Neighbourhoods are a subject close to my heart. For my current project, Kneedl.com (a site aimed at helping people find the right neighbourhood in a new city) I looked into finding accurate neighbourhood boundaries. Zillow and Flickr have both made shapefiles available for free, and with Yahoo Placemaker there’s some really interesting data emerging. Unfortunately when we started on our site, I found the quality of public data just wasn’t good enough.
Miami (a small, but sprawling city) has approximately a hundred distinct neighbourhoods, ranging from large districts (cities–in the American sense–in their own right) down to just a few blocks. But the issue isn’t scale, it’s the fact that neighbourhoods aren’t official boundaries, one person thinks they live in Wynwood, whereas someone else might say they are in Downtown even though they live in the same building. There’s a certain amount of vanity in defining exactly where you live (’Chelsea Borders’ for example in London rather than admitting residing in Fulham).
For Kneedl, I had to manually create the boundaries to Miami’s neighbourhoods. Partly because I couldn’t find sufficiently granular data (most data sets are machine generated from census data and frankly wrong) and partly because we wanted to create the back-end tools for users to define their own neighbourhoods in the future. I did this with official data, neighbourhood websites and by speaking to local residents, a time consuming and manual process. Collaborative neighbourhood boundary definition is definitely the way forward. I believe the New York LA Times is doing some work in this area (thank you Simon for the link) with users drawing boundaries and other users voting on their accuracy. Although there will never be a definitive dataset (because of the subjective nature of neighbourhoods outlined above) I hope that these new sources are used by more and more sites as we’ll all benefit from an increase the relevancy of neighbourhood information.