8 ms·

Transit starts crowd-sourcing real-time transit times as official feed breaks

219 points by ant6n 10y ago

coldshower 10y ago
This is a smart use of crowdsourcing. From the app you can know the number of people who are waiting for the bus and relying on your reporting. Kind of gives you a sense of community. I hope this takes off.
- arjie 10y ago
  Right. Their little smiley telling me that nine people are being helped by my activating their app kept me longer at the stop than otherwise. The bus was four minutes off their prediction, though.
  - samvermette 10y ago
    You're only going to get accurate departure times if someone else is using GO on that bus/train you're waiting for!
    - arjie 10y ago
      Oh, yes, I know. I hope the fellows downstream had better performance.
    - wikibob 10y ago
      What does it take to add real-time data for a new city? I've been interested in trying to scrape the data from the transit provider here. They have what appears to be reasonably accurate arrival predictions which they display through an atrocious app, and electronic signboards at some bus stops. The busses have a terminal for the driver that displays how long they need to dwell at times stops and announces the next stop, so I assume this is (at an unknown interval) transmitting the locations to their server. I haven't gone further in looking at it yet, because google maps will only add data sources directly from the official transit provider.
  - noobermin 10y ago
    I live in a major city of ~800K with one of the lowest public transit ridership in the US, and I've learned to not quite trust their time predictions. We do luckily have GPS location data, and I regularly just look at the bus' actual position to gauge time most of the time.
Terretta 10y ago
I love the competition between Transit and CityMapper. In cities with coverage, they're both so good it's stressful trying to choose! We are spoiled by both in NYC and surrounding counties, in Paris Transit is crushing it with Velib integration, while in Copenhagen CityMapper's local know how is amazing.
- tfigment 10y ago
  Is this the same company? They list only supporting 2 Canadian cities (Victoria, Montreal) in the top link.
  - samvermette 10y ago
    Those 2 cities is where we've been piloting crowdsourcing. The app itself works in 125 metro areas.
  - coldshower 10y ago
    Waterloo region (including Cambridge and Kitchener in Ontario) is supported, as is Guelph. These aren't big cities by any stretch. I see a part list here: https://transitapp.com/region https://transitapp.com/region and more cities listed here: https://play.google.com/store/apps/details?id=com.thetransitapp.droid https://play.google.com/store/apps/details?id=com.thetransit...
Animats 10y ago
NextBus is an old system that uses 2G cellular links. It doesn't need much data, so that's fine. Now 2G service is being discontinued by AT&T in SF, and Muni has to switch. Or switch to T-Mobile, which is offering 2G service for another year. This was an early "Internet of Things" project. I met the inventor once at the Hacker Conference. This is a general problem with "Internet of Things" products - you're dependent on infrastructure which might go away.
- Johnny555 10y ago
  There's no excuse for this happening -- AT&T announced the end of life of 2G four years ago in 2012. Now Muni claims that the shutdown took them by surprise and they are scrambling to replace all of the old 2G equipment https://www.sfmta.com/about-sfmta/blog/why-muni-arrival-times-are-off-week-and-how-we%E2%80%99re-working-fix-them https://www.sfmta.com/about-sfmta/blog/why-muni-arrival-time... Simply put, the deactivation work that affects our vehicles started sooner than expected and outpaced our ongoing upgrade of all Muni vehicles to a new communications and monitoring system. http://usatoday30.usatoday.com/tech/news/story/2012-08-03/att-2g-network/56758432/1 http://usatoday30.usatoday.com/tech/news/story/2012-08-03/at... AT&T Friday [announced] that the shutdown of its second-generation, or "2G," wireless network will be complete by the end of 2016 At the time, I was managing some devices that used 2G to send telemetry -- our AT&T sales rep called us to warn us about the shutdown.
  - xenadu02 10y ago
    I leased an Infinity over a year ago... it has the same "problem": the connected features rely on an ATT 2G radio. They made me sign a form acknowledging it would stop working January 1, 2017. There is zero chance Muni wasn't aware of this. I have no idea why they didn't ask for money to upgrade the system earlier... it seems like a tailor-made situation to get a budget increase: legitimately blame it on someone else and use the largesse to buy a brand new much better system that you normally can't get approved.
    - pnathan 10y ago
eli 10y ago
