A quiz (and it would be helpful to know the area on this one): which of the following street maps, generated from one of the many online mapping services, is accurate? (Focus on the intersection of streets near the centre of each map.)
Google Maps perhaps?

Or
Yahoo! Maps which looks very similar in content and appearance?

Or
MSN Maps / Live Local, also similar in appearance and content?

Or perhaps
Map24 which exhibits a different style and geometry?

. . . which is also similar to
Rand McNally’s geometry . . .

. . . and online map provider leader
MapQuest:

Two different sets of geometry. So who’s right? The answer is NOT
Navteq, even though it seems to have supplied both geometries (how can that be?). What about Navteq’s biggest competitor, TeleAtlas (by the way, who uses
TeleAtlas in the online street mapping world?)?

Something completely different again. The correct alignment and geometry of roads can be seen on this map, provided by the
City of Peterborough on its
online mapping site (
MapGuide plug-in required, only works in IE; ignore the white patch on the right side):

And, interestingly enough, also in the
National Road Network data layer provided by the federal government, free of charge:

(which is not to say the NRN is any better than Navteq or TeleAtlas (it seems to fall down in the more rural areas of the country). Perhaps this little piece of geometry is not a big thing - but it does affect driving directions -
Google Maps gets it right according to its own geometry but wrong according to reality.
Rand McNally gets it right according to reality but wrong according to its own geometry (
but wrong when reverse directions are considered).
All of this suggests that local knowledge can’t be beat which is why, perhaps,
TeleAtlas has initiated an online map feedback tool to elicit such knowledge from its users. How will they determine what information they receive is correct and what is chaff sown by nasty competitors? It will be interesting to watch what happens - and if Navteq picks ups on a similar approach.
As an aside - none of this takes into consideration that, when zoomed in enough, many of the streets which should be straight seem to angle from intersection to intersection. Is it a case of someone setting the snapping tolerances too high?
0 comments