Feature - SimCity, social engineering, and 60 million “people”

The ever-popular SimCity games allow players to create and manipulate their own fictional cities. Now grid computing is allowing researchers to do the same, this time with real social and geographic data. Image © Electronic Arts Inc. All rights reserved.

How long are you expected to live? Where do you go to hospital? Do you own your own home? How often do you go to football games?

If you’re from the UK, chances are Mark Birkin has all of these answers…only he doesn’t know exactly which of his 60 million simulated “people” you are.

A geographer by training, Birkin is part of a University of Leeds team using grid computing to run predictive models in their own real-life version of “SimCity”: integrating real Census data, survey data, healthcare data and more in an anonymous but accurate model of the entire UK population, and then projecting that information into the future.

“We can run alternative scenarios to determine the impact of different policy decisions,” he explains. “We could be concerned with increasing life expectancy, or determining the effect of immigration, or investigating the impacts of the aging population.”

“We’ve been working on grid projects for about four years, starting with basic location optimization problems,” he says. “Things like where to put a new hospital, or the impacts of putting schools in different places.”

Crunching everybody’s numbers 

The scale of each projection is massive: each of the 60 million individuals in the simulated population has several hundred different characteristics, based on real-world behaviors, and any of their myriad potential interactions can be predicted over twenty or thirty years.

“Even if we just focus on hospitals, we need to think about where people are, who they are, their age, morbidity, health characteristics, their needs…we need to integrate information about hospital capacities, transport information, accessibility, and then think about how these different things impact on other things. We’re talking about quite a lot of data.”

Each of the 60 million individuals in the simulated population has several hundred different characteristics, based on real-world behaviors. Stock image from sxu.hu

The project, named MoSES (Modelling and Simulation for e-Social Science), is running on the White Rose grid, a collaboration between the three Yorkshire Universities of Leeds, York and Sheffield.

It is this grid access, says Birkin, that makes such ambitious projections possible: “Historically, people have assembled data on a single PC or workstation. E-science provides exciting opportunities to access multiple databases from remote, virtual locations, making it possible to develop highly generic simulation models which are easy to update.”

So will MoSES replace our politicians? Not exactly. But it can provide them with some useful information.

“It’d be great to build a support framework for all the social services, to provide a decision support system, like a health care planner, where people can access these scenarios to get some decent information for decision making,” Birkin says.

MoSES is funded by the UK Economic and Social Research Council National Centre for e-Social Science.

- Cristy Burne, iSGTW