Check out how well connected Vermonters are to other major ski destinations nationwide. What a neat way of viewing subcultures and travel.

Too bad the data is US-specific. If you could add Canada, you'd see the reality of "Cascadia" -- the empire west of the Rockies that shares natural resources, agriculture, culture, ethnicity, and politics. British Columbia and Washington State are thinking out loud of high-speed train service to link Vancouver and Seattle. We already share coastal protection with the US Coast Guard frequently providing assistance to distressed boats in Canadian waters. And legalized marijuana, along with great vineyards extending northward with every summer, will make Cascadia a very rich region indeed.

Notice how people in NY and CA are not connected to anyone.

In virtually every map in this article, Loving County, Texas shows up dark? It's the exception to all the rules.

I recall from the Johns Hopkins Beginning School Study that Baltimore kids would be sent to the relatives in rural West Virginia to “cool off” when they started to get into trouble in the city. Facebook can be a great aid to Sociology researchers. Graphing databases will be very useful, I think, for tracking drug, gang, and gun flows, disease vectors and other issues connected to human connectivity.

My community is in the United States.

Fascinating! Many of these maps show the old geographic law that “all things are connected, but nearer things are more connected”. Another geographical insight into human connectivity (taken from physics!) is the gravity model. Large objects (in this case, cities) have a greater gravity, and gravity decreases with distance. So: when we study migration, bigger cities generally have the ability to “pull” migrants from greater distances. Migration explains some Facebook interaction patterns, as people stay connected to family and friends left behind. Not surprising to see that large, adjacent cities should have more social interaction than small, distant towns.

Very cool data visualization! The NYT must have an awesome team working on this stuff.

Thanks for publishing this. Could the authors explain what algorithm was used for the "This is what happens when you split the country in N parts"? Thanks

"All politics is local." (widely attributed to Tip O'Neill.

One interesting aspect is the State Line Effect: if you pick a county in the middle of many states, such as Iowa, Oklahoma, and Alabama, you can see the state outlines in darker blue. My guess is the reason there is a sharp dropoff in Facebook friends when you cross out of state is due to state colleges. In state tuition is typically much lower than out of state tuition, so somebody who lives near a state border will likely have more college friends from within his state than just over the border. Plus, rooting for sports teams from your state can be a big connection.

Dear NYTimes: You ask a good question: What is Upstate New York? I'd like to know how you define it, because most journalists define it as anything outside NYC, which is wrong. And something that hits a nerve for those of us who grew up in WESTERN New York and went to grad school in Central New York.

Check out the wide range of multiple contacts from Loving County, TX! (West Texas, south of the NM state line). Wiki says that it has a population of 134 people, but it is connected widely across the west and south with outlier high connection places disbursed across places like the upper Midwest and Florida. Must be a few lonely people Facebooking like mad!

Interesting that Manhattan has a spread out network but they are all other wealthy parts of the country so there's no diversity at all.

Does think include orgs? I'm not seeing how Teton County in WY is heavily linked across the US unless it is links to parks.

Anyone have an explanation for Onslow County, N.C.?

What a great study with great new data. The NYT also did a great service by presenting this research in an accessible, interactive way so the public can learn about it, and reflect on what it means to them. These big data-driven studies face a big challenge. When presented with nice, novel data set like FB data, researchers successfully apply their strong quantitative skillsets to mine the data, and generate the research questions based on common sense and their own experiences (and here, an apt nod to community sociology lit). A third thing does not happen: the refining of existing ideas. Domain specialists call the nearness of friendship “propinquity”. The trends revealed here are known as “distance decay” and/or the “first law of geography”. The connectedness that ties distant places together through relationships (Which this data illustrates! Exciting!) is called “social distance” (or more recently, “logical distance” or “effective distance”). These theories may seem dusty, but they are more valuable today than ever. Still, it seems hard to communicate that they exist across disciplines—here, to finance and economics researchers. Thanks to the study authors, the data providers and NYT for sharing a new way to quantify human “extensibility” (another great communications geography term that measures our social and spatial ‘reach’) and hold a mirror to our nation. The articulate, timely findings will resonate across disciplines for researchers and practitioners.

This is such a deep analysis, and the visualizations provide so much insight! ND, SD and counties with large university populations (Tompkins, NY; Tuscaloosa, AL) are interesting to see. Something with tSNE or UMAP that makes geographical position/distance the variable that changes, while increasing numbers of clusters, could also be fun.

