Paper towns and AI

My friends run an internet-famous dance club that doesn’t exist. It began in 2022- they graduated, moved to beautiful San Mateo California, and wanted an easy way for folks to visit. Instead of sending out their new address, they registered their address on Google Maps as an exclusive dance hall, ‘SMPGHQ’ (bonus points if you can guess the acronym).

Friends and family left reviews on SMPGHQ’s listing over time. Some complimented the staff, others remarked on the food, a few shared that they met their spouse while dancing. I cite HQ as a reason that I moved to San Mateo after college, and in a recent telling of the story my coworker searched “SMPGHQ” in Google and asked me quizically, “Wait, is this is a real thing?”.

The first search result was from a website called evendo.com: “SMPGHQ: The Heartbeat of San Mateo’s Nightlife”. An official-enough looking page opens with this paragraph:

SMPGHQ is an electrifying dance club in Central San Mateo, CA, perfect for those looking to experience vibrant nightlife, live music, and an unforgettable atmosphere. With a reputation for great music and energetic crowds, it’s a must-visit for tourists eager to dance the night away.

It goes on to recommend what to wear, what songs might play, how to be informed of for special events. All mined from the facetious reviews left by our friend group.

Encountering SMPGHQ on a travel website reminded me of “paper towns”. For those who didn’t read the John Green novel, a paper town was a fictious settlement intentionally included in a map for the purpose of exposing copyright infringement. Cartographer A spends time and money to survey a location, Cartographer B copies and re-packages A’s map but includes the non-existent paper town, and now Cartographer A has a case that they are being ripped off. What’s fun is that sometimes people read a map looking for a place to relocate, head to the location of a paper town, find nothing, but stay and make that fictional location a reality.

The internet has long been a lucrative source of data; this phenomenon has only accelerated as companies race to dredge up all of the natural text that they can to force-feed an LLM. I’m not sure of the data mining process that evendo.com used to generate their travel site, but recognize that it’s increasingly trivial to point a coding agent to Google Maps data and get an aesthetic webpage.

So what happens when a tourist shows up to the streets of central San Mateo looking to party?