Justin O'Beirne

Justin O'Beirne of San Francisco, California. Essays, projects, and contact information.



Page 1 · 2 · 3 · 📍 · 5 · 6 · 7 · 8 · 9


How Often do the Maps Label the Same Things?

In Part 1, we compared Google’s and Apple’s Point of Interest labels and found that, on average, the maps have only 10% of their POIs in common:

But there are many other kinds of labels on the maps: countries, states, cities, islands, rivers, oceans, continents, freeways, etc.

How often are the maps labeling the exact same things (e.g., the exact same cities, roads, countries, etc.)?

For example, take these New York maps at z9:

Google labels 25 unique things, while Apple labels 42—and together, they label a combined total of 67 unique things.

But out of everything labeled above, there are only 4 things labeled on both maps: New York, Newark, Hempstead, and Edison. In other words, the two maps have just 5% of their labels in common.

That’s surprising. 

But is it the same for all zooms?

Let’s find out…

*   *   *

First, we’ll take all of our map pairs and count the number of labels that both maps have in common:

Next, taking the total number of unique labels on each map pair, let’s calculate the percentage of labels that the maps have in common at each zoom:

Now that we have our percentages per zoom, let’s plot them on a graph so that they’re easier to compare:

In terms of what they label, it seems that Google Maps and Apple Maps are the most similar early on. But as you zoom in, they diverge, and after z6, they never have more than a third of their labels in common.

In and of itself, it’s surprising that the two maps, on average, never have more than 57% of their labels in common at any one zoom.

*   *   *

The graph above shows that z2 is the zoom where the maps are the most different, having just 4% of their labels in common.

Let’s take a closer look at z2 on our New York map pairs:

Glancing at z2, it’s not hard to see why the maps have so few labels in common: they’re labeling different things.

Google is labeling countries, while Apple is labeling continents and cities. In fact, the only thing the maps have in common are their ocean labels.

That said, it’s puzzling that both maps have so few labels. (Google has 13, while Apple has just 8.) Combining the two maps together, we see that Google could fit some of Apple’s cities (such as New York and Los Angeles); and that Apple could fit some of Google’s country labels (such as the United States and Mexico):

Why not label continents, countries, and cities together?

*   *   *

The zoom where the maps have the most labels in common (z4) is only two zooms down from our most different zoom (z2):

Let’s take a closer look at that zoom (z4) on our New York map pairs:

New York’s z4 map pair gives us a view of the U.S. East Coast and Southeastern Canada.

Let’s highlight the labels that the maps have in common in green:

As expected, there’s a lot of green on both maps. (It’s our most similar zoom, after all.)

However, given that the maps label many of the same things, let’s instead focus on the labels that are different:

There’s a pattern here… do you see it?

I’ve highlighted it below:

On Google Maps, the labels that are different from Apple are all state and province labels (that is, except for Ottawa). And on Apple Maps, the labels that are different from Google are all city labels.

If we combine the two maps together, we start to see why they’re different:

Notice that Google’s labels for West Virginia and Maryland overlap with Apple’s label for the city of Washington.

Google labels West Virginia and Maryland, but in doing so, leaves too small of a space to also label Washington on its map. 

Apple, meanwhile, labels Washington—but Washington’s label occupies the space where West Virginia and Maryland would be labeled, so they don’t appear on Apple’s map.

There are at least three other similar situations on the combined map above:

  • Google labels the state of Connecticut, while Apple labels the city of New York.
  • Google labels Arkansas, while Apple labels Memphis.
  • Google labels Louisiana, while Apple labels New Orleans.

In each situation, it appears that Google is prioritizing state and province labels, while Apple is prioritizing city labels.

And a quick glance at z3—the zoom right before this one—seems to confirm our suspicion:

Notice on z3 that Google labels states and provinces—but no cities. And Apple labels cities—but no states or provinces.

So going back to z4, it seems that Google is indeed prioritizing states and provinces, and that Apple is indeed prioritizing cities.

It’s fascinating, isn’t it?

Even on the zoom where the maps are the most similar (in terms of what they label), we see competing design approaches.

Above, Google is aiming for completeness. Notice that it manages to label each and every U.S. state and Canadian province in view. But it does so at the expense of not labeling the U.S.’s largest city and its capital city—New York and Washington, respectively.

And New York’s and Washington’s absence is even more conspicuous when you consider that Google labels Toronto and Ottawa (Canada’s largest city and capital city, respectively) — without also labeling New York and Washington.

Apple, meanwhile, labels New York and Washington—but at the expense of not labeling several U.S. states. And just as it looks peculiar for Google to label Toronto and Ottawa, but not New York and Washington—it looks equally strange to see an incomplete listing of U.S. states on Apple’s map.

What’s more important on a zoom like this? States or major cities? And are more people looking for Connecticut or New York City? West Virginia or Washington, DC?

And how important is it to label all of the U.S. states at once?

It’s a classic cartography problem, and it’s part of what makes cartography so fun. You don’t have enough space to label everything, so what do you label? What are the top things here?

Each map has voted with its pixels, and it’s interesting to see:

  • Google thinks states are more important—so Google labels all of the states, and if there’s space left over, it then labels cities. (Hence, Ottawa.)
  • Apple thinks major cities are more important—so Apple labels the biggest cities, and then if there’s space left over, it labels states.

What do you think is more important to label at a zoom like z4? States or cities?

Neither is wrong—but each decision comes with its own tradeoffs.

*   *   *

It’s also interesting that z4 is the most similar zoom, given that it’s one of the very earliest zoom-levels. Because it’s so early, our z4 maps show a relatively larger chunk of the world than any of the maps after it:

Notice how wide of an area our z4 maps show—almost the entire Eastern half of the U.S. and the most populated parts of Canada:

And within the area shown in the maps above, there are millions of things that could be labeleddozens of states and provinces, thousands of cities, tens of thousands of roads, and tens of millions of points of interest—among many other things.

In other words, because z4 shows such a wide area, there’s a greater number of label candidates at z4 (i.e., things that could be labeled) than at any zoom after it.

Isn’t it interesting, then, that the maps are most similar where they have the greatest choice in what to label (z3, z4, z5, z6)?


Page 1 · 2 · 3 · 📍 · 5 · 6 · 7 · 8 · 9