What mapping software is being used for your visualizations?

Interesting that Colorado has notably high relationships with a lot of counties as I scanned over.

But...what if I'm not on Facebook? I don't count?

Wow - I find it interesting that a bunch of economists just discovered something that geographer's have known for quite a while. It is one of the few 'Laws of Geography' known as 'Tobler's Law' after Waldo Tobler (https://en.wikipedia.org/wiki/Waldo_R._Tobler ) - Tobler's law states: "everything is related to everything else, but near things are more related than distant things." Perhaps these economists might learn some more geography and psychology and figure out that people are not rational, we all don't have the same quantity of information, economic inequality actually has negative consequences on economic efficiency, and market failures like our failure to value our fundamental dependence on functioning ecosystems here on earth is killing us - literally.

I think that the locations of Facebook friends is a complicated and flawed proxy for "connectedness" in this analysis, especially with respect to how both Facebook as a technology and "connectedness" as a term are used. Most people that I know use Facebook friending for people that they have met in person -- it's not surprising that this methodology, which inscribes physical contact onto a digital space, would produce data that appears to be especially regional. Compare that to something like Twitter, or digital message boards like Reddit, where friendships and information exchange happen less around digitized face-to-face communication and more around subject affinities. "Connectedness," in how at least I have seen it used in news articles, punditry, and casual conversation, more often suggests the latter form of connection. And that's the point: many networked information systems allow people to connect by subject rather than by space -- that's their radically new capability. This is a really interesting data set, but I think that the chosen language to describe it mis-locates the analysis's implications within the current dialogue about public media.

I would have liked to have seen some type of overlap of area codes. I am motivated to this thought by seeing a parsing at 200 parts that I began to see parts of upstate NY essentially coincide with upstate area codes and what New York State's Tourism Office says are the regions of New York State. "The 518" / "Capital-Saratoga" is a good example of that. I lived for most of 30 years up there and actually felt that the parsing shown is this article was a better representation of "The Capital Distict" than either the 518 area code or the the State's "Capital-Saratoga" region.

Interesting. I come from a county adjacent to Clinton in Michigan, but have lived the majority of my adult life in Illinois. I've often noted that Michigan had more of a cohesive sense of place than Illinois, which I ascribed to it's particular geography, surrounded as it is by the Great Lakes. But I'm really surprised at the social connections shown between the Upper Peninsula and the counties bordering Ohio and Indiana; from Temperance to Ironwood, Michigan is a ten-hour drive. Both towns are a couple of minutes from neighboring states. Proximity doesn't seem to matter much in that example.

This data is also a good measure to track and measure where everyone is migrating and transplanting around to which is what this article really should've emphasized. It seems that many are fleeing California in droves for the cheaper, surrounding states, or for more exciting locales as Austin, Brooklyn, Manhattan, and Washington DC have a high amount of people from the SF and LA areas. Lots of people are moving to the Alaskan, Bakken, Oklahoma, and Texan oil fields for work. Atlanta, Austin, DC, NYC seem to the most popular cities in the country. The Sun Belt continues to attract lots of people.

Where I live in New England, people don't even socialize or know those who exist one small town over, mainly due to extreme provincialism. It's really so weird.

Can you try to redraw congress map with more or less equal population with this method? I think it would extremely interesting.

fascinating. thanks for putting this together.

So Mark Zuckerberg's claim that Facebook is connecting us all is false? Gee - imagine that.

So, America is as provincial as ever.

Interesting data. We live in a rural flyover county. By necessity of work as well as a widely distributed family and friends. a vast majority of our contacts are far away. What really makes the difference is the rural telephone cooperatives in both counties where I have property have been leaders and early adopters in high speed internet installation as opposed to much of rural america which does not have it.

fascinating, though data regarding very rural areas may not be as reliable since "100X" in a very sparsely populated area does not mean the same thing as data from densely populated areas. Northern Maine is an obvious example. A great article.

As with any "Big Data" study, there is a tendency to over-emphasize pretty graphics and assert that this is deeper, more thorough, and somehow better. I'd offer for anyone to examine Loving County, TX (SW, near El Paso). The data shows that over 40% of friends are more than 500 miles from this location, and scanning over all counties (just drag your mouse), you will see that this county is dark blue for many places in the USA. What is so special about this place? There are no military bases or prisons there, and a net population of around 100 persons.