Tuesday, August 27, 2019

Yes, charts can and do lie

I've started receiving some mild, reasonable, and predictable pushback to the title of How Charts Lie (example), so allow me to explain my rationale to choose it.

First, some friends have complained that the title sounds too negative, and that it may create a bad impression of data visualization among the public at large. Well, charts can indeed lie. We shouldn't be afraid of saying so. After doing that in How Charts Lie I adopt a cheery and positive tone, as the book is really a manual about becoming better chart readers. It also praises the power of good visualizations.

Another objection I've seen is to claim that “charts don't lie, people lie with charts”. This is true under one strict definition of what lying is: telling someone something you know is not true with the intention to deceive. Lying requires a deliberate decision by a conscious actor (a chart isn't conscious) to convey something untruthful. By this definition, lying is a bit different than deceiving—you can deceive without lying,*—misleading, or bullshitting. More about this later.

My first response is that “charts don't lie, people lie with charts” sounds similar to “guns don't kill people, people kill other people with guns,” a common contention in gun policy debates. I find little merit in this argument. If you're wounded by a bullet shot by someone, it's appropriate to say that the bullet wounded you.

Moreover, the bullet—and the gun—is an artifact designed with the specific goal of hurting. The argument above reminds me of those who claim that artifacts and technology are neutral, and that they only become good or bad when someone uses them. Those arguments and assumptions have been discussed and challenged for decades by philosophers of technology. There's also a growing critical literature about digital tech that you may want to consult: Cathy O'Neil, Meredith Broussard, Virginia Eubanks, or Safiya Noble. I'm not saying you're wrong, just that you shouldn't assume that you can take artifact neutrality for granted without thorough inspection.

Back to the strict definition of lying. Think about cases in which you called a statement a lie. I'm sure you've done so more than once. We all have. Did you read the mind of the actor or actors who made the statement? Or did they explicitly disclose their intentions? Or did you have solid evidence that (a) those actors knew that their statement was false and that (b) they were consciously trying to conceal what they knew to be true with the intention to deceive? Because that's the standard required to call something a lie if you're too nitpicky with words. This standard is what led The New York Times to be (unnecessarily) cautious about calling anything a lie for too long.

How Charts Lie contains examples that I think qualify as lies in this sense, but I cannot tell for certain because liars are usually unwilling to disclose their goals. I'm happy calling those charts lies regardless.

Many other cases in the book are not lies, but instances in which we “lie to ourselves” with a chart. This is the informal way of saying that we often use perfectly fine evidence such as charts in a twisted or inappropriate manner, and we end up extracting faulty inferences or seeing what we want to see. “Lying to ourselves” isn't an entirely accurate term for such a situation, but I believe that everyone understands what it means. Strictly speaking, you can't “lie to yourself without knowing it”, as this article claims. That's a contradiction in terms, but common language isn't as rigorous as philosophical discourse, so I think that the phrase is fine.

I admit that to be 100% accurate I shouldn't have titled the book How Charts Lie, but How Charts Lie, Deceive, and Mislead, and How We Lie with Them, Misread Them, and Use Them to Confirm What We Want to Believe. But the goal of the book isn't to please you, language nerds. It's to attract audiences who don't know much about visualization, get them excited but also a bit skeptical about it, and then show them how it's supposed to work.

Therefore, I'll go with the concise, provocative, and punchy title I chose following Hans Rosling's advice: “You have to be like the worst tabloid newspaper in the front and the Academy of Science in the back.”

*For instance, if I told you something that is true while knowing you're going to get it wrong I wouldn't be telling a lie, but I'd be deceiving you.