A couple of weeks ago, my wife and I were heading across the San Rafael Bridge to downtown Oakland for a show at the Fox Theatre. As all Bay Area drivers know, there’s a historically awful stretch of Interstate 80 along that route – a permanent traffic mess.

I considered taking San Pablo Ave., a major thoroughfare that parallels the freeway. But my wife fired up Waze, and we followed an intricate set of instructions that took us onto frontage roads, side streets, and counter-intuitive detours. Despite our shared unease (unfamiliar streets through some blighted neighborhoods), we trusted the Waze algorithms – and we weren’t alone. A continuous stream of automobiles snaked along the very same improbable route – and inside the cars ahead and behind me, I saw glowing blue screens delivering similar instructions to the drivers within.

About a year or so ago I started regularly using Waze. I started using it on familiar routes: to and from work, going to the ballpark, maneuvering across San Francisco for a meeting. Prior to that, I only used Waze as an occasional replacement for Google Maps – when I wasn’t sure how to get from point A to point B.

Of course, Waze is a revelation for the uninitiated. It essentially turns your car into an autonomous vehicle, with you as a simple robot executing the commands of an extraordinarily sophisticated and crowdsourced AI.

But, as I’m sure you’ve noticed if you’re a regular “Wazer,” the app is driving a tangible “flocking” behavior in a significant percentage of drivers on the road. Waze has built a real-time layer of data and commands over our current traffic infrastructure. This new layer is owned and operated by a for-profit company (Google, which owns Waze), its algorithms necessarily protected as intellectual property. And because it’s so much better than what we had before, nearly everyone is thrilled with the deal (there are some upset homeowners tired of those new traffic flows).

Since the rise of the automobile, we’ve managed traffic flows through a public commons – a slow-moving but accountable ecosystem of local and national ordinances (speed limits, stop signs, traffic lights, etc) that were more or less consistent across all publicly owned roadways.

Information-first tech platforms like Waze, Uber, and Airbnb deliver innovative solutions to real-world problems that were impossible for governments to address (or even imagine).

At what point will Waze or something like it integrate with the traffic grid and start to control the lights?

I’ve written before about how we’re slowly replacing our public commons with corporate, for-profit solutions, but I sense a quickening afoot. There’s an inevitable collision between the public’s right to know and a corporation’s need for profit. How exactly do these algorithms choose how best to guide us around? Is it fair to route traffic past people’s homes and/or away from roadside businesses? Should we just throw up our hands and “trust the tech?”

We’ve already been practicing solutions to these questions, first with the Web, then with Google search and the Facebook news feed, and now with Waze. But absent a more robust dialog addressing these issues, we run a real risk of creating a new kind of regulatory capture – not in the classic sense, where corrupt public officials preference one company over another. Rather, it’s a more private kind, in which a for-profit corporation becomes the regulatory framework itself – not through malicious intent or greed, but simply by offering a better way.

Want to follow the biggest story in business? Get our NewCo Daily newsletter.


  • And then there’s the “What Waze doesn’t know” effect: I was in Sao Paolo last year, and an older couple was shot dead after Waze directed them through a favela – one of the slums there where their black sedan was mistaken for a rival gang’s and shot up something like 100 times.
    The people who live there know Waze isn’t to be trusted – that many of these so called optimal routes can only be somewhat safely used in the daytime, where you might only be stuck up at gunpoint rather than killed.
    The beauty of “AI”.

  • At what point are there enough people using Waze that you’re better off *not* listening to its advice. For example, in the bad old days, I’d listen to radio traffic reports. The first time a delay was announced, I’d go the alternate route. However, by the third or fourth time I’d hear the warning, (1) the accident had already been cleared and (2) sufficient people had diverted so as to make the original route actually be the preferred route. If one person responds to Waze, they get the benefit. If everyone responds, the alternate quickly becomes the un-preferred route. What you really need is intelligence at the head end which optimizes traffic flow, sending some people to the alternate, sending others to another alternate and the remainder to stay the course.

Leave a Reply

Your email address will not be published. Required fields are marked *