Risk wheels within risk wheels

An excellent article that highlights the importance of argument framing. A remarkably small idea that could reduce distracted driving by Emily Badger. The issue is that there are inherent dangers to distracted driving, that cars are becoming more complex and therefore create greater distraction. The suggested solution is to use design, and specifically font design, to reduce the degree of distraction in the car.

There are at least two ways to think about the problem of distracted driving. We could try to get people to cut down on all of the stuff that's distracting them -- texting, fielding phone calls, fiddling with in-car navigation screens at 50 miles an hour. Or we could acknowledge that drivers will probably keep doing all of those things anyway and try to mitigate the harm.

[snip]

Two years ago, the MIT AgeLab and Monotype began to study whether more legible typefaces could make a difference in in-car media. For men, at least, the answer has been yes. In driving simulations run by the lab, male drivers took their eyes off the road for less time when the text on a small navigation screen appeared in a typeface from what's known as the humanist genre. The difference between humanist and grotesque typefaces amounted to the equivalent of turning away from the road over a distance of 50 feet at highway speeds.

But the issue is more complicated than the straight-forward framing above. The additional concerns are raised in the comments section.

People based systems are complex with uncertainty surrounding incentives, communication, feedback mechanisms, inconsistent objectives, cognitive biases which distort the gap between stated intentions and actual actions, etc. The article misses these but they are pointed out in the comments.

First - The heart of economics is incentives. People respond to incentives (and disincentives). If you make something faster, easier, cheaper, or more rewarding, ceteris paribus, you increase demand for it. So if you make it easier and faster to comprehend text on car displays, even though you are decreasing the amount of time per instance of looking, you may also be increasing the number of instances. Such is the complexity of the human system.

But what this also highlights is the human tendency to polarize things into a dichotomy. Some of us want rules that are enforced in support of a shared fixed goal. That leads to a desire for clear rules and clear punishments that attach to breaking those rules. Others are more concerned about the outcome than the process. If people are going to text regardless of what we do, then let's make it safer to text is the logic.

There are obviously circumstances where both perspectives are useful. The challenge in this case is to determine whether it is more likely that making it safer to read and text (therefore making it more likely that people will do so) outweighs the per instance benefit by making clearer type. In other words, which is the greater number, deaths and injuries from more frequent distractions or deaths and injuries from longer duration distractions? The important thing is to anticipate that the process does have feedback loops and people do change their behaviors based on incentives. Without further data, there is no way to tell whether this is potentially a good approach.

What we do know is that it is a bad approach not to take into account the consequences of anticipated change on the human incentive structure.