Comparing Google and Microsoft's Mapping Coverage

I wrote recently about Google's applaudable improvements to their mapping system, remarking that they are still trailing other mapping services (such as Microsoft's). My observations about the data coverage of Google's mapping is based on many interactions with the system, but it is not remotely scientific.

For example, I wrote about the complete absence of Cozumel - a post which received a comment from Andrew Hitchcock about the missing island of Hong Kong.

Someone going by the name of Jon posted this useful comment on my latest post on the topic:

Trailing it's competitors? What are you smoking? Microsoft doesn't have more coverage, and never did.

It is an interesting comment - particularly to me, after having thought about and read about the power of brands to help users fantasize about the quality of service. Jon is, of course, not very well informed.

However, I thought it might be a good time to consider a more methodological approach. Taking a list of countries ordered by population, and starting at the bottom, I reviewed the data found on Google and Live (Microsoft). I scored each location for the presence of the shape of the country, the presence of populated places (PPL) - that is to say cities, towns and so on, the presence of roads, and which system had the better imagery - meaning the highest resolution, not the most up to date.

In the table below, the presence of features is marked for Google and Live. For example, Y/N means that Google has the feature and that Live doesn't.

Country	Population	Shape	PPL	Roads	Best Imagery
Pitcairn	50	Y/Y	N/N	N/N	Live
Vatican City	800	Y/Y	Y/Y	Y/Y	Live
Tokelau	1400	N/Y			Google
Niue	1600	Y/Y	N/Y		Google
Falkland Islands	3000	Y/Y	N/Y	N/N
Montserrat	5900	Y/Y	N/Y	N/Y	Google
Saint-Pierre and Miquelon	6125	Y/Y	N/Y	N/Y	Google
Saint Helena	6600	N/Y	N/Y	N/Y	Live
Saint-Barthélemy	6852	Y/Y	N/Y	N/Y	?
Nauru	10000	Y/Y	N/Y	N/Y	Live
Tuvalu	11000
Anguilla	13000	Y/Y	N/Y	N/Y	Google
Cook Islands	13000	Y/Y	N/N	N/N	Google
Wallis and Futuna	15000	Y/Y	N/Y	N/Y	Google
Palau	20000	Y/Y	N/Y		Live
British Virgin Islands	23000	Y/Y	N/Y	N/Y	Google
Turks and Caicos Islands	26000	Y/Y	N/Y	N/N	Google
Gibraltar	29000	Y/Y	Y/Y	Y/Y	Google
San Marino	31000	Y/Y	Y/Y	Y/Y	Google
Monaco	33000	Y/Y	Y/Y	Y/Y	Google

It is interesting to note the trend indicated by this summary analysis: Microsoft is wining on country and road data, while Google is ahead with imagery. Of course, there are many other ways to sample the data (e.g. via areas with high tourist traffic), and it would be nice to actually build a comprehensive version of the table above - or even display graphically coverage for different system.

I'll finish with this beautiful image from Google of Tokelau, a tiny nation in the south Pacific (and by nation, I mean non-self-governing colonial territory of New Zealand).

Comments

[Nick Johnson]

You're missing a couple of datapoints: Does it have geocoding (can you enter an address and have it find it) and does it have routing? In general, routing is more difficult than geocoding, which is more difficult than mapping. Also, countries with non-western addressing schemes make geocoding really hard, so it's all the more impressive if they can do it for more out-of-the-way places.

Disclaimer, for what it's worth: I work for Google (though not on mapping), and until recently I worked for a GIS software vendor.

[Raj Rao]

It is also interesting to note the differences in imagery that Google and MS have for US itself... I find it extremely hard to categorize which company has better imagery - though both of them seem to have just as good road coverage. If you check out my latest post - http://therajahs.blogspot.com/2007/09/creating-custom-maps-in-microsoft-live.html - you will see that I have 2 examples - in one Google has better imagery and in the other Microsoft. So it looks like tools like http://www.jonasson.org/maps/ that allow you to compare the 2 maps are of immense use to figure out which company's mapping solution one should use at any given time.
Another interesting thing to see is that when you zoom out - I have typically seen that Google seems to have better imagery.... but as you zoom in - it can go either way.
But as far as searches go - MS seems to return a lot more relevant searches for the US atleast.