BriFiTwiBo (Brian’s First Twitter Bot)
I live in Cobble Hill, Brooklyn, but you might know of it as BoCoCa, which is, as Roman Mars of the 99% Invisible podcast coined, an “acroname” of Boerum Hill, Cobble Hill, and Carroll Gardens. Now, I sure hope nobody actually calls it that, but the name exists. And it’s dumb. And waiting to be made fun of.
The aforementioned podcast had a recent episode on this very issue:
San Francisco, the area South of Market Street is called SoMa. The part of town North of the Panhandle is known as NoPa. Around the intersection of North Oakland, Berkeley and Emeryville, real estate brokers are pitching properties as part of NOBE. An area of downtown Oakland is being branded as KoNo, short for Koreatown Northgate. But no one actually calls it that, or at least, not yet.
These types of names have spiraled to the point of parody. In an episode of South Park, part of the town gets rebranded as SoDoSoPa (South of Downtown South Park):
After seeing many bots by the famous Darius Kazemi, from the Sorting Hat bot to the Two Headlines bot, I thought this could be my opportunity to learn to make a rando Twitter bot. Thus, Belinda Botrick was born.
The first step was to get to the roots of the acronames—the original neighborhoods themselves. Luckily, Zillow has a huge database of all of the neighborhoods in the US, albeit in a form that had to be massively converted into usable data. This left me with 6,920 neighborhoods to play with.
From there, it was just programmatically changing them into the camel cased acronames: so I broke them up by the first vowel, and if they started with a vowel, went one letter forward. Then, they get a 50/50 chance of becoming “North of” or “South of.” In effect:
North of Tampa International Airport Area, Tampa, FL
Rolls off the tongue, doesn’t it? Well, buy now before the housing value skyrockets!
From there, we have to render a map of our up-and-coming neighborhood full of coffee shops and vinyl stores. I used the Google Maps Static API to render a map of the location, but with one twist: I’d move the map north or south a few clicks on the GPS coordinates depending on which direction the new neighborhood was to the old. Then, with a text overlay, we end up with the map of NoTaInAiAr:
Bringing it all together
From there, it’s time to toss in a little extra flair and get this bot on the road. What real estate agent is without their portrait? Luckily, my friends have been making a habit of photoshopping my face onto real estate agents’ heads. Check.
The language is processed with a series of random real-estate-y talk generated by yours truly tossing what I think the most annoying broker might say.
Then, a little bit of gentrification magic seals the deal. Utilizing the Bing Image Search API, I tack on a random image from the neighboring neighborhood and then one additional annoying stock photo to really get the feel for the place. It’s pretty rough, but here’s how the sausage is made:
Toss ’em all together using the ImageMagick binding for python, and baby, you got a stew goin’:
Belinda the Twitter bot is set to tweet out a new “neighborhood” she’s made up 5 times a day.
There are additional things that I want to add that hopefully I’ll add in the next bit of time. One is having neighborhood portmanteaus—ones that “pop up” between two existing neighborhoods. The other is allowing people to tweet an address at Belinda and she’ll reply with a new neighborhood name based on your surroundings.
But until then, enjoy!
Be sure to follow Belinda Botrick on Twitter @belindabotrick and maybe you’ll find the new neighborhood of your dreams.