I came across this very interesting behavior, while messing around (a technical term) with Google and Bing Map’s geocoding services (see a previous post here). I always thought that a location on the ground defined in Lat/Lon would be decoded to – more or less- the same address. I was proven wrong. It seems to depend on the geocoding service you will use. Especially when it comes down to the location of Microsoft’s or Google’s Headquarters.
Take for example this little pair of Lat/Lon values: 47.64054,-122.12934. Now open your web browser and call Bing Map’s REST API using this URL replacing the key value with your key if you have one. If not, see how you can get one here: Getting a Bing Maps Key
<a title=”http://dev.virtualearth.net/REST/v1/Locations/47.64054,-122.12934?o=xml&key=” href=”http://dev.virtualearth.net/REST/v1/Locations/47.64054,-122.12934?o=xml&key=”>”>http://dev.virtualearth.net/REST/v1/Locations/47.64054,-122.12934?o=xml&key=<your_key>
You will get returned an XML file saying that the Lat/Lon you entered corresponds to 1 Microsoft Way, Redmond, WA 98052.
This gets returned with a “Medium” confidence as per the XML. Have a look at the Lat/Lon which are not the ones I entered. Strange…
Now do the same but this time use the Google Service:
The address is now coming back as: 3635 157th Ave NE, Redmond, WA 98052, USA. What’s going on here?
So I had a look at the map searching for the same coordinate pair. In the left image, the green arrow shows where this pair of coordinates are according to Google Maps and the pushpin the address it was matched to. The right image is from Bing Maps.
The location on the ground is the same, but Bing Maps does not seem to attempt to decode it to an address. Searching for 1 Microsoft Way,Redmond,WA 98052 in Google, doesn’t return any exact results and the closest address it gets matched to is around 1,4m north at “Microsoft w Campus Acrd”
Searching for Googleplex on 1600 Amphitheatre Parkway, Mountain View, CA 94043, returned slightly more expected results only a few meters away from each other as shown below (left image from Google, right one from Bing):
So while at first I put my conspiracy hat on, thinking that I am on to something and that the two companies would be biased against each other’s HQ locations, it turned out that it might not be as bad(?), but it still shows that address geocoding is far from an exact science. Having said that, I still find it weird that Google will map a Lat/Lon point to a completely different street than the one the point is on. A behavior which I could not really replicate when trying other locations. Note also that the Street the initial point is on is called “157th Pine” in Google Maps, but “Microsoft Way” in Bing Maps. Older/different data or maybe a conspiracy?