Incen;ve Mechanisms for Societal Networks. Balaji Prabhakar Depts of EE and CS, Stanford

Transcription

1 Incen;ve Mechanisms for Societal Networks Balaji Prabhakar Depts of EE and CS, Stanford

2 y Research Background Un-l 2008: Computer Networks Internet Algos, Cloud Compu-ng : Societal Networks Transporta-on, Wellness, Recycli Dec 2007 I visited Bangalore

3 me

4 : Stanford Research Developed large- scale nudge engines for: Infosys (Bangalore) Singapore Public Transit Stanford (U.S. DoT) Bay Area Rapid Transit (BART) Accenture, USA à Wellness

5 12 now: Urban Engines Co- founded with some students and ex- Googlers Developed a big data system for large- scale mobility networks Transporta-on systems: road, bus, train (metro) systems Logis-cs and delivery chains Ø Urban Engines acquired by Google (Google Maps) in Sep 2016

6 Smart City Programs: A Personal Perspec;ve Engineering at the beginning of the 20th century Civil Engineering, Mechanical Engineering, Power Engineering à Call it the Edison- Ford type of engineering à Highly resource- constrained, built to last

7 Smart City Programs: A Personal Perspec;ve Engineering in the middle of the 20th century Electrical Engineering (VLSI, Comm) and Computer Science à Call it the Bell- Watson type of engineering à Moore s law, short product cycles, disrup;on

8 art Ci-es = Bell- Watson- ites working with Edison- Ford-

9 Urban Immobility

10

11

12

13

14 Not just commuter frustra;on and indignity

15 The Cost of Conges-on Time and fuel wasted (TTI) $115 billion In 2007 $121 billion in 2011 Emissions due to traffic 27% of all U.S. emissions (EIA) 25% for U.K. (DoE&CC, UK) Safety,

16 The Visible Problem Demand >> Supply

17 The Real Problem Dated Transporta;on Architecture

18 Transporta-on Architecture of the 20th Century Networks gencies / Operators Departments of Transporta-on Public transit operators Taxi companies Commuters RAIL Operate BUS Use CAR TAXI No flow of informa;on Demand doesn t talk to supply!

19 Transporta-on Systems of the 21st Century Networks gencies / Operators Departments of Transporta-on Public transit operators Taxi companies Commuters RAIL Operate BUS CAR TAXI Use

20 Transporta-on Systems of the 21st Century Networks gencies / Operators Operate Real- -me analy-cs RAIL BUS CAR Commuters TAXI Real ;me usage data Use Incen-ves User behavio Cloud, Big Data, IoT

21 Transporta-on Systems of the 21st Century Networks gencies / Operators Operate RAIL BUS CAR Commuters TAXI Use Key Enablers Cheap, ubiquitous sensors GPS RFID/NFC Bluetooth Technological advances Abundant bandwidth Cellular WiFi Internet of Things Cloud compu;ng Big data

22 Commerce and Retail in the 20th Century The Producers gners/manufacturers Stores Stock The Consumer Customers Shop

23 Commerce and Retail in the 21st Century The Producers gners/manufacturers Online + Delivery List ecommerce + Delivery The Consumers Customers Buy Online Home Delivery Pickup + Delivery

24 Basic Problem Demand >> Supply

25 Our Work Increase supply using big data, matching markets Demand >> Supply Reduce demand using incen;ves, not penal;es

26 Incen-ves: Singapore MRT The INSINC project Primary goal: Incen-vize offpeak travel Launch and current status Stanford + NUS, Jan Jun 2012 Urban Engines, July 2012 Currently: Travel Smart Rewards 380,000+ registered par-cipants & 75+ corporate members Shil in peak load: ~ 10%

27 INSINC Demo st Commu-ng History Electronic Ticket kms to credits 3x for off- peak ommuter The Insinc portal Credit History Date 15th June 16th June 16th June 18th June Time Credits 09:00:19 08:10:45 16:20:17 06:15:20 Rewards

28 INSINC: Jan 2012 June 2012 registered/admioed and ac-vated 22867/20319 random/determinis-c redeemers 87.6%/12.4% number of recommend- a- friend s sent (excluding reminders) people with friends (59.9%) total rewards (including referral and joining bonuses) $137,639

29 5 FRACTION OF COMMUTERS IN 5-MINUTE SLOTS Before After -7.49% B A % TRIP START TIME TRIP START TIME FRACTION OF COMMUTERS IN 5-MINUTE SLOTS Before After % 0.90 B A % TRIP START TIME TRIP START TIME 10 11

30 f par;cipants All in the group Mild peakers Medium peakers Heavy pe ar-cipants e with Insinc friends without Insinc friends me players d exchange rt distance mmuters g distance mmuters

31 Our Work Increase supply using big data, matching markets Demand >> Supply Reduce demand using incen;ves, not penal;es

32 Things that move in a city GPS Smart travel card Smart phone Barcode / R

33 Lots of data, but it is Piecemeal: tap- in/tap- out, train loads, Error- prone and noisy: needs healing and curing Siloed: different orgs, database technologies and formats What s needed: A system and algorithms for solving a massive jigsaw puzzle