Nice first pass, but its clearly calculating these as the pigeon flies and not taking the various stitched together street grids into account. You should get diamond and square shapes, not circles, for the most part. I'm working on a redesigned subway map / nomenclature an indicators of walking time to nearest stations is a subtle feature I want to build in. Something between a watershed like this and a vornoi diagram depending on which edge comes first.
It's all bravado commenting on something you're working on in the comments of someone else's actual working site. Let's see the comments from a link to your site? Thought not
No, it is clearly not as the pigeon flies. That is incorrect.
When I click on my local subway stop it seems entirely accurate, and the stations don't form any kind of circle at all. In fact I can't find any subway stop origin that seems to form anything even close to a circle.
Also, according to the "about" it's calculating from actual subway schedules.
And it has nothing to do with street grids. It's subway trips, not walking.
The station-to-station part does indeed use subway travel times, but the website says "Hover over a station to see how much of the city is accessible within 40 minutes by subway and walking."
And the "walking" part clearly shows a perfect circle around each destination station, so it is 100% doing an as-the-pigeon-flies estimation for the walking.
"Isochrones are manually calculated using turf.js assuming 1.2m/s walking speed after the subway trip. These are simple buffers around each station/prior isochrone and do not take the street network into account."
Other people already corrected you, but I feel like being obtuse so here it goes. The circle goes from Jamaica, along Queens Blvd, down 8th Ave to W 4th St, along the F line to Jay St, down Fulton St to Broadway Junction, past Cypress hills on the Nassau line (J/Z), and finally returns to Jamaica. This loop is possible without adding any additional tracks or switches to the system. Thats how the pigeon flies. Checkmate.
It's using some hybrid of isochrones & walking time "as the crow flies." The subway travel is in isochrones, but once you exit the subway the walking time assumes walking in straight lines with no obstructions in any direction. So yes that's inaccurate but it's relevant that only the walking time is done so. I assume subway ride time is based on subway schedule times.
I did this exact thing in 1st year geography. Gotta use the network analysis tool, not spatial analysis. Distance along road segments, not the Euclidean distance tool.
You can take a shortcut, because NYC is several large grids stitched together. If your isochron line doesn't cross a border between two grids its just a diamond instead of a circle. If it does cross the border between two grids, then you just have to figure out where along the border the "diamond" of one grid system meets the tilted diamond of the other. But you don't need a general purpose road network analysis tool to do it. It may even be tractable analytically.
Why would you go to subawaysheds.com to learn about taxicab stands? that would obviously be taxicabstands.com.
Why just NYC? Because does anything else exist outside of NYC? Maybe atlantic city, or the catskills, or niagra falls, but everything else is just unimportant
As a New Yorker: 40 minutes is roughly my boundary between "ordinary trip with no prior thought" and "I'll set aside a dedicated time and/or day to make this trip."
The city's bus lines are much more variable in service, and don't offer as many interborough routes.
I had the same surprise. Which raises an interesting question: does that count as "breaking the back button"?
Theoretically, no: the back button is working as expected. At the same time, it was a significantly worse experience for me because it took a lot longer to go back to the "parent" page (HN).
It seems to make the assumption that you’ll get any particular train instantly. Factor in expected wait time for each line and the local/express variants and the sheds will shrink substantially.
it’ll also get confusing because the sheds will change as a function of time of day
It’s still interesting/useful, though. Once you require factoring in average train wait times and stuff like that the project is going to be way too complex to ever be achievable.
for sure, it's still useful, but i think it might be worth some sort of down-weighting for trains that run less frequently.
for example, if the ratio of local:express trains is 5:1, the watershed should be more heavily influenced by what's accessible in 40 minutes via local.
this is also in part due to the way people use the subway in NY -- nobody looks at a timetable and says "i'll head down to the station to catch the 6:37 express" like you would in some cities with reliable timetables/schedules. in NY, you just go to the station and wait and hope the train you want shows up next/soon.
> this is also in part due to the way people use the subway in NY -- nobody looks at a timetable and says "i'll head down to the station to catch the 6:37 express" like you would in some cities with reliable timetables/schedules. in NY, you just go to the station and wait and hope the train you want shows up next/soon.
That was true 15 years ago, but it's less true today now that there's a multitude of apps (both the first-party MyMTA app and many third-party apps which are much better) which make this easy.
Even Google Maps gives train times in their directions, and they're mostly accurately updated as delays happen, etc.
I've used the arrival time on Google maps and it seems accurate enough but the A train run so often it's not really an issue so I rarely if ever look. Plus I use mta.info for updates to check for delays.
