Technology moves in mysterious Waze, its wonders to perform…
Los Angeles, a city I lived in for four years, has a traffic problem. So when Waze, a GPS app that allows users to share information about traffic in real time, Angelinos were enthusiastic adopters.
Waze was launched in Los Angeles in 2011 through a media for equity agreement at the height of the so-called Carmageddon, the result of maintenance work on Interstate 405, one of the city’s main arteries and that threatened to bring traffic to a halt (in the end, drivers took so many precautions, that on some of those days, traffic flow was better than usual). Waze, which was eventually bought by Google, became hugely popular as a way of avoiding gridlock.
Many of the thousands of people who use Waze do so not to find out how to get somewhere they don’t know, but to find out the quickest route to work or back home. In a city like Los Angeles, which is laid out in a grid pattern, it isn’t hard to alternative routes to Wilshire Blvd or Sunset Blvd. The outcome is that normally-peaceful streets in residential areas are suddenly filled with drivers trying to avoid traffic jams and creating new ones instead.
Faced with this noisy, smelly takeover, and sometimes unable even to enter or leave their own driveways, residents have taken the offensive, signing up to Waze and reporting accidents, speed checks and roadworks in the hope it will put people off using their street.
In response, Waze has begun monitoring users who report a lot of incidents, checking them, and if they turn out to be false, closing their account. Which prompts the question as to how easy it is to trick an app in this way. The answer is that it’s not easy. One way is for local residents and friends to agree to report problems, but if another driver passing by happens to say that everything is fine, then the company will likely cancel their accounts. Of course anybody can always sign up again under a false name, and the whole process starts up once more.
As is increasingly the case, this is about developing a sufficiently sophisticated casuistic so as to be able to train a machine learning algorithm: it’s possible to trick a machine a number of times, but not systematically over time, because its learning processes are more consistent and generate rules about patterns of all types. There are hackers able to create patterns to simulate a traffic jam by sending updates that look as though they come from multiple users simply to show that it can be done, but even these kinds of patterns are easy to detect if they become repetitive, and let’s face it, most people have better things to do.
Events in Los Angeles illustrate the many paradoxes that technology creates: a GPS provides the best route, but it generates less-used routes, which can make other people’s lives a misery, or even create new traffic jams when too many people receive the same information. Waze is typical of the dynamic systems we’re going to be seeing more of and that present challenges in terms of establishing a happy medium. Right now, we’re still in the smartphone and app phase of Waze and Google Maps, but more and more new vehicles will have them built in. Of the 2.1 million SIM cards that AT&T have activated over the first quarter of 2015, more than half were for cars. We’re in the midst of a revolution, the consequences of which we’re not always aware, until we find a traffic jam outside our front door!
(En español, aquí)
