🔎 INVESTIGATION #6
Do Google & Apple Label the Same Kinds of Places?
In a dense city like New York, there are so many different kinds of places: hospitals, parks, homes, schools, grocery stores, fire stations, skyscrapers, government buildings, cathedrals, universities—the list goes on and on, and it’s seemingly endless.
Take the maps below: there are thousands of different places that could’ve been labeled on either of these maps:
But similar to what we saw with all of the Bay Area’s roads, it would be impossible to show everything that’s actually there and still have a legible map:
Instead, the maps show just a tiny selection of everything that’s actually there.
Remember Faisal’s picture of the Empire State Building at the beginning of this essay? That’s a picture of the same area shown in the maps above. There are a lot of places there!
* * *
So out of thousands of possible places in Midtown Manhattan, Google and Apple choose to label these specific ones:
But how do Google and Apple choose these places? (They had thousands to choose from—how did they end up with these?)
And are there specific kinds of places favored by each map?
Let’s dig into this more…
* * *
First, we’ll divide all of the places on Google Maps and Apple Maps into eight simple categories:
- Things to Do & See: Landmarks & attractions, entertainment (theaters, stadiums, and nightclubs), and recreation (parks and other spaces).
- Eat & Drink: Restaurants, bars, coffee shops, etc.—anywhere you can be served prepared food and beverages.
- Shopping & Services: Banks, hotels, shops, stores, markets, etc.
- Transit Stations: Train and subway stations, bus and streetcar stops.
- Other: Anything that doesn’t fit into the other seven categories: corporate headquarters, apartment buildings, homes, etc.
Now that we have our eight categories, let’s count how many categories appear on each map.
We’ll do it like this for each pair:
So above, Google shows four categories, while Apple shows five.
Now, let’s count the number of categories on the rest of the zooms:
And next, let’s plot and average those counts:
Interesting. At nearly every zoom, Apple Maps is showing a greater number of categories.
Though given what we saw in New York (Apple showed five categories, while Google showed four), perhaps it isn’t too surprising?
Let’s take a closer look at each zoom, and see if we can find some other patterns…
* * *
While we were counting the number of categories on each zoom, I also recorded the number of places for each category shown:
This is cool because it’ll allow us to see which categories appear most at each zoom.
Like here in New York:
Interesting, isn’t it? The two maps are really different.
Early on, Apple shows airports, while Google shows transit stations. And then as you continue zooming in, both start showing a greater variety of places—though Apple shows a greater variety overall (just as we saw in our graphs).
And this certainly matches the map we just looked at—there are lots of transit stations on Google’s map and a greater variety of place categories on Apple’s:
Let’s see if we see the same pattern in San Francisco:
Here in San Francisco, we’re generally seeing fewer places per zoom than we did in New York, but the results are largely the same: Google is prioritizing transit after the first few zooms, while Apple is prioritizing landmarks.
Looking at the bar graphs above, z13 looks like one of the most different zooms between the two—let’s take a closer look:
Ah, this is that zoom where Apple showed very few text labels for roads. Remember that?
But perhaps now we see why: Apple is actually showing a greater number of places here than Google (15 for Apple, versus 6 for Google).
With so many more places labeled on the map, there’s probably less room for road labels—hence, the lower road label count we saw earlier.
But let’s test this theory.
First, we’ll merge the two z13 maps together:
Now, let’s look and see if any of Google’s road labels collide with Apple’s places.
Look at that! Apple’s place labels collide with seven of Google’s road labels. In other words, those place labels are occupying space on the map where roads could’ve been labeled. That’s why we saw the lower road label count on Apple earlier.
It’s interesting, isn’t it? Google is prioritizing roads, while Apple is prioritizing places. And that’s why were seeing such noticeable differences in the counts.
But even when it comes to places, the maps are prioritizing different things:
Above, Google is again prioritizing transit places, while Apple is prioritizing landmark places. And Apple is also showing a couple of hospitals and a famous bakery.
What’s more important on a map like this? Transit? Or landmarks and hospitals?
Each map voted with its pixels, and it’s interesting to see.
(And here we’re also seeing the argument for map personalization: Not everyone uses transit, so there might be places—such as landmarks—that are more important to some people. And when it’s life-and-death, hospitals are the most important places in the world. But how often does the average person visit a hospital? Personalized maps are better at surfacing the most appropriate places for each person; but for the default map, something still has to be chosen… and that’s what we saw above.)
* * *
Do the patterns we’ve seen for New York and San Francisco also hold for London?
Whoa — look at all of those transit stations shown in London at z13!
Let’s take a closer look at that zoom and see what’s going on:
Ah, it’s all of those Tube and rail stations we saw earlier.
The icons don’t have text attached to them, so Google is able to pack lots of them onto the map. (34 in total.)
It’s a clever way of increasing the map’s information density: the icons give just enough information that you know that they’re Tube stations—and without text, Google can fit even more of them on the map.
That’s why Google’s count is so high at this zoom.
* * *
Now that we’ve looked at each city individually, let’s average them together to see the overall pattern:
Google Maps prioritizes transit stations for several zooms, while Apple generally labels more landmarks. And as you zoom in, both maps show a increasing variety of categories.
Both maps also show a surprising number of restaurants (the yellow parts of the bar graphs). But notice that the restaurants don’t start appearing until later zooms.
And that’s another interesting point, isn’t it?
The places that each map feels are more important are shown earlier, while the places that each map feels are less important are shown later.
It’s clear that Google thinks transit is important, while Apple thinks that airports, hospitals, and landmarks are important.
Two very different views of the world!
* * *
Now that we have data on the kinds of places shown at each zoom, let’s look at how balanced an average zoom is on both maps:
We’ll use pie charts:
A zoom is generally “balanced” if no single category comprises more than 50% of its places.
And as you can see from the pie charts above, Google isn’t particularly balanced for the first nine zooms it labels places on. And in comparison to Google, Apple generally has a better balance of place labels per zoom, overall.
We’ve looked at the place categories that are shown on each zoom — but what about all zooms, taken together?
Which category does each map show the most, across all zooms? In other words, which category is each map’s favorite?
Let’s investigate this one last thing…