Neat. I had a similar idea a few years ago to make NextBus times more accurate... but fell down a rabbit hole trying to implement it as a cross-platform HTML app and never actually launched. If anyone wants my idea: My perception of NextBus's arrival times are that they're based on a very naive algorithm. I think it might literally be assuming that all buses always move at an average 15 mph. Lots of ways to make that more accurate with collecting just a teensy bit of actual arrival time data.
- stuckagain 10y ago
  I don't know about other systems but I've spent time in the field analyzing NextMuni's predictions and they have no predictive power whatsoever. The first arrival time is weakly correlated with actual arrival, and the 2nd and 3rd times are not even correlated. It's useless, essentially.
  - danaliv 10y ago
    I learned early on to discard the units when using NextMuni. The J will arrive in three. Three what? Nobody knows, but there will be three of them.
- skuhn 10y ago
  Back when NextMuni was only 99% worthless, Transit had a method of actually getting reasonable arrival times out of it. Since each bus has a published GPS location (in addition to the NextMuni time), you can use that with driving direction data to loosely get travel time. When the app did this you would initially see Next Muni times -- 2 minutes away! -- which would update a few seconds later for lines you were particularly looking at -- oh wait 20 minutes away! Apparently NextMuni either just displays the schedule (which is literally never right) or does some kind of extremely basic estimation like you said. Many times I have seen a bus arrival time listed that was physically impossible -- a bus a mile away arriving in 1 minute, for instance. Transit's method still fell down in the case of buses which appear on the route but are just sitting at the beginning of the line. It treats that as a bus that's en route, but in my experience they just hang out for as long as they feel like it. I learned to look up the locations on the map and discard those, just annoyance #475 of public transportation.
JCzynski 10y ago
>Today we’re excited and proud to be launching in San Francisco. Not that we planned it or anything. Literally we just decided an hour ago to do this because somebody asked us to go live and we said “cool okay.” Startup culture at its finest right here.
- arielm 10y ago
  That statement actually makes me trust them less... I'm not saying they shouldn't move quickly on such opportunity, but don't make it so obvious. I hope it works flawlessly.
  - samvermette 10y ago
    Only reason we were able to deploy this quickly is because of the 2 pilots we've been successfully running for the past 3 weeks.
    - arielm 10y ago
      I don't for a second doubt the ability. I'm a big believer in moving fast, especially when there's such a bit opportunity for engagement and PR. It's just the wording which I found a bit "loose". But I'm a stickler for that sort of stuff ;)
Camillo 10y ago
I tried downloading the app, but the app store was offline. Not sure if that's considered ironic.
lanewinfield 10y ago
With all of NYC's subway stations receiving cell reception as of less than a week ago, this would be amazing for train times here. Besides a couple lines, the train times here are all estimated—and essentially useless.
- nhf 10y ago
  They're working on a beacon-based train time system now [1]. Bluetooth beacons are installed on cars with trackside sensors as trains enter and exit stations. Station wi-fi is used to relay data about entering and exiting trains. Since the MTA already has train speed and track length data, it gives a pretty accurate guess at where the train is between stations. It's not as accurate as full trackside sensors throughout the tunnels, but much faster and cheaper to spin up. Pretty clever overlay IMO. [1] http://www.metro-magazine.com/rail/article/715387/nyc-transit-uses-new-tech-tests-real-time-train-arrival http://www.metro-magazine.com/rail/article/715387/nyc-transi...
swang 10y ago
One of my lines I can ride into work is the N-Judah line. This line is the busiest MUNI line. It is faster to get to my work because it goes underground and the MUNI buses I have to ride go slower, especially when there are a lot of pickups. It is the most frustrating experience to walk to the N line because it is such a huge gamble. It use to get so crowded that you had to shove in there like you were on a line going into Tokyo. Just wall-to-wall people. They alleviated some of that by having more stops come back earlier going the other way. Only took them 2+ years since I've been here to decide that was an actual GOOD IDEA. Equally frustrating riding on the line is that the N-Judah is sometimes actually a J-Church train that comes back right before you enter the tunnel into Market St. The only sign it is not the one you want is the last car on the train is locked out, and the badging for the train still shows N-Judah. So sometimes you get on that expecting to go through