What’s a pretty street anyway?

How do you build the most beautiful path?

One of the unique aspects of the app we have in mind, is that it’s NOT all about POIs. Sure Points of Interest matter. You don’t want to miss the Eiffel Tower or Empire State Building when you visit the city for the first time.

But often the occasional narrow street, without a landmark, monument, or famous place, but with ambiance, will make your day.

So we just want to make sure, while you go from POI to POI, that you walk through as many of these charming streets as possible.

Now, how the heck are we going to do that?

We identified 3 possible ways:
1- manual: either ourselves, users, or professional guides, would determine what are the pretty places, like we could use a highlighter pen to show on a map where to go in our city to one of our friends
2- automatic: based on big data (ha! here’s the most overused expression of the past 5 years), run a pass on where the most likely “pretty places" are
3- a mix between the two

For option 1, and from our own past experience and failures, we decided that getting enough people to participate in such a grand project from scratch would probably take quite a long time before we have enough usable data. So we decided to go big. Data. And scalable (yup, buzzwords galore).

So how can we determine what’s a pretty street? After talking with people, reading stuff (http://en.wikipedia.org/wiki/Walkability) and pondering things, we identified the following pointers:

- number of geolocated pictures (with the need to handle false positives, like a demonstration, sports or business event, etc).
- Quiet (based on amount of car traffic)
- Pedestrian areas
- Monuments, Historical or Architectural buildings (yep, POIs can be positive moderators to determine if an area is interesting)
- Trees
- Restaurants (in particular when they offer outdoor seating)
- Specific outlets (design stores, flowers, etc.)
- Presence of water (promenades near the water tend to be favorites)
- And other small moderators like cobblestones. OK, some people hate cobblestones (ask my GF when she’s wearing high-heels…)

Agreed, no system, manual or automatic, will be for everyone. Because beauty also depends on the light, season, climate, mood, and ultimately the eye of the beholder.

However some places are generally considered as beautiful or pleasant, and we believe that if we can find relevant data for at least some of these points, we should be able to build a first pass of our model on a large scale, hopefully providing a base for people to help feed a free, open and participative map of beautiful places. Which we can then use to build not the fastest walking itineraries, but the most beautiful.