Andreas Weigend, Social Data Revolution | MS&E 237, Stanford University, Spring 2011 | Course Wiki

The Impact of Social Data on Mobility

[Ruchi Varshney rvarshne] [Eva Petrova-Ibarria evapi] [Sebastiaan Boer sboer]

Introduction


Many issues related to the mobility problem are caused by the inability of information to reach the individuals who need it most. A person who is passing the scene of a highway accident knows exactly what happened, but the thousands of people in the traffic jam behind them have no clue what they are waiting for. Drivers in a parking lot know that there are only a few spots left, but those on their way to the city have no clue that they might be searching for parking forever. Commuters on the bus know that their bus is running 20 minutes late, but those waiting at the bus stop ahead are just getting more and more frustrated.

If only information about mobility would spread easier, the mobility problem might just be alleviated slightly. Social data on mobility has the potential to change mobility-related behavior, particularly at times of peak activity when it matters most. Social data should be used to control the flow of traffic on the streets, highways and parking lots to ensure that our transportation systems work efficiently by alleviating congestions. New data sources shared by the government, information shared through social networks and instant updates to mobile devices can enable everyone to tap into the wisdom of the crowd and make smarter decisions.

On this Wiki page, we present a PHAME analysis of the mobility problem and how social data can help solve it. We will present our hypothesis on how social data can impact mobility and what metrics and experiments can be used to measure the impact of socializing traffic and parking related data. At the end, we will present a reality check of the ideas we have presented. Would people socialize data without incentives? Is dynamic pricing of tolls and parking the only way to control people's behavior? Is it fair that only people who can afford to pay high parking fees and tolls benefit from the system? What happens if everyone tries to optimize their travel route? Does that only increase congestion? This wiki intends to answer some of those questions and also leave some food for thought.

Problem

Millions of people around the world suffer from mobility problems every day. Cities are facing exponential increases in traffic. Most traffic congestion is attributed to the inability of roads to accommodate the increased amount of cars and drivers during peak times; the rest is attributed to traffic incidents, road work, weather conditions and public events. Traffic jams are significantly worsened by the increasing amount of drivers looking for parking in the big cities.
traffic.png
The impact of the mobility problem is massive. Having bad traffic conditions means many accidents, heavy pollution, intense crowding, long commutes and a host of other ugly conditions and experiences. The Texas Transportation Institute estimated that in the year 2000 the 75 largest metropolitan areas of the USA experienced 3.6 billion vehicle-hours of delay, resulting in 5.7 billion gallons in wasted fuel and $67.5 billion in lost productivity, or about 0.7% of the nation's GDP. It also estimated that the annual cost of congestion for each driver was approximately $1,000 in very large cities and $200 in small cities. Traffic experts estimate that more than 30 percent of traffic congestion in cities is caused by drivers circling the block looking for a parking space. Traffic congestion increases environmental pollution, affects people’s health and harms communities. Studies have linked traffic to high blood pressure, stroke, and heart disease.

By late 2010, the US cities with the worst rush hour congestions were Los Angeles, San Francisco-Oakland, Washington, D.C., Atlanta, GA, and Houston TX. The cities with the world's worst traffic conditions include Tokyo, Moscow, Shanghai and Mumbai. The full list of the top 20 world's cities with worst traffic jams can be found here. Some initiatives of the US government to help understand congestion behavior can be found here.

Hypothesis

The traditional ways proposed by governments to solve the mobility problem are: increasing capacity, increasing the efficiency of the existing system, and better management of construction and maintenance projects. But the reality is that cities are rarely able to build sufficient infrastructure to cope with the mobility problem. Therefore, there is need for creative thinking to solve the mobility problem. Our hypothesis is therefore:

Social data can be used to impact mobility-related behavior and significantly help alleviate the mobility problem.