the tunnel except it dumps you right outside of it. Now you have to pray there is room leftover on the next N-Judah line. Or alternatively you can try for the J-Church train that is entering near where the N dropped you off. So N-Judah is the closest line, but 50% of the time I have to use the J-Church train to actually get into work. And that's even assuming the tunnel into Market St. is not bogged down. Whose genius idea was it to have all lines converge into one line? What's the point of all this and what's it have to do with this article? Well fucking MUNI that's what. They recently increased fees for cash payment/limited payment card to $2.50. It is still just "$2.25" if you have a Clipper card. When I arrived here 4 years ago it was $1? $1.25? to ride. Literally for the increase of $1 more per ride I have seen service get progressively worse. When this bug hit, it said 30 minutes 'til the next N-Judah train. I tweeted to MUNI's twitter about what was going on and it took them 20 minutes to respond. And all they can say is "technical" problem as though 40% of the population in SF doesn't know what that really means. It is an embarrassment. SF is suppose to be some tech capital of the world yet we can't even ride into work without pulling our hair out over getting there. Sorry, ranting, but it is unbelievable how incompetent our public transportation is given how much money they have.
- wcummings 10y ago
  Do they really have that much money? Public transit has been chronically underfunded in most of the US, including progressive metro areas. Boston has billions of dollars of backlogged maintenance, which contributes a lot to reliability problems.
aristus 10y ago
This looks nice, but why does a bus app need to read my calendar? And photos? And identity?
- samvermette 10y ago
  Calendar to show a "Show Upcoming Events" option in the search. If you don't tap it we don't access it. Photos to allow installing the app on an SD card (popular request). Identity to use your name in the Send Feedback option. Here again if you don't tap it we don't access it.
- skybrian 10y ago
  In the latest version of Android you can turn these off under Settings->Apps->Transit->App Permissions
pronoiac 10y ago
I like this, but then it highlights the underground train stops without mobile signal.
pronoiac 10y ago
Checking, I see a lot of "invalid dates" alerts on lines.
glibgil 10y ago
The Transit app should pay Muni drivers to use the app. Maybe $5 to $10 a shift would be enough motivation. That would be an extra $100 to $200 a month income
- CoachRufus87 10y ago
  I'd be surprised if they were allowed to use their phones while working.
  - Zaheer 10y ago
    I think OP meant so that they have data for all transportation vs relying on users crowdsourced. Still crowdsourced but with better paid sourcing :) Note that it's passive - users don't have to do anything I believe aside from having the app.
    - glibgil 10y ago
      Yes, this is what I mean. Drivers would not have to do anything if Transit is smart about it. "Looks like you are a driver. You have taken two return trips on the 14 today. Enter your debit card and keep the app installed and we'll pay you $10 every day you drive"
      - jzwinck 10y ago
        Homeless people could just sit on buses all day and get paid for it too. And within a few days some disruptive people will duct tape their phones plus large batteries to the bottom of buses. Then terror alert.
        janekm 10y ago
        The homeless part sounds like a pretty good idea. You hear a lot about startup employees looking the other way with regard to social problems in the Bay Area, so partial solutions like this could be interesting? Of course $10 per day doesn't really raise anyone out of homelessness...
dkarapetyan 10y ago
SF public transit continues to boggle the mind.
Zigurd 10y ago
This should be crowdsourced. The official data here in the "T" bus service area is unreliable. Crowdsourcing would also enable measuring real-world service performance.
headcanon 10y ago
Very impressive. The word 'agile' has bad connotations, but I think it applies here. I hope that Transit and others can use this to jump-start their service in smaller municipalities. My hometown's bus system isn't terrible, but they're not the kind of organization that would see the value in making a proper app, or by publishing real-time bus data. I would love to be able to use an app like that where I live.
bonniemuffin 10y ago
Twice in the last few days I tried to use NextMuni and it gave me super wrong estimates, way more wrong than I've ever seen it. I wonder if this was why. I'm really grateful to have another option available.
bfrog 10y ago
Nifty, I know moovit has been doing this for a few years now as well as crowd sourcing route information for a bit. Transit has a pretty nice UI and a few interesting integrations. Fun to see some competition in this app space.