That might not be an entirely wrong assumption to make. An office worker's working hours usually has some amount of slack, and people are willing to adjust their schedule to minimize wait time. No point waiting 25 minutes every day when you can get up 5 minutes earlier and wait basically zero.
The subway doesn’t keep to a strict schedule (it has one in theory but no one would depend on it), and has anywhere between 3 and 20 minute headways (time between trains).
You can sort of kind of get close to 1-2 minutes of slack if you have an app telling you how far away the subway is (CityMapper has this feature), but you can’t really time your sleep or known work hours for making a specific subway.
I mean super-digital people really do. If your train only comes every 20 min, you absolutely have your eye on the app and change the speed you eat breakfast in the morning, or decide you can reply to 4 more emails before leaving work.
There are tons of apps with live train countdowns. They're probably the biggest improvement to NYers quality of life in the past 30 years honestly, with ridesharing in second place.
Schedules are tough for a transit system to keep. Plus you aren't guaranteed to have the same walk to the station every day. Factor in maybe a couple light cycles of signals you need to wait for as a pedestrian to get down to the station, and your travel time on foot to the station can vary by a significant proportion, adding one or a couple three minute light cycles into the mix that might or might not hit for you. I try and anticipate getting to the platform at least 10 minutes before the scheduled train as a result. If I am closer to 5 minutes then I am rushing, feeling late, and stressed, and have missed plenty of trains in moments in this situation before.
Transfers take a very relevant amount of time, even during rush hour when the trains are running frequently. Aside from waiting on the train, which does not keep a strict schedule and can get delayed for many reasons there is the walking time to the other train, which can be a decent amount even in locations where you don't have to exit the subway to transfer. Factoring a minimum 5-10 minutes extra time per transfer is a safe bet.
You misunderstand. It's not about one seat rides or about schedules. Nobody knows the schedule in NY, the train is frequent enough to not care. But changing lines is basically a statistical 5 min of travel time each time you do it.
Since it's Mapbox I wonder if they are using the Isochrone API[1], or if this is more custom b/c it's subways? I agree an adjustable duration of trip would be neat.
I'm especially impressed by how it handles the local and express routes: North Brooklyn is only about 15 minutes door-to-door from Lower Manhattan on the Fulton Line[1] express A train, and the map correctly shows that. Nice work!
I'm not from NYC, so maybe this is obvious, but why can you not get from Staten Island to Manhattan in 40mins? Is there really not bridge/tunnel that can get you there?
There is no direct bridge or tunnel between Manhattan and Staten Island, only a ferry. Staten Island is connected to Brooklyn via the Verrazano Bridge.
The only bridges approaching that length over water are causeways. The harbor is too deep for that, and again a causeway would not permit shipping traffic.
On top of what 'sempron64 said: there was a plan to construct a subway tunnel between Brooklyn and Staten Island[1], but construction was canceled nearly a century ago.
The Verrazano Bridge itself was designed under Robert Moses's eye, and he shut down the idea of a subway link early on[2].
It’s kind of a political hot potato few want to touch these days. A tunnel feels like a no brainier but a lot of NIMBY types in Staten Island don’t want a fast connection to the rest of the city as it would likely change the area dramatically.
It's also very across the harbor between Staten Island and Manhattan. Any bridge or tunnel would be multiple times longer than any other bridge or tunnel in the area (and correspondingly expensive).
The original plans were for a tunnel between SI and Brooklyn. Still probably the most realistic of any plan though it does lose the benefits of a direct connection to Manhattan.
We should be asking “why not both?” but unfortunately NY construction costs are inflated so ridiculously we can’t even afford one let alone get politicians with a strong enough backbone to support these common sense projects that would alleviate strain on NYC housing stock.
The Verrazano bridge is usually baked up with traffic as is the two major highways feeding into it: Brooklyn-Queens Expressway (BQE) and the Belt Parkway. To get from SI to Manhattan by car you take the bridge to the BQE then battery tunnel into Manhattan but that's over and hour in traffic. Ferry is another option. There's no connecting subway tunnel and a single small train line runs north south to the east. Weird borough - feels like long island or upstate. Even had a shanty ghost town to the south.
