One of my guilty pleasures is Pokémon go. It’s a fun game that gets me outside.
For no players, all you need to understand is that Pokémon go places virtual objects across the globe; you need a phone with the game to find and interact with them. But there must be millions of them.
One thing I find incredible is the richness and accuracy of the placement of objects. It speaks to an incredibly powerful dataset.
A simple example are the trails here in Washington state. Granite Mountain is a 7 mile hike up to a peak near Snoqualmie pass. It’s a rigorous trail, popular in summer, but by the standards of a city location it’s very lightly trafficked.
Yet there are three Pokestops on the trail! One at the trailhead/parking area, one at a viewpoint halfway up, and one at the summit fire tower lookout.
Data. Probably Google data which niantic’s historical relationship would enable.
Here’s my hypothesis: First, Google knows from smartphone users that this trail is a high traffic route for the middle of the woods. So they should place some Pokestops. But where? The start and turnaround points make sense, but did they really build trail logic in? And how did they find the viewpoint halfway? A better hypothesis would bring new data in: Photos. Google would know where the most photos are taken, and I’d bet the start, viewpoint, and finish are huge clusters.
I haven’t figured it all out yet.
Pokestops all have a photo associated with them. They do a great job of providing an illustrative one; no selfies, no off angle or poorly lit. This too can’t be manual, so how?
Another mystery is how Pokestops are named. Always right, but how did Niantic know this was a reader board??! It’s impressive.
What all this really shows is the power of data, in this case Google’s data. The game would be impossible without it. Google didn’t collect it for this purpose, but once collected and catalogued, you can do all kinds of interesting things!