219 points by ant6n 10y ago

This is a smart use of crowdsourcing. From the app you can know the number of people who are waiting for the bus and relying on your reporting. Kind of gives you a sense of community. I hope this takes off.

arjie 10y ago
Right. Their little smiley telling me that nine people are being helped by my activating their app kept me longer at the stop than otherwise. The bus was four minutes off their prediction, though.
- samvermette 10y ago
  You're only going to get accurate departure times if someone else is using GO on that bus/train you're waiting for!
  - arjie 10y ago
    Oh, yes, I know. I hope the fellows downstream had better performance.
  - wikibob 10y ago
    What does it take to add real-time data for a new city? I've been interested in trying to scrape the data from the transit provider here. They have what appears to be reasonably accurate arrival predictions which they display through an atrocious app, and electronic signboards at some bus stops. The busses have a terminal for the driver that displays how long they need to dwell at times stops and announces the next stop, so I assume this is (at an unknown interval) transmitting the locations to their server. I haven't gone further in looking at it yet, because google maps will only add data sources directly from the official transit provider.
- noobermin 10y ago
  I live in a major city of ~800K with one of the lowest public transit ridership in the US, and I've learned to not quite trust their time predictions. We do luckily have GPS location data, and I regularly just look at the bus' actual position to gauge time most of the time.

Terretta 10y ago

I love the competition between Transit and CityMapper. In cities with coverage, they're both so good it's stressful trying to choose! We are spoiled by both in NYC and surrounding counties, in Paris Transit is crushing it with Velib integration, while in Copenhagen CityMapper's local know how is amazing.

tfigment 10y ago
Is this the same company? They list only supporting 2 Canadian cities (Victoria, Montreal) in the top link.
- samvermette 10y ago
  Those 2 cities is where we've been piloting crowdsourcing. The app itself works in 125 metro areas.
- coldshower 10y ago
  Waterloo region (including Cambridge and Kitchener in Ontario) is supported, as is Guelph. These aren't big cities by any stretch. I see a part list here: https://transitapp.com/region https://transitapp.com/region and more cities listed here: https://play.google.com/store/apps/details?id=com.thetransitapp.droid https://play.google.com/store/apps/details?id=com.thetransit...

Animats 10y ago

NextBus is an old system that uses 2G cellular links. It doesn't need much data, so that's fine. Now 2G service is being discontinued by AT&T in SF, and Muni has to switch. Or switch to T-Mobile, which is offering 2G service for another year. This was an early "Internet of Things" project. I met the inventor once at the Hacker Conference. This is a general problem with "Internet of Things" products - you're dependent on infrastructure which might go away.

Johnny555 10y ago
There's no excuse for this happening -- AT&T announced the end of life of 2G four years ago in 2012. Now Muni claims that the shutdown took them by surprise and they are scrambling to replace all of the old 2G equipment https://www.sfmta.com/about-sfmta/blog/why-muni-arrival-times-are-off-week-and-how-we%E2%80%99re-working-fix-them https://www.sfmta.com/about-sfmta/blog/why-muni-arrival-time... Simply put, the deactivation work that affects our vehicles started sooner than expected and outpaced our ongoing upgrade of all Muni vehicles to a new communications and monitoring system. http://usatoday30.usatoday.com/tech/news/story/2012-08-03/att-2g-network/56758432/1 http://usatoday30.usatoday.com/tech/news/story/2012-08-03/at... AT&T Friday [announced] that the shutdown of its second-generation, or "2G," wireless network will be complete by the end of 2016 At the time, I was managing some devices that used 2G to send telemetry -- our AT&T sales rep called us to warn us about the shutdown.
- xenadu02 10y ago
  I leased an Infinity over a year ago... it has the same "problem": the connected features rely on an ATT 2G radio. They made me sign a form acknowledging it would stop working January 1, 2017. There is zero chance Muni wasn't aware of this. I have no idea why they didn't ask for money to upgrade the system earlier... it seems like a tailor-made situation to get a budget increase: legitimately blame it on someone else and use the largesse to buy a brand new much better system that you normally can't get approved.
  - pnathan 10y ago

eli 10y ago

Neat. I had a similar idea a few years ago to make NextBus times more accurate... but fell down a rabbit hole trying to implement it as a cross-platform HTML app and never actually launched. If anyone wants my idea: My perception of NextBus's arrival times are that they're based on a very naive algorithm. I think it might literally be assuming that all buses always move at an average 15 mph. Lots of ways to make that more accurate with collecting just a teensy bit of actual arrival time data.