Staten Island is south of Brooklyn, as others have noted, and a good distance away — so even if we had a rail link via bridge or tunnel to Brooklyn, you’d need more work to bring it to Manhattan. You might think of using existing lines, but the closest credible line, the R train, has capacity for adding such trains, and for taking those trains into Manhattan… but it runs local along Brooklyn’s 4th Ave, making the overall trip much slower. It does join up with the D/N lines after a short distance (4 Ave Express), but the D/N has capacity limitations both at that point and later when they join the B/Q and cross the Manhattan Bridge to the 6th Ave Express/Broadway Express line. You could displace either of those services to the local track and send them into lower Manhattan via the Montague Tube link, at the cost of making trips down to Coney Island / Stillwell Ave much slower and serving areas in lower Manhattan which are less relevant than midtown.
There is some remote possibility of going out of the way to link to the F train, which even has an essentially-unused express track (portions of the system are 3-track but more of the relevant route is a full 4-track). The big problem with this is that it eventually dumps you onto the 6 Ave Local, and 6 Ave is pretty much full, local or otherwise. The only other option is the IND Crosstown, which goes to Bed-Stuy, Williamsburg, Greenpoint and Queens instead of Manhattan. The smaller problem is that the express train stops here are awkwardly located, and the potential transfer points for other lines tend to be crowded.
In short, all the good subway capacity into Manhattan is already claimed by existing services. This makes sense! The best ones are all full and the ones with spare capacity are less popular for good reason. It’d be a waste otherwise.
If you’re looking to build more capacity, you’ll need something dramatic like connecting the Second Avenue Subway. At that point I’d want to hook it up to the B/D tracks via the Rutgers Tube, send the D in Manhattan onto the new line, and divert the D in Brooklyn via the F express tracks, instead of the Brooklyn 4 Ave Express. (You could trade the B line with some clever swaps near Chrystie as an alternative.) This would free up capacity for the Staten Island service on the 4 Ave link, a much better proposition.
But given that building the existing Phase 1 mile-and-a-bit of the Second Avenue Subway ran billions of dollars and took over 70 years to happen… you may see the size of the problem you face adding many new miles in Phase 3 to connect it to the area near the Christie St Connection.
Now the pandemic might have mitigated some of this, but it’s also mitigated the benefit of the scheme in the first place, and it has strained the budgets for such projects too.
very cool. but I noticed some of the stop information is incorrect.
for example, the franklin stop on the C train in Brooklyn, isn't connected to the franklin stop on the shuttle.
Is this true? I haven't taken the Franklin Ave shuttle in a decade, but I remember it being on the top floor of the Franklin Ave station (with the C platform being in the basement).
I’ve taken that connection so I assure you it is! The S train is elevated and quite a walk from the C platform but they are accessible from each other (including via elevator, which is a rarity)
Edit: in the map they aren't connected. It shows them as two separate stations. Im arguing that they are the same station. And one can transfer easily from the "C" to the "Shuttle" and vice versa.
WNYC had a cool one a few years ago that also included time to get to/from stations but it seems their MapBox account isn't working anymore:
https://project.wnyc.org/transit-time/
This is pretty cool and well done, but perhaps a bit optimistic on the time estimates. Starting from my station, there's really no way to get to Brooklyn in 40 min. The timing seems off by 25% or so
Interesting, and it's cool to see an illustration of how weird it is that there's no quick way to cross the park! It takes the same ballpark amount of time to get from east 86th St & 2nd to Kew Gardens (~14 miles apart) as it does to get from east 86th & 2nd to west 86th St & Broadway (~1.5 miles apart)...
This kind of tool would be absolutely brilliant as part of Zillow.
I'd love to see various distances from properties assuming bike, or walking, or driving, etc.
Everyone who lives in a neighborhood knows these things, but it's really hard to figure it out before without traveling there and attempting some things.
Zillow already has that. You input an address and it tells you the distance in time by transportation method on each property details screen. You can also sort properties according to distance from the given address. The software labels it a "work" address, but you will face no consequences for using an address that you do not work at.
The problem is that this shows you how long it takes to get to a single place, rather than showing a map visualization like this site. I've often really wanted this on StreetEasy especially. I don't want to know how long it takes just to get to work, I want to know how long it takes to every neighborhood I regularly travel to.
Agree it could be better. I recently moved and wanted to be close to work but also close to a few friends. It would be cool if you could put in two (or more) locations and it would map out areas. I'm sure some ones would have popped up I didnt consider.
I've sort of done this using the Walk score transit heat maps (they have similar for biking and walking as well) for getting a more broad feel for what areas are walkable/bikeable/less car dependant while house shopping, using a user script that takes the map tiles from Walkscore and injects them into the Google maps instance on Zillow. However, you could definitely generate the distance from a point heat/contour maps using QGIS/OSM data and overlay the tiles that way as well.
Would be super nice if this was built into Zillow, or any MLS browsing tool for that matter...
