A Tool's Got to Know Its Limitations

Written by: N.S. Palmer | Posted on: | Category:

What do DNA, psychological questionnaires, and my new car have in common?

DNA (deoxyribonucleic acid) is a molecule in which plants and animals encode the basic characteristics of their bodies. It's what makes a rose a rose instead of a carrot. It makes some people tall and other people short. It affects almost every trait of living things -- even people's psychological traits.

As a result, we can now use DNA to make predictions about people. Because environment also affects how people turn out, the predictions can only give probabilities. But someone with a high "polygenic score" for height is more likely to be tall than someone with a low score.

The trouble with shiny new tools is that we tend to over-estimate what they tell us. For example, in his book Blueprint: How DNA Makes Us Who We Are, geneticist Robert Plomin reveals what his score for obesity told him:

"Knowing my BMI polygenic score helps me realize that I can’t let my guard down, because it is in those weak moments – for example, when I am tired after a long day – that I sometimes give in to those siren snacks in the cupboard whispering to me."

That's good to know. But did he not already know that? Of course he knew it. His polygenic score didn't tell him anything new about his behavior. It only provided a partial explanation for it.

Polygenic scores aren't nothing, but they're not everything, either. They're one piece of a puzzle.

The same is true of psychological questionnaires such as the Myers-Briggs test. They were a fad in corporate HR departments when they were first introduced. A lot of large companies still use them, and you can see employees' Myers-Briggs personality diagrams posted near their desks.

The theory is that everyone has personality traits (well, duh!) that make them better for some jobs than for others (also a "well, duh!"). Myers-Briggs and similar tests try to measure those traits.

But there are two problems. First, anyone who's smart enough to hold a job can easily learn to game the tests. If you know how to answer the questions, you can get whatever result you want.

Second, and more relevant, the tests rarely tell you anything about people that you couldn't discern just by talking to them for 15 minutes. I'm fairly obtuse about people, but even I can tell you more after a short conversation than by having you take a Myers-Briggs test. Sure, you could deceive me if you knew how, but the same problem applies to the test.

And then there's my new car. Did I mention that I have a new car? It's quite a nice toy. It's got so many features that I still don't know what half of them do.

But it definitely has cameras. Front, rear, visual, radar, probably even sonar and laser sensors -- it wouldn't surprise me a bit. "Laser," by the way, is an acronym for Light Amplification through Stimulated Emission of Radiation.

But I digress. I complained to one of my brothers that the cameras were neat but that they didn't add much. While driving, I still have to watch the area around the car in exactly the same way as I would without the cameras.

And then I realized: the cameras are like DNA and psychological tests. They simply give you more data about what you already knew. Most of the time, they might not add much to your understanding -- though in a few cases, they might. You just can't rely on them by themselves. A tool's got to know its limitations.

I like my new car, anyway. I think I'll keep it.