In June 2017, a blind pedestrian named Sabriye Tenberken was using a popular navigation app to walk from her hotel in Nairobi to a conference venue twelve minutes away. The app told her to turn left. She turned left into a construction trench.
In June 2017, a tech reviewer at a major outlet compared the same app against its closest rival. He tested commute times, traffic rerouting, fuel cost estimates, satellite imagery. He declared a winner. He never left his car.
I read these comparison reviews the way some people read restaurant guides written by someone who has never been hungry. The vocabulary is fluent. The methodology is clean. The conclusion is confident. And it has nothing to do with me.
Google Maps versus Waze. Waze versus Apple Maps. The reviewers test speed, accuracy, interface, voice prompts. They compare arrival times across identical routes. They measure how quickly each app reroutes around a traffic jam. They never once test whether the suggested route includes a curb cut. They never ask whether the pedestrian directions send you down a street with no sidewalk. They never check if the cycling route assumes you can hear a truck approaching from behind.
You might say: these apps weren’t built for that. Exactly. That is the argument.
Mike Oliver drew a line in 1983 between impairment and disability. Impairment lives in my body. Disability lives in the built world. A T6 spinal cord injury is an impairment. A navigation app that routes me down a cobblestone street with a 15-percent grade and calls it “accessible” is disability. The distinction matters because it tells you where to aim the fix. Not at my spine. At the software.
I tried this myself. February 2023, central Rotterdam. I toggled on the “wheelchair accessible” layer in Google Maps — a feature the company launched with some fanfare in select cities. The app routed me to a tram stop. The tram stop had been under renovation for four months. No temporary ramp. No alternative. The app didn’t know. The app doesn’t update for temporary barriers because temporary barriers are not part of the data architecture. Permanent infrastructure gets mapped. The sandwich board, the construction fence, the delivery truck parked across the curb cut — these don’t exist in the model. They exist on the ground, which is where I am.
Sunaura Taylor wrote about how the category “normal” functions as an unmarked default, an invisible standard that only becomes visible when a body deviates from it. Navigation apps are normal in this exact sense. They assume a body that can see the screen, hear the audio, climb stairs, cross unpaved ground, react in real time to a reroute. Every review that compares two apps without naming these assumptions is reviewing a product for one kind of body and calling it universal.
The economics make this worse. Waze is crowd-sourced. Its data comes from drivers reporting conditions. Wheelchair users, blind pedestrians, Deaf cyclists — none of these populations generate data at the scale the algorithm needs. The app improves for the people who already use it. It gets worse, or stays static, for everyone else. This is not a bug. It is the funding model. Crowd-sourced platforms extract value from the crowd that shows up. The crowd that can’t show up — because the platform wasn’t built for them — never enters the feedback loop.
I know what this costs because I’ve sat in rooms where people price it. In October 2022, I attended an urban mobility workshop in Brussels where a project lead from a European mapping consortium admitted, on the record, that adding real-time accessibility data to their platform would cost roughly twelve percent more than their current budget. Twelve percent. He said, plainly: “The market isn’t there.” He meant: not enough wheelchair users generate enough trips to justify the spend. He was honest. I appreciated the honesty. I did not appreciate the math.
Solnit wrote beautifully about walking as a political act, as a way of claiming public space. I read her Wanderlust in 2009 and loved it and felt the familiar ache of reading someone who assumes the body she inhabits is the body everyone has. Walking is political. Rolling is political in a different register — it exposes infrastructure that walkers never notice. Every crack, every gradient, every missing curb cut. My chair is a diagnostic instrument. It reads the city the way a doctor reads a scan. The navigation app cannot do this because it was never asked to.
The review that inspired this piece tested fuel estimates. It compared how accurately each app predicted gas costs for a cross-country drive. The reviewer spent three paragraphs on this. He did not spend one sentence on whether either app could route a person in a wheelchair from a train station to a hospital entrance without encountering stairs. That silence is the review. Not what he measured. What he didn’t think to measure.
I don’t want better apps. I want the reviewer to say, in the first paragraph: “I tested this as a sighted, hearing, ambulatory driver, and my conclusions apply only to people like me.” That sentence would cost nothing. It would change what the reader understands they are reading. It would make the unmarked body marked.
Tenberken, who is also the co-founder of Braille Without Borders, found the conference venue eventually. Someone on the street walked her there. The app is still rated 4.6 stars.
This article was inspired by Google Maps vs. Waze: I compared the two best navigation apps, and this one’s better - ZDNET from news.google.com.