Design and Google Maps
A few days after reading Irene Au’s piece on design, I found myself reconsidering it as I drove through the streets of Los Angeles, feeling somewhat frustrated. I wasn’t stuck in the city’s legendary traffic. I was actually moving steadily toward my destination thanks to Google Maps on my phone and its GPS navigation. These are amazingly powerful tools, and I’m thankful to have them as I learn a new city. But following the real-time directions, I kept encountering steps that felt like manifestations of a design flaw. If as Au says, “what we create reflects our inner state,” then Maps must think it was designed by and is now giving directions to robots.
I am not an engineer. I have no understanding of the formulas and algorithms that Maps uses to find the most efficient route. I speak math conversationally, but I’m not fluent, and I certainly don’t think mathematically unless I have to. Which is why I find it so odd that Google Maps frequently gives directions as if they were inputs for another computer to process. Most obvious is this frequent step: “In 1000 feet, turn right.” I have no sense of what that means. Three blocks? Five? At different speeds, 1000 feet feels very different, and without the street name to clarify, I’m inevitably taking my eyes off the road to figure out where I need to turn. If I had a human passenger reading me the directions instead, he or she would never tell me to turn in x feet. More likely, “turn in three blocks” or “a right at the next light.” I want Maps to measure distances like a human, not like a machine.
Changing the “language” of the spoken instructions might not be that difficult, but I think this design problem goes deeper into the code. Maps uses live traffic data to find the fastest route, which sometimes means deviating from the most direct route. Avoiding a traffic jam by taking an entirely different highway can save several minutes, but I don’t think the optimization is worthwhile for city driving. I often find that I’m being directed through narrow side streets, just to avoid a stoplight that Maps has confused for traffic. Unless you’re saving me 5 minutes or more, I don’t think a detour is helpful, especially since I’m probably using Maps because I’m unfamiliar with the streets. An example:
Once again, a human passenger would have given different directions, this time choosing simplicity over speed. But for a computer, there’s no difference — turn when instructed. Is there a way to design Maps so that it strikes a reasonable balance between finding the fastest route and the least complex?
I don’t know if there’s any data relating safety to all of this, but I definitely feel more nervous when the directions seem too roundabout. I end up paying more attention to my phone or the street signs, desperately trying to avoid a wrong turn. The other cars on the road tend to fade into the background. And while safety should probably be the most important consideration in Maps’ design, it may also be the most difficult to program. Here’s a very specific situation that I found myself in, once again frustrated that simple instructions for a computer were actually very difficult for a human to follow:
Coming down another side street, I approached a stop sign. I was instructed to make a right, followed by a quick left onto another side street. Simple enough. But in between was a major road, two lanes of fairly consistent, fast-moving traffic that I needed to cross to pull into the left turn lane. I could see the side street I was supposed to turn onto. It was an incredibly quick turn, just over two car lengths along the busy road, which meant I couldn’t get into the turn lane while there were other cars waiting. So I’m sitting at this stop sign for a long time, waiting for there to be no cars on the two lanes that were going through the intersection and for there to be no cars in the left turn lane. I eventually made it, but this was a complicated maneuver that could have been unsafe had I turned right too early and blocked one or both lanes of oncoming traffic. This entire scenario could have been avoided by using a side street with a stoplight or by entering the busy road a few blocks down to give me time to change lanes. Is it possible to design Maps so that it doesn’t put drivers in risky scenarios like this one?
I’m not sure whether this design problem is the result of “fear, greed, attachment, [or] ego.” More likely, it’s just the best we can do with the technology we have. But I hope that Google takes Au’s ideas about design and the Self into consideration as it improves upon Maps. For now, it’s hard to imagine the kind of person that Maps would be if it were a passenger in your car instead of an app on your phone. To me, it’s still too robotic to be human. Using it feels more like obeying commands than following directions.
Maybe the discrepancy lies in how the data is viewed. I think Maps sees the world from above, a bright grid of yellow and white roads against a city greyed into irrelevance. But people only get to see what is visible through their windshields, and the buildings tend to get in the way. It’s amazing that Maps can give us the bird’s eye view and use it to calculate the best route. But it’s frustrating when it can’t communicate that knowledge to human drivers who are physically incapable of seeing the entire picture. We need a design that speaks our language and acknowledges our limitations. As often as possible, it should let us focus on the road we’re on now and not the road we need to turn onto next. Mostly, it shouldn’t forget that we’re human.