With the exception of three stops on the UWS that took about a century to build[0] and one additional stop on the 7, ~all of the subway stops were built 80+ years ago, via one of three independent private train companies. Those three companies eventually went bankrupt and were taken over by the city, and then later the state.
So I wouldn't call anything a "design decision" in this case, insofar as it was never designed intentionally to be a comprehensive public transit system the way we think of it today.
[0] Not an exaggeration - Phase 1 was originally supposed to be completed before WWII.
It would be fantastic if Google Maps and the like could take this into account when searching instead of just naively relying on distance. 2 miles away along a well-developed transit corridor is much "closer" to me than 2 miles away in a direction with no bus or rail options.
Sometimes I'd be happy to go halfway across town if I can just take a single train there and it's close by the station. Especially if it involves picking up packages or heavily items.
Google maps is pretty basic for its routing. For example, I use a skateboard to commute to transit stops. It effectively cuts walk times in half or more and is easy to carry on a crowded train. There is no way to set this average walking/skating speed, however, so the routes they choose are not optimal and require my own localized knowledge of the transit network and schedule, and knowing the best routes for skating fast. This doesn't just affect skaters, but people who might have a bike or scooter they can use between transit trips. You can imagine all the much more optimal routings that become possible when you can suddenly move 10-15mph on a whim between point A and B.
Definitely. Offline maps are great but the only offline routing option is by car. If you want to walk, you have to figure it out yourself based on the driving directions.
I can see from a business point of view that it might make sense to optimize for drivers, but it's irresponsible of them to ignore the impact they have on users. They're creating a self-perpetuating cycle.
Google only draws on the offline maps for me when I shut off my data. If I have a spotty connection, google maps will try and pull fresh data including the road mapping. Shut off LTE and suddenly the app works instantly.
Apple didn’t exactly do a speedy rollout of bike directions in their Maps app. I find it hard to blame Google here. Seems more like a symptom of our society to me.
The best transit app I use is this ancient mobile website from my local transit agency. It just has a live map of a chose route and shows where the busses are, and it looks like its some original mobile web app type of stuff like from 15 years ago. What makes it great is how lightweight it is. If I have a spotty data connection or I want to preserve battery, I know I can rely on it a lot more than google maps or other bloated transit apps.
114 comments
[ 0.15 ms ] story [ 171 ms ] threadIt's all bravado commenting on something you're working on in the comments of someone else's actual working site. Let's see the comments from a link to your site? Thought not
When I click on my local subway stop it seems entirely accurate, and the stations don't form any kind of circle at all. In fact I can't find any subway stop origin that seems to form anything even close to a circle.
Also, according to the "about" it's calculating from actual subway schedules.
And it has nothing to do with street grids. It's subway trips, not walking.
The site says "accessible within 40 minutes by subway and walking"
The station-to-station part does indeed use subway travel times, but the website says "Hover over a station to see how much of the city is accessible within 40 minutes by subway and walking."
And the "walking" part clearly shows a perfect circle around each destination station, so it is 100% doing an as-the-pigeon-flies estimation for the walking.
Click on Rockaway park. A pigeon could reach those islands, but a walking person could not.
This is still a great visualization.
"Isochrones are manually calculated using turf.js assuming 1.2m/s walking speed after the subway trip. These are simple buffers around each station/prior isochrone and do not take the street network into account."
“Interesting niche site about birds in the authors backyard. But why doesn’t it also talk about the trees in my backyard?”
Why just NYC? Because does anything else exist outside of NYC? Maybe atlantic city, or the catskills, or niagra falls, but everything else is just unimportant
The city's bus lines are much more variable in service, and don't offer as many interborough routes.
One thing surprised me after I clicked around, I did not expect each station to become a separate browser history entry.
Theoretically, no: the back button is working as expected. At the same time, it was a significantly worse experience for me because it took a lot longer to go back to the "parent" page (HN).
it’ll also get confusing because the sheds will change as a function of time of day
for example, if the ratio of local:express trains is 5:1, the watershed should be more heavily influenced by what's accessible in 40 minutes via local.
this is also in part due to the way people use the subway in NY -- nobody looks at a timetable and says "i'll head down to the station to catch the 6:37 express" like you would in some cities with reliable timetables/schedules. in NY, you just go to the station and wait and hope the train you want shows up next/soon.
That was true 15 years ago, but it's less true today now that there's a multitude of apps (both the first-party MyMTA app and many third-party apps which are much better) which make this easy.
Even Google Maps gives train times in their directions, and they're mostly accurately updated as delays happen, etc.