Jay Nath, the Director of Innovation at the city of San Francisco, strongly believes in the capability of social data to change people's behavior. He said: "I think the signals broadcasted via social data have the ability to make significant change in mobility. For example, I’ve seen the growth of the biking culture here in SF and I think it’s in large part due to biking being seen as hip. Social data generated location apps have the potential to amplify that trend inside/outside SF. You can also think of people sharing their carbon footprint and driving behavior change that way. Of course that’s one example and I think that’s the beauty of data. The next big idea in this area is likely to be a unique combination of datasets packaged in the right way."

Analysis

As a first introduction to the impact Social Data can have on the mobility problem, this chapter will look at the ways Social Data is used today to change the behavior of individuals and their relation to mobility. Behavior that relates to mobility can only change in four distinct ways. People can either change their means of transportation, change their path, their departure time or their destination. The following paragraphs contain examples of how Social Data currently impacts each of those types of mobility-related behaviors.

Changing the means

Knowledge of parking information and traffic congestion could play a pivotal role in helping commuters decide their means of transport, such as driving to SF or taking the Caltrain. If people began to share information about when they were driving to different destinations or the kind of vehicles they own, they could save money and keep the earth green by ride-sharing or vehicle-sharing. Some services like Avego and Relay Rides have already made strides into this method of changing the means by which people travel. Avego offers a method to match drivers and riders in real time as they travel. They have a mechanism in place to check the background of riders and drivers and enable online payments for rides. Relay Rides enables people to borrow cars from car owners in their neighborhood by the hour or by the day. Through this method car owners can monetize the time their car is sitting in a parking lot, while borrowers can borrow cars at low rates. They have a similar mechanism of verifying the renter's driving record and insurance. Another well known application is Roadify, that can simultaneously uses crowdsourced information to connects people to real-time parking and public transit information in your neighborhood through mobile phones. With Roadify, commuters send and receive updates that help them find available parking, catch the next bus or train, or avoid traffic jams before it is too late.

Changing the path

Crowdsourcing of current traffic data can be key in optimizing routes of travel. Waze is a social mobile application that provides free turn-by-turn navigation based on the live conditions of the road. It is completely powered by users automatically providing data about the traffic conditions around them when they have the Waze app turned on. Live map modeling based on real-time GPS information enables everyone to benefit. For instance, this application enables people to know when Highway 101 is clogged up so that they can steer towards Highway 280. This app enjoys strong networking effects, as the route guidance and accuracy becomes better and better as more people use the app.

Similar initiatives are taking place in other countries like China and Japan. The Beijing Transportation Information Center started a project called Star Wings that generates real-time traffic information based on Probe car traffic information collected from car navigation systems. This information is then broadcasted back to navigation systems or mobile phones. Results from driving tests show that the average reduction in travel time was about 20%. The Sky Project in Japan started by Nissan uses the same mechanism.

Changing the time

A large number of people have some flexibility when making travel plans, and would be open to adjusting the time of their travel if that would save them the long and agonizing hours sitting in traffic. Before leaving on a trip or heading off for the office people have the option of analyzing existing sources of social data such as Google maps, the radio, or traffic news, before starting to move. If you find out that your particular route is in traffic then you have the option of not leaving until the problem is recovered. The majority of traffic jams occur around 7am and 5pm as these are the common hours of work schedules. The best way to save time is to listen to social data and adjust your work or trip schedule accordingly.

Changing the destination or intent of travel

A large number of people have several alternatives when deciding how to spend their free time. Some people may hesitate whether to spend their date night going to a movie in the neighborhood or driving to the city. In those cases, social data can significantly impact their decision. Finding out that the route to the city is plagued with traffic would probably make the movie alternative a much more desirable option. Furthermore, social data can not only change people’s mind about where they are heading but also has the power to change their decision whether to go at all. Many employers today offer telecommuting, and more and more employees are choosing to stay home in order to avoid traffic headaches and increase their productivity.

