The Shape of Flickr Nation

US alpha shape

What is the correct way of representing our cities and countries on a map? This is a question long debated by cartographers. The website WorldMapper provides hundreds of different views. Flickr has just shared their own unique view of the world, built from database of 90 Million geotagged photos. Each of those photos comes in with a set of lat-long coordinates. If we clustr all of those photos together we can see what Flickr Nation really looks like.

There’s more to geotagged photos than just putting them on a map. About a year ago, Flickr introduced Places which shows iconic photos of a location and lets you view tags with a geographic twist (see this Radar post on the “chicken” tag in the US vs. China for an example). The Flickr team has used them to create alpha shapes of neighborhoods, cities, states, countries and continents. The shapes are not always the ones that you normally see on a map, but they definitely define a recognizable area.

The image above shows the United States. Not all of the boundaries are accurate. It includes the coastal waters and parts of Canada and Mexico. Where should shapes boundaries stop? Should it include places that can view the US? Or is it just the results of a faulty algorithm?

The Flickr Dev Team reverse geocodes the lat-long points to come up with a human-readable location. There are up to 6 locations stored with each photo. These locations can span from neighborhood to continent. Each of these locations are associated with a WOE IDs (Radar post) viaYahoo’s publicly available location service. You will be able access the alpha shapes via the Flickr API (with its WOE ID) very shortly.

Sure these shapes look cool, but what are they good for? As Aaron writes “These are not maps you should use to guide your spaceship back to Earth but they’re probably good enough to explore the possibilities.”. Perhaps they are the geographic shapes of social media. How closely aligned will Twitter Nation and be to Flickr Nation?

Image Credits: Aaron Straup Cope (US)

Apologies to Thomas Hawk for the use of Flickr Nation.

Update from Aaron: Just an FYI that the polylines for clustr shapes have been available via the API since launch.

It’s just the actual shapefiles that have been missing but the refeed is finally underway as I write this. Here’s Germany, for example.

tags: ,
  • http://blog.planeteye.com/2008/11/05/crowdsourcing-geodata/ Juan Gonzalez

    Why is this important?

    Building a service around geocoded data has become easier with the proliferation of mapping technology…..as long as all we do is display data on a map. If the service tries to do any calculations based on the geographical context, the maps are useless as they are nothing else but an image.

    To really leverage the hundreds of thousands data points in our database to deliver great travel recommendations, we need to be able to answer questions such as “Is this hotel located in the city of Toronto?” or “What is the closest town with a good variety of restaurants?” or “Can I get a similarly-priced hotel closer to downtown?“.

    Aside from some serious technology to manage large volumes of data, the key to solve these is what is known in the GIS world as shapefiles. While there are many sources for them, very few are available at low cost. Until now.