You can sort of kind of get close to 1-2 minutes of slack if you have an app telling you how far away the subway is (CityMapper has this feature), but you can’t really time your sleep or known work hours for making a specific subway.
There are tons of apps with live train countdowns. They're probably the biggest improvement to NYers quality of life in the past 30 years honestly, with ridesharing in second place.
[1] - https://docs.mapbox.com/api/navigation/isochrone/
If they were using the Isochrone API, the buffers around the subway stations would take in account the street grid vs a circle.
That's really cool that Mapbox now offers an API to do this. It seems like it also factors in elevation as well!
I'm especially impressed by how it handles the local and express routes: North Brooklyn is only about 15 minutes door-to-door from Lower Manhattan on the Fulton Line[1] express A train, and the map correctly shows that. Nice work!
[1]: https://en.wikipedia.org/wiki/IND_Fulton_Street_Line
This feels like a 10th percentile stat. Change the 10-40m range to be 15-55m and it will be closer to what I experience on a daily basis.
Great resource for visitors to NYC to set expectations.
Bridges can also be tall, if the concern is cutting boat traffic.
It would also be a massive eyesore
The Verrazano Bridge itself was designed under Robert Moses's eye, and he shut down the idea of a subway link early on[2].
[1]: https://en.wikipedia.org/wiki/Staten_Island_Tunnel
[2]: https://www.nytimes.com/1956/04/09/archives/moses-bars-train...
It’s kind of a political hot potato few want to touch these days. A tunnel feels like a no brainier but a lot of NIMBY types in Staten Island don’t want a fast connection to the rest of the city as it would likely change the area dramatically.
There is some remote possibility of going out of the way to link to the F train, which even has an essentially-unused express track (portions of the system are 3-track but more of the relevant route is a full 4-track). The big problem with this is that it eventually dumps you onto the 6 Ave Local, and 6 Ave is pretty much full, local or otherwise. The only other option is the IND Crosstown, which goes to Bed-Stuy, Williamsburg, Greenpoint and Queens instead of Manhattan. The smaller problem is that the express train stops here are awkwardly located, and the potential transfer points for other lines tend to be crowded.
In short, all the good subway capacity into Manhattan is already claimed by existing services. This makes sense! The best ones are all full and the ones with spare capacity are less popular for good reason. It’d be a waste otherwise.
If you’re looking to build more capacity, you’ll need something dramatic like connecting the Second Avenue Subway. At that point I’d want to hook it up to the B/D tracks via the Rutgers Tube, send the D in Manhattan onto the new line, and divert the D in Brooklyn via the F express tracks, instead of the Brooklyn 4 Ave Express. (You could trade the B line with some clever swaps near Chrystie as an alternative.) This would free up capacity for the Staten Island service on the 4 Ave link, a much better proposition.
But given that building the existing Phase 1 mile-and-a-bit of the Second Avenue Subway ran billions of dollars and took over 70 years to happen… you may see the size of the problem you face adding many new miles in Phase 3 to connect it to the area near the Christie St Connection.
Now the pandemic might have mitigated some of this, but it’s also mitigated the benefit of the scheme in the first place, and it has strained the budgets for such projects too.
(I live like 2 blocks from that station)
You can see what it looked like here https://gothamist.com/news/map-find-out-how-long-it-takes-to...
Also it's not clear if it's counting waiting time when transferring, which would make a big difference.
I'd love to see various distances from properties assuming bike, or walking, or driving, etc.
Everyone who lives in a neighborhood knows these things, but it's really hard to figure it out before without traveling there and attempting some things.
I want to know all the places I don’t know about that are X minutes away.
Would be super nice if this was built into Zillow, or any MLS browsing tool for that matter...
So I wouldn't call anything a "design decision" in this case, insofar as it was never designed intentionally to be a comprehensive public transit system the way we think of it today.
[0] Not an exaggeration - Phase 1 was originally supposed to be completed before WWII.
It's about time, super useful!
Sometimes I'd be happy to go halfway across town if I can just take a single train there and it's close by the station. Especially if it involves picking up packages or heavily items.
I can see from a business point of view that it might make sense to optimize for drivers, but it's irresponsible of them to ignore the impact they have on users. They're creating a self-perpetuating cycle.
Times Square [0] has the most connections, but Atlantic Av [1] has better coverage of Brooklyn. Hard to tell which is larger, though.
0: https://subwaysheds.com/stop/r16/times-sq-42-st
1: https://subwaysheds.com/stop/r31/atlantic-av-barclays-ctr