Metrics

One of the biggest concerns of deploying a social data based application is measuring success. Justifying efforts placed to incentivize people and governments to share information is extremely important. Some of the ways the impact of socializing traffic and parking data can be measured are:
  • Amount of real-time data socialized by commuters on social media sites
  • The amount of successful carpool arrangements made by services such as Avego
  • Application of computer analytics on data collected from a host of wireless sensors and remote cameras to determine the change in speed and volume of cars that are on the street over time or the number of parking spots available at different times in the day
  • The download/usage rates and ratings of applications like Waze, SFPark etc.
  • The number of people hitting parking and traffic related websites, reduction of calls made for finding out public transit timing information
  • Satisfaction of drivers on the street who are on Twitter or Facebook when they are stuck in traffic. This approach would be similar to the Gross National Happiness calculation done by Facebook

Experiment

In this section we will discuss several experiments that could change people’s behavior in regards to mobility. First, we will propose how Caltrain and BART could increase the efficiency of their transportation services by releasing utilization data to the public. Then we’re going to look at the SFPark initiative by the San Francisco Municipal Transportation Agency, and finally look at several new data intelligence services, such as SpotRank, SimpleGeo and Inrix.

  • BART/Caltrain

BART_Bay_Area_Rapid_Transit-logo-51743C1461-seeklogo.com.gifOne of the main benefits of public transportation compared to driving your own car is the ability to do work during your commute. You can’t work, however, when there is no seating space available. If people would know when seating is available, they can adjust their travel time accordingly.
Experiment BART and Caltrain both have electronic check-in systems, and hence are sitting on a massive dataset that contains information on past seating availability. Our experiment is to aggregate that seating availability and create an API or website that displays estimated seat utilization. This information could be used in applications such as 511.org and Google Maps. Currently, transportation websites show information on pricing and travel time, but could easily include a metric indicating the chance of seat availability.
Behavioral Changes We expect to see a decrease of utilization of the public transportation system during peak hours. Individuals that have flexible schedules can arrange their travel times outside the peak hour.

  • SFPark

external image logo.png
SFPark is an experimental initiative by SFMTA (San Francisco Municipal Transportation Agency) to make parking data public. This organization has created a publicly available API that allows developers to receive real-time information about the approximate availability and price of street parking in certain busy areas of the city. The availability for street parking is given in grades of utilization, such as ‘High’, ‘Medium’ and ‘Low’, and is aggregated to a street level. The original idea behind this initiative revolves around the fact that drivers will have to spend less time searching for an available parking spot when they use the information provided by this API.
Experiment SFPark itself is a governmental Social Data experiment. The upcoming months will tell us how the API will change the behavior of drivers in San Francisco. If we were able to take charge of SFPark, however, we would make the following changes to the implementation of this system. First of all, we would improve the API to show the real time availability of every individual parking spot in the selected areas. Secondly, we would introduce a pricing model that takes into account real-time supply and demand. The price of a parking spot would depend on the utilization of the parking spots in the same street. A parking spot would then become very cheap on Monday at 4AM and quite expensive at 1PM.
Behavioral Changes If our proposed changes are implemented, we expect to see two behavioral changes in mobility-related behavior that are directly related to the experiment. First of all, we expect commuters to be more inclined to take public transportation during the busiest hours of the day (as parking will be really expensive). Secondly, drivers will spend less time circling around the city searching for a parking spot, as they will be able to quickly find a parking spot that fits their preferences and hence decrease their time spent on the busy roads.

  • SpotRank/SimpleGeo

SpotPank.png
SpotRank is a new data intelligence service from Skyhook. SpotRank predicts the density of people in predefined urban square-block areas worldwide at any hour, any day of the week. It uses a massive reference network comprised of the known locations of over 250 million Wi-Fi access points and cellular towers. To develop this database, Skyhook has deployed drivers to survey every single street, highway, and alley in tens of thousands of cities and towns worldwide, scanning for Wi-Fi access points and cell towers plotting their precise geographic locations. SimpleGeo is an API service that surfaces SpotRank density and other real-time location data on platforms such a iOS and Android so that developers can easily build location-aware apps.
Experiment Data from SpotRank is currently used to serve location-based content and ads, but it could also easily be used to spot traffic congestion, and even people congestion as a result of sport games, concerts, and other public events. The challenge will be how to incentivize private companies, such as SpotRank to share their data for free.
Behavioral Changes If Skyhook releases its data, it has the potential to significantly impact people's behavior. Commuters will be more inclined to choose routes with lower density of people. In addition, this data could be used by the police to plan its resources on certain days and place more police staff in the most crowded spots in town or the busiest freeways, which can have a secondary impact on the mobility problem.

