Exploring Local
Mike Dobson of TeleMapics on Local Search and All Things Geospatial

Map Accuracy, Google, NAVTEQ and “Free”

February 11th, 2010 by MDob

In one of my last columns, a reader raised the issue of the real “value” of free data, suggesting that with Google’s free map data that you get what you pay for. Since that comment, Nokia decided to enter the “free” map market by provisioning a subset of its higher-end phones with NAVTEQ Maps (Nokia owns NAVTEQ).

I think there are at least two strategies at work here. First, Nokia is attempting to protect its market position as a handset provider in the United States and Europe by offering a service competitive or superior to Google in terms of the quality of its map data. The second part of the strategy may be an attempt to expand its market dominance in areas where Google currently does not have the navigable map database coverage that NAVTEQ can provide Nokia. On the other hand, Google, through inputs from Google Map Maker, can provide navigation for pedestrians in more locations than those in which NAVTEQ has car-navigation databases and this may allow Google to be more successful in countries where cars are not the principal means of transportation.

However, as we have postulated before the key differentiator in many geographical markets may not be the map data or the navigation services themselves, but the supporting role they play in local search and, perhaps, social networking in respect to geo-targeted advertising.

In an interview with Kevin Dennehy of GPS World http://www.gpsworld.com/lbs/did-googles-market-grab-spur-tcs-purchase-nim-9246 I indicated that “…Google is not competing directly with TeleNav, TeleAtlas, MapQuest or anyone else who is not in the advertising business. All the poor souls in mapping, navigation, LBS and Local Search are being crushed by Google’s relentless innovation to produce a better advertising engine. It is impossible for most application companies to compete with Google today, because Google regards their “apps” as a means to an end, not as the core business. They are making so much money that they can afford to innovate, spend 50 million on a project and “dump it” if it does not work they way the want it to work.” I added, “While the end user may think of Google as a utility providing search services, map services, routing services and information services, Google thinks of itself as an advertising engine that has an opportunity to touch its end-users with advertising everywhere they go.”

If my insights on this topic have any value, it is in raising two interesting questions. First, “Just what level of accuracy does Google require to support its advertising business?” Next, “What influence will Google’s data quality have on the rest of the industry?” The most troubling aspect of this latter question, at least to me, is the difficulty of increasing the accuracy of map databases while dealing with the increasing costs of data acquisition in a market that seems destined to provide free map data and routing services. However, let’s get talk about the Google data quality issue and then move on to other topics next week.

As noted previously in this blog, Google dumped TeleAtlas because they were unhappy with the accuracy of the product and thought they could do better. Perhaps this is where we can start the trail of trying to determine Google’s map accuracy goals. Although it is still very early in the game, Google is not providing data that is more accurate than its competitors are providing and in most cases is not providing data as accurate as its competitors map data. Google seems to have adopted its “free” model based on this sentiment – “We will provide the map base and keep it relatively current, but you, Mr. or Ms. User, need to fix what doesn’t work for you.” Is it possible that this will become the entry-level standard in the mapping industry? Of course, that means that you can provide input on map errors that is actually accepted, but let’s discuss that some other time.

Although Google has built an map compilation engine that should be able to solve the map accuracy and currentntess problems of the industry, having built the engine is neither the measure of an ability to integrate it into a system that can solve the problem, nor a measure of the ability of the Google Team to manage a workable map update system to produce the desired level of results.

Some of the errors in the Google-mapbase are unforgivable. For instance, its maps are missing the I-195/I-95 re-alignment in Providence, even though their Street View offering provides imagery of the new alignment (I guess they just didn’t know that the road segment they were driving on was a new section of I-195). If you want more detail, take a look at this note from the Rhode Island DOT website. The DOT’s note ends with this interesting sentence, “Keep in mind that online mapping programs and GPS systems will not reflect these changes.” Thanks to Jim for pointing this out.

According to the Rhode Island DOT this is the situation on the ground. as of last October. You can download a PDF of this image that you can zoom to your hearts delight from the Rhode Island DOT.
This is a map showing the re-alignment of I-195  late in 2009

Of course, if you look online at Bing Maps, you will see this
This is the Bing Maps (NAVTEQ data) map of the area in February 2009

If you look at Navteq, you will see this (at least the traffic is stopped).
This is the same area shown on the NAVTEQ corporate website in February 2009

And Google
No, Google missed it.

and look at the aerial image Google provides.
Hmmm, looks like new construction, doesn't it?

Ouch! If this is the best of premium data and free data, well, bring back the days of paper maps. Maybe there is an alternative. Let’s talk about that next time.

Click for contact Information for TeleMapics

Bookmark and Share

Posted in Bing maps, Data Sources, Google, map updating, Mapping, Microsoft, Mike Dobson, Navteq, Nokia, Personal Navigation, place based advertising, routing and navigation, TeleAtlas, TomTom

14 Responses

  1. Richard Fairhurst

    Ouch indeed. But the best of free data is surely at OpenStreetMap – http://osm.org/go/Zecfw64P– – where the data appears (to my eye, an entire ocean away) to be fully up-to-date.

