Google Maps Is Manually Put Together, A Lot More Than You'd Imagine

Street View images are used to extract street names and signs

  How the Google Maps data is represented internally
Google Maps is something most people use, maybe once a week, maybe several times in a day. But it's not really something most people think about. What's more, there's a growing trend of ditching Google Maps, the new iPhone comes with its own built-in app, and several other websites are dropping Google Maps in favor of Open Street Map for example.

Google Maps is something most people use, maybe once a week, maybe several times in a day. But it's not really something most people think about. What's more, there's a growing trend of ditching Google Maps, the new iPhone comes with its own built-in app, and several other websites are dropping Google Maps in favor of Open Street Map for example.

Building a new map app isn't easy, but how hard can it be, really? Well, pretty damn hard it turns out. The Atlantic has managed to get a short lesson on how Google Maps is put together and what the data powering the service looks like.

It's the first time anyone outside of Google has been able to hear about "Ground Truth," the codename of the program that builds the maps everyone sees.

There's a lot more to it than just getting some map data, layering some satellite photos on top of it and calling it a day. In fact, the single most impressive thing is just how much human labor is involved.

The data that Google acquires is far from perfect and deviates from reality in many details. A road may be a few meters off, it's not a huge difference in the grand scheme of things, but when you draw a road in the middle of what anyone can see is a field in the satellite images, it creates a problem.

Google uses a variety of ways of improving that data, for example, it relies on several data sources. But it's also building its own, the Street View cars that scour the Earth do more than collect pretty pictures, they collect GPS data that can be used to accurately draw any road that the cars go through.

But that's just half the story, the photos themselves are a huge source of data. For one, any detail can be double-checked by looking at the photos.

Google employs its image recognition technology to read street signs and determine what's the name of the road, what's the house number, whether it's a one-way street and so on.

On top of this, Google hires a lot of people to manually fix problems, problems with the original data, or things like new roads, road works and so on. Adding it all up, Google is sitting on a huge amount of information on every corner of the world.

Comments