We love the power of cartograms to show thematic data maps because they overcome some of the problems of classic choropleth maps. And they can be beautiful. So it’s exciting to see a promising new tool to create nice cartogram hexmaps automatically. It’s called Tilegrams (for “tiled cartograms”) and it has been developed by Pitch Interactive in collaboration with Google News Labs. But let’s back up a little bit, since many people are not familiar with cartograms in the first place!

A cartogram is a map that actually functions as a chart by distorting the size of geographic areas (such as countries or states) in proportion to numerical values they represent. Here are two nice examples by John Tomanio, Director of Graphics at National Geographic. In the first map, each dot represents a specific number of people—2 million— living in a country. In the second map, each dot represents $20 billion in GDP for that country, as a proxy for consumption. The dramatic, yet entirely accurate data representation allows two different stories to emerge nicely as we see the relative weight of each country for that particular dataset. That’s the power of cartograms (the black circles here represent the same variables a few decades back, for a nice representation of growth over time).

You may say “ok, these look a little weird, why not represent these data in a choropleth map, or in other words using different shades of color in a “normal” map?”. Like the example below. Choroplet maps are ubiquitous and, while they are not incorrect, they are seen by many as misleading. They overrepresent the importance or large areas, and diminish the importance of small ones, just by virtue of their size in the page. If I plot GDP for different countries, for example, a large country with a medium value (say Brazil) “lights up” much more prominently than a country with a higher value such as Singapore, simply because Singapore is tiny and hard to see in the map. See a nice video from Vox here explaining the problem with choropleth maps.

One solution is to give all entities equal area and then use the choropleth technique (different shades of color represent different values). That’s more fair. But we can only show a few classes or “buckets” of data ranges, rather than precise values in each area.

So it may be interesting to resize each area according to the number we want to represent. Some cartograms can be fairly abstract. Instead of using the approximate shapes of countries or states, they use squares, circles or other geometric shapes to represent quantities in highly stylized maps. See this map we did for Fortune, with the number of millionaire households by state in the U.S. (color here represents a second variable, the median household income).

I love this type of cartogram. Here is another one by John Tomanio, a few years ago, for Fortune.

Cartograms can be really striking and powerful. Lately we have seen a comeback, and they were often used by different news organizations in the UK in the recent election cycles. A cartogram of the U.S. can be a nice way to represent that a small state in the East Coast may have many more electoral votes in the general election than a sparsely populated but large Midwest state. Here is the tilegram showing how much each state actually matters (how many electoral votes it holds) in the U.S. Much more telling than showing actual size in the context of election stories! Don’t be misled by how much blue or red you see in typical election maps.

Complex cartograms have been done manually forever, for example in Adobe Illustrator, as in the case of the National Geographic maps. It’s a painstaking process, but human judgement is critical to make sure the composites of little shapes will still resemble the actual map of the area, which is entirely the point if we want to avoid confusion. You have to put them together as a jigsaw puzzle.

We have seen tools that create cartograms working in conjunction with other GIS software tools, but if you ask me, they produce grotesquely deformed maps that are a hard to decipher mess (with that strange fisheye lens effect. You have probably seen them around). I think these are useless:

Tilegrams is a really nice concept, still evolving. It plots U.S. maps using hexagon shapes, which are more versatile than squares to assemble them together in adjoining shapes with different configurations. You can use some preloaded maps (electoral votes, population, etc) or load your own data. Tilegrams allows you to show different levels of resolution. For example, the first map below shows U.S. population with 4 million people represented in each tile. The one below shows 300,000 people per tile. As you would expect, the higher resolution allows us to resemble the shape of the U.S. better (but remember states are meant to be oversized or undersized to represent the values, so the unusual proportions are to be expected.

If the map starts looking a bit too weird you can drag hexagons to move them around. The tool tells the user about the statistical accuracy of each region’s surface area, as it tries to be both accurate and resemble the actual map in a difficult trade-off. You can export the map as a TopoJSON or as a SVG file, a wonderful option since it’s fully editable in Illustrator!

Pitch Interactive explains the tool here, and it continues to develop it. The company is now working on a U.S. county map. It sounds like a daunting task (there are over 3,000 counties in the U.S.). And hopefully one day we can show any country and their divisions in this simple way. I can’t wait to see what they come up with!