Here is an interesting study that looked at the impact of urban highways on social connections within the 50 largest US cities. To measure this, the researchers used Twitter data from 2012-2013, which is a period of time where the default setting in the mobile app was to tag each tweet with the user's precise geographic coordinates.

This allowed the team to generally figure out where a user is likely to live. If you're often tweeting from the same residential address, then there's a good chance that's home. They then looked at things like mutual followship as a measure of social ties. And what they ultimately found was that in all 50 cities, urban highways exhibit a strong barrier effect. They measured this using something they call a "barrier score."

Now this sounds right and supports lots of other evidence that highways divide cities; but Twitter isn't necessarily a place where mutual followship means you actually know the person in real life and you regularly walk down the street to see if they can come out and play. So one of the things that the researchers also did was work to replicate their findings using data from another social network called Gowalla.

I very vaguely remember this platform, but it is/was a social network where users are supposed to connect with people they actually know and share their locations through check-ins. With this data they found that their "barrier score" was even more pronounced, which makes sense given that the platform's social graph should have had, in theory, stronger real-life ties.

But even if you don't believe the social data, these results should make intuitive sense. Highway underpasses and overpasses tend not to be the best environments for pedestrians. They're usually a clear break in a city's urban fabric, which can make people second guess whether they really want or need to cross it.

Cover photo by Tom Barrett on Unsplash

One of the interesting things about street networks is that -- once laid out -- they rarely change in a substantial way. Not unless Haussmann comes along and decides to renovate. That can be a good thing if you get it right like in the case of Philadelphia or Barcelona. But it can also be a bad thing if you're maybe wishing the founders went with design option B.

On that note, here's an interesting set of polar histograms from Geoff Boeing, who is a professor of urban planning and spatial analysis at the University of Southern California, that sorts city street networks by most-ordered/gridded to most-disordered:

What these diagrams don't tell you is how walkable or dense a city might be. But they do tell you how many of its streets follow a strict orthogonal grid -- see Chicago and Beijing -- and how many of its streets are all over the place, like in the case of Rome and Charlotte.

They also tell you when the grid is askew, like in the case of Toronto (because of the angle of our shoreline) and in the case of Manhattan (because of the angle of the whole island). And if you were to revisit these diagrams well into the future, I suspect you wouldn't see much difference from what you see here.

We've spoken before about Saudi Arabia's "The Line" project. At first, I wasn't sure if it was real, but it is, and it's now under construction. We then spoke about whether a 170-kilometer line is an optimal urban form for a city, and the answer, according to this study, is that it's not. The problem with a line is that it actually maximizes the average distance between inhabitants. This makes sense because you could have two people living and working 170 kilometers apart.

On the other hand, if you maintain the same built-up area and take the opposite kind of geometry -- a circle -- you actually minimize the average distance between inhabitants. It's for this reason that older cities (the ones that weren't masterplanned) have tended to grow radially and not linearly (unless there were geographic features forcing it to grow in a certain way). So there is a strong argument to be made that The Line is a suboptimal plan for a new city.

But here's what's interesting: many cities already follow a somewhat similar approach. They don't do it as absolutely as The Line, but they do it in the way that they zone for higher densities and a mix of uses only on their main corridors. Example:

This creates a similar kind of effect when it comes to walkability, ability to support higher-order transit, and overall agglomeration economies. All of the urban activity gets concentrated along one corridor, maximizing the distance between people. In extreme examples, you also get inhabitants that are forced into different mobility options. The corridor is supposed to be transit-oriented, but all of the surrounding areas are really only conducive to driving. This creates a mismatch that is less an ideal for everyone.

So this post is our regular reminder that, when it comes to planning cities and bringing people together, circles tend to be better than lines. This doesn't necessarily mean that you need to adopt some sort of radial street network, à la French model. (Although I'm now thinking about the effects of this vs. an orthogonal grid.) It just means that urban density works a lot better when it's clustered, especially around transit. And generally, circles make for better clusters.

