Coffeehouse Thread

8 posts

Forum Read Only

This forum has been made read only by the site admins. No new threads or comments can be added.

Mapquest Going Downhill

Back to Forum: Coffeehouse
  • User profile image
    Cornelius Ellsonpeter

    Or at least it seems like it lately. Recently, I put in a couple locations on Mapquest to get directions because I wanted to travel out of state. Fine. If I pull out a regular old highway map (they still make those, ya know), it looks very straightforward to me that I only need to take 2 or three highways to get to my destination. If I run it through Mapquest, it gives me a huge list of directions that involves 26 seperate turns and about fourteen different roads.

    And that's the "shortest distance" option!

    I've already had experiences where Mapquest has told me to take roads that do not exist (one of them ended at a river, but Mapquest insisted I could drive through!), or where it fails to have updated roads. It just seems like lately they are tweaking their algorithms or something, and it's taking a turn for the worse.

    I also tried using the "avoid highways" option...it gave me fewer roads to travel on, yet the roads included state and U.S. Highway routes. It apparently only avoided interstates. Oddly, it's closer to one of the routes I was thinking of taking.

    Perplexed

  • User profile image
    DoomBringer

    Thing is, by looking at the map, you only see the roads, and not the details of getting on them.  Mapquest will tell you ever single detail, so you won't get lost finding how to get on that road.

  • User profile image
    Cornelius Ellsonpeter

    Mapquest Geocoding

    I thought I also read somewhere that in some cases, they do actually drive some of these routes, and note several things such as stoplights, number of turns, etc. It's just that when you move outside of large cities (as a starting or ending point), the data must be a lot more incomplete. I've seen it though, where it will have me go through a suburb, taking several turns, for no apparent reason. In that particular case, it did not turn out to be the shorter route, and instead took longer because I had to go through more traffic and stoplights.

  • User profile image
    Antitorgo

    Are you sure that the directions MapQuest provided isn't the shortest distance?

    "shortest distance" will actually try pretty hard to give you the shortest distance route, and they rarely involve highways...

    I'm curious as to what route you plugged in, care to share them after you've genericized them and took out the details?

  • User profile image
    Rossj

    DoomBringer wrote:
    Thing is, by looking at the map, you only see the roads, and not the details of getting on them. 


    Well given time I suspect Microsoft Senseweb could solve this problem, well time and a couple million sensors. Basically (in theory) it will give you a real-time view of the world from the current weather patterns on your route to the density of traffic.


  • User profile image
    Cornelius Ellsonpeter

    Antitorgo wrote:
    Are you sure that the directions MapQuest provided isn't the shortest distance?

    "shortest distance" will actually try pretty hard to give you the shortest distance route, and they rarely involve highways...

    I'm curious as to what route you plugged in, care to share them after you've genericized them and took out the details?
    Here. I don't live anywhere near these towns, but I picked them more or less at random to help get the point across. From my understanding these towns are only about four miles apart on the same major highway. The shortest time algorithm is whacked if you do the math:

    Point A to point C using shortest time algorithm:

    25 directions, Total Est. Time: 4 hours, 17 minutes; Total Est. Distance: 239.44 miles

    Now drive to a town about 3-4 miles over and you get this...

    Point B to Point C using shortest time algorithm:

    12 directions, Total Est. Time: 4 hours, 19 minutes; Total Est. Distance: 285.99 miles

    Yes folks, that's half the directions in the second example, 46.55 more miles but only two minutes more drive time!

    Perplexed

    Now let's try the shortest distance algorithm.

    Point A to point C using shortest distance algorithm:

    27 directions, Total Est. Time: 5 hours, 30 minutes; Total Est. Distance: 218.37 miles

    Point B to Point C using shortest distance algorithm:

    26 directions, Total Est. Time: 5 hours, 28 minutes;Total Est. Distance: 221.33 miles

    The distance ones in this case make a bit more sense, although you'd spend an extra hour on the road (and only save about 18-20 miles). Perhaps they should offer several choices of which route you'd like to take instead of assuming they always know the best way?

    [A]

  • User profile image
    Antitorgo

    Cornelius Ellsonpeter wrote:
    

    Here. I don't live anywhere near these towns, but I picked them more or less at random to help get the point across. From my understanding these towns are only about four miles apart on the same major highway. The shortest time algorithm is whacked if you do the math:



    I take it you meant four hours apart right?

    Cornelius Ellsonpeter wrote:
    
    The distance ones in this case make a bit more sense, although you'd spend an extra hour on the road (and only save about 18-20 miles). Perhaps they should offer several choices of which route you'd like to take instead of assuming they always know the best way?



    It's been a while since I worked for a company that did this stuff (back in 2000). From what I remember of it, the "weight" given to roads has a fairly large influence on the "shortest time" route. My guess in this case would be that Hwy 12 in this case has a lower weight than the major US hwys thus the slightly longer trip on it caused the algorithm to opt against going all the way to I-35.

    In any case, I've seen the algorithms get really messed up sometimes (like taking the most bizarre roundabout routes) just because of how the math of the graph worked out. So I wouldn't be too surprised.

    End result is was the Google and Yahoo both chose the "non-US highway" route for both starting points... So overall MapQuest didn't do half-bad in my book.

    I'd definitely like to see an "easiest directions" option though.

  • User profile image
    Xaero_​Vincent

    Buying a decent car GPS system will demonstrate just how poor Mapquest is.

    A good GPS system will offer better accuracy and lead you to your destination with the fastest or closest trip distance and tell exactly you where to turn and stop. You wont travel out of your way nor fiddle with printouts like with Mapquest.

    However like all good things there is a price to pay. A good GPS system could set you back $400+. Our GPS system costed $500 and will store and play MP3s on its mini 20 GB hard drive.


    Regards,
    Vincent

Conversation locked

This conversation has been locked by the site admins. No new comments can be made.