stuckagain 10y ago
I don't know about other systems but I've spent time in the field analyzing NextMuni's predictions and they have no predictive power whatsoever. The first arrival time is weakly correlated with actual arrival, and the 2nd and 3rd times are not even correlated. It's useless, essentially.
- danaliv 10y ago
  I learned early on to discard the units when using NextMuni. The J will arrive in three. Three what? Nobody knows, but there will be three of them.
skuhn 10y ago
Back when NextMuni was only 99% worthless, Transit had a method of actually getting reasonable arrival times out of it. Since each bus has a published GPS location (in addition to the NextMuni time), you can use that with driving direction data to loosely get travel time. When the app did this you would initially see Next Muni times -- 2 minutes away! -- which would update a few seconds later for lines you were particularly looking at -- oh wait 20 minutes away! Apparently NextMuni either just displays the schedule (which is literally never right) or does some kind of extremely basic estimation like you said. Many times I have seen a bus arrival time listed that was physically impossible -- a bus a mile away arriving in 1 minute, for instance. Transit's method still fell down in the case of buses which appear on the route but are just sitting at the beginning of the line. It treats that as a bus that's en route, but in my experience they just hang out for as long as they feel like it. I learned to look up the locations on the map and discard those, just annoyance #475 of public transportation.

JCzynski 10y ago

>Today we’re excited and proud to be launching in San Francisco. Not that we planned it or anything. Literally we just decided an hour ago to do this because somebody asked us to go live and we said “cool okay.” Startup culture at its finest right here.

arielm 10y ago
That statement actually makes me trust them less... I'm not saying they shouldn't move quickly on such opportunity, but don't make it so obvious. I hope it works flawlessly.
- samvermette 10y ago
  Only reason we were able to deploy this quickly is because of the 2 pilots we've been successfully running for the past 3 weeks.
  - arielm 10y ago
    I don't for a second doubt the ability. I'm a big believer in moving fast, especially when there's such a bit opportunity for engagement and PR. It's just the wording which I found a bit "loose". But I'm a stickler for that sort of stuff ;)

Camillo 10y ago

I tried downloading the app, but the app store was offline. Not sure if that's considered ironic.

lanewinfield 10y ago

With all of NYC's subway stations receiving cell reception as of less than a week ago, this would be amazing for train times here. Besides a couple lines, the train times here are all estimated—and essentially useless.

nhf 10y ago
They're working on a beacon-based train time system now [1]. Bluetooth beacons are installed on cars with trackside sensors as trains enter and exit stations. Station wi-fi is used to relay data about entering and exiting trains. Since the MTA already has train speed and track length data, it gives a pretty accurate guess at where the train is between stations. It's not as accurate as full trackside sensors throughout the tunnels, but much faster and cheaper to spin up. Pretty clever overlay IMO. [1] http://www.metro-magazine.com/rail/article/715387/nyc-transit-uses-new-tech-tests-real-time-train-arrival http://www.metro-magazine.com/rail/article/715387/nyc-transi...