  • Inrix

INRIX_-_Google_Chrome_2011-05-30_16-00-01.png
Inrix is a traffic service company that provides traffic information based on historical tracking mixed with real time traffic information. Inrix uses information gathered by clients on Android phones, the iPhone, as well as Microsoft’s Sync, used by Ford. Inrix also collects information on traffic congestion from over 3 million trucks, delivery vans, and other fleet vehicles that are equipped with GPS.
Experiment The wealth of information gathered by Inrix each day goes beyond what the traditional GPS traffic companies do. Inrix includes statistics such as school bus routes, concerts, sporting events, and even takes into account the legislative calendar in Washington D.C. Inrix is currently planning to conduct an experiment and use all of this information to gather a clear picture of what drivers should expect to see on their commute.
Behavioral Changes Inrix consolidates data from several different sources to deliver a new type of solution that can become more valuable that the traditional traffic services.
The ability to inform users of what should be coming up on their drive will be of significant value and can change people's driving behavior.

  • Social Media

facebook.png
Ford’s Sync as well as the newly announced Toyota Entune will be implementing car mobile apps that will include many social media apps. General Motors OnStar has announced an upcoming feature on their latest model that will read aloud Facebook and Twitter updates.
Experiment The emergence of these technologies creates opportunities for people to share social data, and tell everyone of a major accident or traffic jam that came up on their drive. The premise for this experiment is that having social media available on the in car console will allow drivers to pinpoint a new spot of congestion. Users of this new social media will be able to redirect their route to steer clear of traffic.
Behavior Changes Having an ever evolving map of your routes allows drivers to be more informed and less likely to cause further congestion by inducing another accident or rubbernecking.Having a city of traffic updaters will allow for each street and freeway to be Facebook updated or Tweeted up to the minute offering drivers at home a new route to avoid the current traffic jam. While each new driver is made aware of the oncoming traffic, drivers 30 minutes away will be able to alter their routes.

In Reality

According to “TRAFFIC, Why we drive the way we do” by Tom Vanderbilt, many optimizations of the traffic network have effects that only last for a very short period of time. If everyone decides to drive on the fast lane, this lane will get congested. Tom Vanderbilt stated “The classic issue is if everyone is told road A is suddenly not congested, everyone shifts to road A, and it becomes congested. Past a certain point, when roads reach the sort of capacities that spill into stop-and-go traffic, route information isn’t going to matter much.”

We could imagine that if everyone in San Francisco has the SFPark App, a desirable free parking spot will get hordes of interested drivers rushing to get it. This could cause collisions, aggressive driving and road rage.

Also, a big hurdle in the way of implementing socialized data applications is the incentive for people to share their geo-location or tweet information about traffic conditions in their whereabouts. Is pricing the only incentive? Singapore was the first city in the world to implement an electronic road toll collection system for purposes of congestion pricing. All cars are manufactured with transponders that are used to charge people entering downtown during peak hours. Would using a Foursquare like method of giving badges incentive enough?

Finally, as we’ve seen in Class 16, each and every dataset that is made public can be used in a destructive way by either criminals or organized crime. One could imagine using seat availability data to plan the optimal time for an attack on the public transportation system. In our opinion, these long tail probabilities are mostly impossible to resolve and are not valid reasons to hold back the data we discussed above.

References

  • Thoughts on a Smarter Planet, Wired Magazine
  • TRAFFIC, Why we drive the way we do, Tom Vanderbilt
  • Traffic Congestion, Wikipedia, here
  • Bart’s Open Data Initiative, for Developers
  • Bart's Public Data API
  • The Need for Creative Thinking to solve the Traffic Crisis, here
  • Nissan's new Intelligent Transport Systems (ITS) technology, here
  • Can Social media help to reduce traffic congestion?, here
  • Official blog of Tom Vanderbilt – author of Traffic, here