Here is an interesting study that looked at the impact of urban highways on social connections within the 50 largest US cities. To measure this, the researchers used Twitter data from 2012-2013, which is a period of time where the default setting in the mobile app was to tag each tweet with the user's precise geographic coordinates.

This allowed the team to generally figure out where a user is likely to live. If you're often tweeting from the same residential address, then there's a good chance that's home. They then looked at things like mutual followship as a measure of social ties. And what they ultimately found was that in all 50 cities, urban highways exhibit a strong barrier effect. They measured this using something they call a "barrier score."

Now this sounds right and supports lots of other evidence that highways divide cities; but Twitter isn't necessarily a place where mutual followship means you actually know the person in real life and you regularly walk down the street to see if they can come out and play. So one of the things that the researchers also did was work to replicate their findings using data from another social network called Gowalla.

I very vaguely remember this platform, but it is/was a social network where users are supposed to connect with people they actually know and share their locations through check-ins. With this data they found that their "barrier score" was even more pronounced, which makes sense given that the platform's social graph should have had, in theory, stronger real-life ties.

But even if you don't believe the social data, these results should make intuitive sense. Highway underpasses and overpasses tend not to be the best environments for pedestrians. They're usually a clear break in a city's urban fabric, which can make people second guess whether they really want or need to cross it.

Cover photo by Tom Barrett on Unsplash

One of the interesting things about street networks is that -- once laid out -- they rarely change in a substantial way. Not unless Haussmann comes along and decides to renovate. That can be a good thing if you get it right like in the case of Philadelphia or Barcelona. But it can also be a bad thing if you're maybe wishing the founders went with design option B.

On that note, here's an interesting set of polar histograms from Geoff Boeing, who is a professor of urban planning and spatial analysis at the University of Southern California, that sorts city street networks by most-ordered/gridded to most-disordered:

What these diagrams don't tell you is how walkable or dense a city might be. But they do tell you how many of its streets follow a strict orthogonal grid -- see Chicago and Beijing -- and how many of its streets are all over the place, like in the case of Rome and Charlotte.

They also tell you when the grid is askew, like in the case of Toronto (because of the angle of our shoreline) and in the case of Manhattan (because of the angle of the whole island). And if you were to revisit these diagrams well into the future, I suspect you wouldn't see much difference from what you see here.

We've spoken before about Saudi Arabia's "The Line" project. At first, I wasn't sure if it was real, but it is, and it's now under construction. We then spoke about whether a 170-kilometer line is an optimal urban form for a city, and the answer, according to this study, is that it's not. The problem with a line is that it actually maximizes the average distance between inhabitants. This makes sense because you could have two people living and working 170 kilometers apart.

On the other hand, if you maintain the same built-up area and take the opposite kind of geometry -- a circle -- you actually minimize the average distance between inhabitants. It's for this reason that older cities (the ones that weren't masterplanned) have tended to grow radially and not linearly (unless there were geographic features forcing it to grow in a certain way). So there is a strong argument to be made that The Line is a suboptimal plan for a new city.

But here's what's interesting: many cities already follow a somewhat similar approach. They don't do it as absolutely as The Line, but they do it in the way that they zone for higher densities and a mix of uses only on their main corridors. Example:

This creates a similar kind of effect when it comes to walkability, ability to support higher-order transit, and overall agglomeration economies. All of the urban activity gets concentrated along one corridor, maximizing the distance between people. In extreme examples, you also get inhabitants that are forced into different mobility options. The corridor is supposed to be transit-oriented, but all of the surrounding areas are really only conducive to driving. This creates a mismatch that is less an ideal for everyone.

So this post is our regular reminder that, when it comes to planning cities and bringing people together, circles tend to be better than lines. This doesn't necessarily mean that you need to adopt some sort of radial street network, à la French model. (Although I'm now thinking about the effects of this vs. an orthogonal grid.) It just means that urban density works a lot better when it's clustered, especially around transit. And generally, circles make for better clusters.