    Great comment Richard! Although the OSM data is not navigable, it appears up-to-date and certainly provides a more accurate representation of the road geometry for the I-195 re-alignment than that provided by the commercial firms Thanks. Mike

  2. Russ Nelson

    The OSM data can be used for routing, although the OSM servers do not provide any routing services on their own. You need to go to a third-party site, such as Cloudmade, for routing services.

    Thanks Russ, I appreciate the correction. You may want to see the blog above this in which I discuss the situation.

  3. Harry Wood

    What d’you mean by “Not navigable”? Are you talking about data connectedness for routing algorithms? Try routing around your example using CloudMade routing on OpenStreetMap data here: http://bit.ly/apyooy Seems to work pretty well.

    Mainstream satnav makers are generally not interested in using OpenStreetMap yet because they’re all busy trying to get consumers locked into their paid-for maps, but there’s a few smartphone apps using OSM data. These get better and better as the data improves.

    To be fair there are some connectivity issues in the U.S. OSM data which can be fairly disastrous for rout-ability. The project needs to attract more U.S. contributors to tackle this. See http://wiki.openstreetmap.org/wiki/TIGER_fixup

    Thanks Harry. Very informative and I appreciate the correction. You may want to see the blog above this in which I discuss the situation. –Mike

  4. Komяpa

    There are a lot of services that provide navigation on OSM, for example, http://maps.cloudmade.com or http://yournavigation.org/

    Thanks Komzpa. I appreciate the correction. You may want to see the blog above this in which I discuss the situation. — Mike

  5. Plepe

    The OSM data is very well routable, only the OSM page itself does not provide routing. Check out:

    * http://maps.cloudmade.com/
    * http://www.yournavigation.org/

    More information at: http://wiki.openstreetmap.org/wiki/Routing .

    Plepe.

    Thanks Plepe Very informative and I appreciate the correction. You may want to see the blog above this in which I discuss the situation. — Mike

  6. Peter Dörrie

    May I point out, that osm data is fully routable on all levels (cars, bicycles, pedestrian and even wheelchair)? Have a look at http://www.openrouteservice.org/ . While not as polished as commercial routing services, it offers more for its money (which is free). Oh, and you can import OSM-Data on Garmin devices as well.

    Greetings,

    Peter

    Thanks Peter. I appreciate the correction. You may want to see the blog above this in which I discuss the situation. –Mike

  7. Jeff Ollie

    An interesting tool for comparing maps is Geofabrik’s map compare:

    http://tools.geofabrik.de/mc/?mt0=mapnik&mt1=googlemap&lon=-71.40075&lat=41.81547&zoom=14

    They don’t have Bing or Navteq on there yet though.

    Thanks Jeff. Thanks for the link. I appreciate the correction. You may want to see the blog above this in which I discuss the situation. — Mike

  8. Tim Waters

    Mike,
    Oh, OSM data is indeed navigable.

    There’s a few online routing services that use the data, for example, Cloudmade’s offering, using OSM data, which correctly routes cars over the bridge.
    http://is.gd/8wfeS

    Thanks Tim. Thanks for the link and I appreciate the correction. You may want to see the blog above this in which I discuss the situation. — Mike

  9. Shaun McDonald

    There is currently no routing on the OpenStreetMap site, however there is an increasing number of other people who are providing routing facilities using the open data.
    http://wiki.openstreetmap.org/wiki/Routing

    CloudMade: http://maps.cloudmade.com/?lat=41.819496&lng=-71.405897&zoom=15&directions=41.824420759325385,-71.41993045806885,41.81361055134571,-71.38516902923584&travel=car&styleId=1&opened_tab=1

    Thanks Shaun. I appreciate your comment. For more on the topic, see today’s blog (February 16, 2010). – Mike

  10. Matt Blackler

    Topological consistency in the UK is a bit off – I have used data in 9 different study areas for calculating drive time isocrones and populations falling therein. After splitting down the data (where an end node meets another line mid-line) it worked rather admirably!

    So, yes, routing can be done, but I think a certain amount of cleansing is required to get really good results (but I can only speak for the UK). But, as a result of my mainly positive experience of osm, I am keen to contribute to fix some pockets of non-mapped areas that I found! I have got the OSM bug!

  11. vanomel

    Have to say about OSM routable maps on Cloudmade ^_^

  12. MDob

    Thanks for your comment, Matt. I think that “OSM bug” is catching and that’s good.

    –Mike

  13. MDob

    Vanomel – Thanks for your comment. –Mike

  14. Don

    Hey Mike, I can’t believe you wrote this without reporting on Tele Atlas. Look at that area in Providence on routes.tomtom.com ****** Don

    Don – you shamed me into doing just what you suggested. Today’s blog describes the result of my examination. I guess I am going to have to apologize again, but this time for including TomTom TeleAtlas. The route that I created using the site was one of those “impossible” routes. If I had run it a year ago, it would have worked, but the Rhode Island DOT tells me that the old north loop of I-195 was closed in November, 2009 and is now being demolished.

    Other than that, it was, as always, good to hear from you. — Mike