swang 10y ago

One of my lines I can ride into work is the N-Judah line. This line is the busiest MUNI line. It is faster to get to my work because it goes underground and the MUNI buses I have to ride go slower, especially when there are a lot of pickups. It is the most frustrating experience to walk to the N line because it is such a huge gamble. It use to get so crowded that you had to shove in there like you were on a line going into Tokyo. Just wall-to-wall people. They alleviated some of that by having more stops come back earlier going the other way. Only took them 2+ years since I've been here to decide that was an actual GOOD IDEA. Equally frustrating riding on the line is that the N-Judah is sometimes actually a J-Church train that comes back right before you enter the tunnel into Market St. The only sign it is not the one you want is the last car on the train is locked out, and the badging for the train still shows N-Judah. So sometimes you get on that expecting to go through the tunnel except it dumps you right outside of it. Now you have to pray there is room leftover on the next N-Judah line. Or alternatively you can try for the J-Church train that is entering near where the N dropped you off. So N-Judah is the closest line, but 50% of the time I have to use the J-Church train to actually get into work. And that's even assuming the tunnel into Market St. is not bogged down. Whose genius idea was it to have all lines converge into one line? What's the point of all this and what's it have to do with this article? Well fucking MUNI that's what. They recently increased fees for cash payment/limited payment card to $2.50. It is still just "$2.25" if you have a Clipper card. When I arrived here 4 years ago it was $1? $1.25? to ride. Literally for the increase of $1 more per ride I have seen service get progressively worse. When this bug hit, it said 30 minutes 'til the next N-Judah train. I tweeted to MUNI's twitter about what was going on and it took them 20 minutes to respond. And all they can say is "technical" problem as though 40% of the population in SF doesn't know what that really means. It is an embarrassment. SF is suppose to be some tech capital of the world yet we can't even ride into work without pulling our hair out over getting there. Sorry, ranting, but it is unbelievable how incompetent our public transportation is given how much money they have.

wcummings 10y ago
Do they really have that much money? Public transit has been chronically underfunded in most of the US, including progressive metro areas. Boston has billions of dollars of backlogged maintenance, which contributes a lot to reliability problems.

aristus 10y ago

This looks nice, but why does a bus app need to read my calendar? And photos? And identity?

samvermette 10y ago
Calendar to show a "Show Upcoming Events" option in the search. If you don't tap it we don't access it. Photos to allow installing the app on an SD card (popular request). Identity to use your name in the Send Feedback option. Here again if you don't tap it we don't access it.
skybrian 10y ago
In the latest version of Android you can turn these off under Settings->Apps->Transit->App Permissions

pronoiac 10y ago

I like this, but then it highlights the underground train stops without mobile signal.

pronoiac 10y ago

Checking, I see a lot of "invalid dates" alerts on lines.

glibgil 10y ago

The Transit app should pay Muni drivers to use the app. Maybe $5 to $10 a shift would be enough motivation. That would be an extra $100 to $200 a month income

CoachRufus87 10y ago
I'd be surprised if they were allowed to use their phones while working.
- Zaheer 10y ago
  I think OP meant so that they have data for all transportation vs relying on users crowdsourced. Still crowdsourced but with better paid sourcing :) Note that it's passive - users don't have to do anything I believe aside from having the app.
  - glibgil 10y ago
    Yes, this is what I mean. Drivers would not have to do anything if Transit is smart about it. "Looks like you are a driver. You have taken two return trips on the 14 today. Enter your debit card and keep the app installed and we'll pay you $10 every day you drive"
    - jzwinck 10y ago
      Homeless people could just sit on buses all day and get paid for it too. And within a few days some disruptive people will duct tape their phones plus large batteries to the bottom of buses. Then terror alert.
      - janekm 10y ago
        The homeless part sounds like a pretty good idea. You hear a lot about startup employees looking the other way with regard to social problems in the Bay Area, so partial solutions like this could be interesting? Of course $10 per day doesn't really raise anyone out of homelessness...

dkarapetyan 10y ago

SF public transit continues to boggle the mind.

Zigurd 10y ago

This should be crowdsourced. The official data here in the "T" bus service area is unreliable. Crowdsourcing would also enable measuring real-world service performance.

headcanon 10y ago

Very impressive. The word 'agile' has bad connotations, but I think it applies here. I hope that Transit and others can use this to jump-start their service in smaller municipalities. My hometown's bus system isn't terrible, but they're not the kind of organization that would see the value in making a proper app, or by publishing real-time bus data. I would love to be able to use an app like that where I live.

bonniemuffin 10y ago

Twice in the last few days I tried to use NextMuni and it gave me super wrong estimates, way more wrong than I've ever seen it. I wonder if this was why. I'm really grateful to have another option available.

bfrog 10y ago

Nifty, I know moovit has been doing this for a few years now as well as crowd sourcing route information for a bit. Transit has a pretty nice UI and a few interesting integrations. Fun to see some competition in this app space.