Google Maps, traffic, and estimating time

I find the subject of how people use time fascinating — and particularly the sub-topic of how people estimate time. The gist is that it’s more complicated than it seems.

On Tuesday, I drove from my home outside Philadelphia to Tyson’s Corner, VA to give a lunch time speech. This is a 153 mile trip, and in perfect conditions, should take 2 hours and 33 minutes (per Google Maps). However, on a route that includes I-476, I-95 and I-495 (the Beltway around Washington D.C.) perfect conditions occur at roughly 3 a.m. And maybe not even then, because there’s a lot of roadwork going on that’s often done overnight.

Traffic is not an unknowable variable. Google Maps can tell you the traffic delays at different points in your route. However, traffic changes over time, which means that on a long drive, accuracy becomes a tricky thing. The answer Google gives you when you ask the distance between two points is to the question “What are the total traffic delays over my route?” It is not the answer to the question “How long will it take me to get between these two points?”  

I was reminded of this yesterday morning when I typed in my two points at 8:30 in the morning, and had Google tell me that my route would take 4 hours and 32 minutes with traffic. With that level of delay, I would miss my speech! But I soon realized this calculation was assuming I’d be facing 8:30 a.m. rush hour traffic in Philadelphia, Baltimore, and Washington DC simultaneously. Rush hour doesn’t last that long, thank goodness, and in fact the trip took 3 hours and 10 minutes, including a bathroom stop. I experienced 35 minutes of Philadelphia-area traffic delays, then basically didn’t need to use my brakes until I pulled off the highway in VA. The additional 85 minutes of traffic delays were gone by the time I got to the problem.

It would be a much more complicated algorithm to tell someone that if you leave right now, the traffic 100 miles away from you has a 90 percent chance of dissipating by the time you get there. My in-car Garmin “solves” this problem by, as far as I can tell, not even telling you about traffic delays that are more than an hour ahead of you. (That’s just my guess based on when I learned about different delays along the route, but I’d be interested in their actual algorithm). This isn’t a perfect system either, as sometimes you’d make different decisions in major metropolitan areas based on anticipated traffic. If you know, with a 90 percent probability, that there will be delays of more than 20 minutes on a certain bridge 1 hour in the future, you might choose a different route from the get go. But for the most part, this aspect of estimation still seems to require human skill.

In other news: I wrote a post for Fast Company a few weeks ago on How to Manufacture More Time in Your Day that covers the skill of time estimation.



6 Responses to Google Maps, traffic, and estimating time


  1. Nother Barb says:

    I need to test whether Google maps can tell me whether I’m going to catch one of the freight trains that might cross my path, on two separate tracks, how long the trains are, and how slowly they are moving, and whether taking my detour for one of them will help my time or not. On a 150-mile trip it doesn’t make a big difference, but on a 15-minute trip, it means all the difference.

  2. oldmdgirl says:

    This post is so apropos! I just drove up to my parent’s house, and as they are in NW CT, Google maps tells me that I should take the NJ turnpike and go through NYC. However any idiot knows this is foolish. Annoyingly, we’ve found the fastest alternative route is to take 206 through Princeton to get on 287 to go around the city, but the traffic in that area is so incredibly annoying that I was contemplating yesterday whether it would have been worth going through Allentown and taking 476 instead. Google maps says this is 25 minutes longer than the Princeton route, but it is all highway driving and has no annoying lights. Maybe next time?

    And I agree with Barb having freight trains built into the algorithm would be very useful.

    • Laura says:

      @oldmdgirl – you should give it a try. Any route that goes over the GWB is just fraught with problems from the get go. 206 through Princeton slows you a lot. Unfortunately, some of the ways we go north from Philly have a lot of lights too — driving through Doylestown to get to points north…

  3. oilandgarlic says:

    And oftentimes you have to add time for finding parking, too!

    • Laura says:

      @oilandgarlic – this, alone, is why I am late to everything in downtown Philly. Getting there is fine. Parking is a whole different matter.

  4. Katy says:

    You might want to try the Waze app -the company has been recently purchased by Google and the app is VERY good at predicting travel time as it takes into account other people (running the app ) travel time in real time

    https://www.waze.com/about

    (I’m in no way associated, just a very happy user :) )