"The algorithm for drawing neighborhoods is the "blobby" algorithm, well known in computer graphics. You can think of each point in a neighborhood as a little magnet, and the neighborhood is the region where the combined attraction of all those magnets is above a certain strength. A single point makes a small circle on the map. The influence of a number of nearby points will combine to make a curved blob. Read more about blobbies:

• Implicit Surfaces
• Metaballs 
• An Overview of Metaballs/Blobby Objects"

"The Neighborhood Project
compiled by Nora Parker, Senior Managing Editor 

Matt Chisholm and Ross Cohen are working on a project to define neighborhood boundaries, so far only in San Francisco, but eventually in other cities as well. (If you get impatient and want to launch this project in your city, they suggest you can download the software and get busy.) Neighborhood boundaries? What does that mean? Generally neighborhoods are not defined by exact boundaries - they are defined by what geographers call "fuzzy lines" - lines that are not well-defined. (An example of "fuzzy line" might be the line between two bioregions. These are generalized, mappable regions, that might involve factors such as precipitation, soils, topography, etc., but there is no defined line on the ground when you cross from one region to the other - the change between regions is more gradual. This contrasts with the line between two countries, for example, where you might literally be able to stand at a border crossing with one foot in each country, or even closer to home, the line between my neighbor's property and mine.)  The Neighborhood Project is an excellent illustration of the concept of fuzzy lines - if you go to the site and look at the map, you'll note quite a bit of fuzziness to the neighborhood definitions. You can even begin to perceive it in the very small scale map included below. People living next door to each other might consider themselves to be in two different neighborhoods.

The site uses housing post data from craigslist, which includes addresses and neighborhoods, as well as a public poll on the site, to generate address/neighborhood pairs. Open source products (geocoder.us, Python and PostgreSQL) are used to geocode and map the data.

The Neighborhood Project's map of San Francisco. The dots are colored based on what neighborhood residents considered their "home." Used by permission. (Click for larger view.)

---