Stephen! Always appreciate how you are trying to move the dataviz industry forward. Keep up the good work.


By Esther. September 25th, 2017 at 6:50 am

Hi Stephen,

Sorry for my cheeky curiosity, but…
Could you please share how many (technical/data visualization) books do you read on a monthly basis? And how many reflection time do you put through them?

I’m astonished by the body of knowledge you have accumulated in your library, and can’t help but wonder how the heck can you have so many good, useful things to say, all the time!…



By Stephen Few. September 25th, 2017 at 8:43 am

Hi Esther,

I spend much of my time reading, and have done so consistently since starting Perceptual Edge in 2003. Only a fraction of my book reading, however, involves books that were written about data visualization in particular. To date, the total library of worthwhile books written about data visualization numbers no more than about 25. Reading 25 books would not require a great deal of time. Most of my reading ventures into other areas that inform data visualization—especially brain science (perception and cognition). Additional areas include statistics, design, communication, critical thinking, systems theory, scientific method, ethics, and potentially related technologies (e.g., AI).

Reading alone, however, is not the key to learning. You can read everything worthwhile that’s been written and learn nothing. Thinking deeply about what you read and then putting it into practice over and over is required for learning. Thoughtful practice is essential. You can churn out data visualizations for years without developing expertise if you do so without thought. When I create data visualizations, I constantly evaluate their worth by stepping back and asking myself what works, what doesn’t, and how they might be improved. Given the public nature of my work, I also benefit from the thoughts of others who share their perspectives with me.

The best experts in any field are those who are the best students.