1

Urban Encounters: The game of real life

Eamonn O’NeillUniversity of Bath

Department of Computer Science

Vassilis KostakosUniversity of Madeira / Carnegie

Mellon University

CHI 2008 Proceedings Works In Progress

2

Outline

• Introduction• Modelling human encounter in urban space• The game of real life• Describing the cell dynamics• Modelling life, death and survival• Discussion• Conclusion and ongoing work

3

Introduction(1/2)

• A game that helps us model urban encounter• Developing models of human movement and

patterns of encounter in cities• Aim:– a systemic understanding of cities and urban life– use this understanding to aid in the development

of urban pervasive applications

4

Introduction(2/2)

• “The Game of Real Life” is an extension of Conway’s Game of Life

• User plays game on the mobile device and the game gathers empirical data by Bluetooth

5

Modelling human encounter in urban space

• Models can be broadly categorised into three levels:– Macro– Meso– Micro

• Applications of encounter– Delay Tolerant Networks– a mobile application

6

The game of real life(1/4)

• Understanding the patterns of urban encounters

• The Game of Life as a basis for simulating how people encounter and interact with each other

7

The game of real life(2/4)

• The game of life:– Survival– Death– Life

8

The game of real life(3/4)

• The game of real life:– The playing board becomes an undirected graph– Cells are not limited to two states

9

The game of real life(4/4)

• Address two issues:– define the set of neighbours that a cell can have at

iteration n of the game.– define the rules that drive life, death and survival.

10

Describing the cell dynamics

• Data-gathering mobile application that utilises Bluetooth

• Implements a version of the Game of Real Life in order to motivate users to keep it running on their mobile devices.

11

Modelling life, death and survival(1/2)

• A cell represents a specific mobile Bluetooth device

• The number of neighbouring devices is given by the results of a Bluetooth discovery scan

• The cell’s state may decrease or increase• Analysed a week’s data from three participants• From this data we derived a set of rules for

changing the cells’ state

12

Modelling life, death and survival(2/2)

• Identified as optimum the following rules:– Under-crowding is 2 or fewer neighbouring

Bluetooth devices– Desirable number of neighbouring devices is 3-5– Over-crowding is 6 or more neighbouring devices

13

Modelling life, death and survival(1/2)

• Adding memory to the cell

14

Modelling life, death and survival(2/2)

15

Discussion(1/2)

• Inevitably most of the time was spent on death (states 0 and 1) and state 6, which was the maximum possible.

• The states between the two extremes typically act as buffers and are occupied only temporarily

16

Discussion(2/2)

• Identified that adding memory to the cells produces graphs with smaller and fewer fluctuations.

17

Conclusion and ongoing work(1/2)

• Describe extensions to the game that make it more closely resemble human encounters

• Present a mobile application that acts both as a game for users and a data collection tool.

• Cell spend most time on the extreme states• Equipping each cell with memory enables it to

predict its state by utilising its own memory

18

Conclusion and ongoing work(2/2)

• This paper provides the groundwork for developing such an account.

• Attempting to determine the asymptotic behaviour of our model as influenced by different set of values for life and death

• Ultimately develop a mathematical account of cell dynamics that closely matches the encounters recorded by our Bluetooth scanners.