Waze, the “…world’s largest community-based traffic and navigation app” failed its users in the Santa Cruz, California area during the month of February 2017. These users who depend on Waze to find out traffic conditions and alternate routes were not able to do so. The same was true for Google Maps. For example, when traffic stalled for up to three hours, Waze and Google Maps happily reported that conditions were just fine.
Many Santa Cruz, CA residents commute from the beach community, and ever farther, to the Silicon Valley. They commute over highway 17, a four-lane twisty thoroughfare (two lanes in each direction), built over a mountain range. Normal drive times span between 30 and 45 minutes each way. Once your vehicle enters this highway, you are committed; you cannot turn around until you reach the other side of the mountain.
With the heavy rain storms, mud slides, and road closures that began in February 2017, traffic ground to a halt and commute times skyrocketed to multiple hours.
Many commuters depend on live traffic alerts from websites like sigalert.com, cruz511.org or www.dot.ca.gov, and twitter feeds like @TrafficOn17.
In other words, the 20,000 daily commuters1 on Highway 17 rely on the dependability and accuracy of Waze and Google Maps.
But the apps failed.
Thousands missed their job interviews, medical procedures, child care pickups, and so on.
At IWL our motto is “Saving the World From Network Failure.” Many Santa Cruz residents turned to us and asked us to make good on our motto! They asked us to explain what happened; they assumed that the problem was some type of network outage or failure. So we looked into the problem.
So, why did Waze and Google Maps report that road conditions were just fine, when they so very clearly were not?
Waze and Google Maps have been mysteriously silent on the subject. There have been no explanations, no apologies, no plan for correcting the problems that are obviously there. They have not even acknowledged that there was a problem!
Debunking the Myths …
There’s been much speculation about the cause of the failures.
Was spotty cell phone coverage to blame? No! Most apps on iPhones and Androids were working just fine on highway 17 during the mud slides and road closures. Cellular connectivity did not appear to be an issue. While some spots on highway 17 are cellular dead zones, there was plenty of coverage in the aggregate.
Was there insufficient data on traffic flow? No! At any point in time, the cell phone carriers (Verizon, T-Mobile, etc.), know exactly where your car is on highway 17 and they can compute your transit time from point A to B.
One could calculate commute time based on either the car or the user’s phone GPS location; where was the car 20 minutes ago and where is it now?
Waze and Google Maps got this wrong; they reported typical commute times, not the actual commute times.
Were alternate routes suggested to these users? Sometimes. But if a driver believes that the commute time is normal, the driver would not pursue an alternate route, which would be very long on a highway 17 commute.
Waze / Google Maps Expectations from Users
The Waze website suggests that Waze is merely a fun, social media app for drivers to connect and engage with each other. However, Waze users do not see it that way. Users perceive Waze as the source of mission critical information required to make key planning decisions for appointments, childcare, and a host of other obligations. Google Maps, on the other hand suggests that you “…navigate the roads like a pro. Google Maps dynamically re-routes based on real-time traffic information, and even suggests which lane to be in.”
If Waze / Google Maps continue on this trajectory, then perhaps it is time for public agencies to get involved to create an alternative solution consistent with the public’s expectations for critical road condition information. Or perhaps we need more competition with better designed and better tested applications.
Our Questions for Waze / Google Maps
- What is the underlying theory of operation of your systems? We had thought that you must incorporate machine learning and probabilistic reasoning systems to come up with alternative routes, but perhaps not. Perhaps you rely on crowd sourcing and that failed.
- Good computer science / computer networking practice is to “back off” in a failure condition and either send a message like “Waze not available” or “Estimated Google Driving Time not available”. Why did the apps continue to report best case commuting time?
As climate change continues unabated, we can expect to see more degraded traffic conditions like the landslides that occurred on Highway 17 in February. Commuters must be informed to make intelligent driving choices. Waze / Google Maps must provide greater accountability for their systems, particularly if they plan to expand to ride sharing applications. Clearly more competition is required in the navigation app space.