Yahoo Can Pick The Most Beautiful Route For You Instead Of The Quickest

Stroll the nice way home

By
Head Image
© 2017 Neil Howard Flickr

I never truly realised how much Google maps was dictating my urban experience of a city until I moved to Edinburgh a few years ago. New to the ancient city, unsure where anything was, my smartphone became my trusty guide. 

For months I diligently followed the routes to get from A to B. Until one day, walking with a friend to one of my favourite pubs, things took an unexpected turn. She directed us up a small, cobbled side street, which was a supposed short cut to our destination. I realised that all of this time, I had been following Google's advice up main roads and streets, and unknowingly missed all of the closes, alleyways and cobbled sidestreets. It was a hard realisation, and one I tried to remedy immediately. 

While Google's intentions are good, and often extremely helpful, we have to be aware of how these tools shape our experiences, and can let us miss out on some of the best parts of a city. 

That's why I'm interested in the new project being undertaken by Daniele Quercia at Yahoo Labs in Barcelona. A team of researchers came up with the idea to make an algorithm that will figure out more pleasant or scenic routes from A to B, which don't take infinitely longer than the ordinary suggestion (Quercia says that on average these routes take just 12% longer than the shortest routes, which makes them plausable alternatives for someone on foot). 

They built the project initially by testing out photos along London routes from Google Street View and Geograph on UrbanGems.org, with people choosing the image that they thought was more beautiful, happy, or peaceful.

Then the images were plotted with their scores of pleasantness on a map. The research team's algorithm chose the most beautiful route among several alternatives based on pleasantness scores.

Then, the reasearchers asked  30 people from London to assess the paths the computer guessed were the nicest. The participants agreed with with the computer’s decision.

Obviously using this method for every city would take an age, so the team used metadata from millions of photos from Flicker taken of the same routes in London, and evaluated the routes according to positive or negative comments, how many times a location was photographed (which might infer beauty/interest), and other kind of endorsement-like data. 

To test this method, they carried out the London project in reverse on Bostonians, letting the Flickr-determined routes be appraised by people. Again, the people agreed with the computer. 

If this software becomes available soon, I guess the next time I want to explore a new city, I wont have to worry.