Model Citizens

Visualization

In the busy world of urban planning, important information and data can get buried in spreadsheets and reports. When conveyed visually, data delivers more powerful and meaningful messages.

In 2011, the Guardian newspaper's interactive team developed an interactive visualization (tinyurl.com/olltpbf) of rumors that proliferated during the UK riots that year. Unsubstantiated stories about various occurrences abounded, and people took to social media to show support, opposition, or skepticism regarding a particular occurrence. The interactive team created rumor timelines that showed key tweeters, information that was retweeted, and the associated sentiments, all of which were color-coded.

Each cluster of colors — green, red, or gold — reflected a specific tweet that was significantly retweeted. It allowed them to follow a rumor from its inception, including who started it and when, how it proliferated, and how others reacted to it over time.

One rumor they followed was that police beat a 16-year-old girl. The visual timeline clearly shows when the rumor started, and that a few hours later questions about its accuracy arose. That was followed by more people retweeting the rumor. When someone posted a possible video of the situation, tweeters began to question its validity again, since the video showed a disturbance but no beating. Still, without evidence of a beating, the video seemed to breathe new life into the rumor, questioning stopped, and the video was widely disseminated over social media.

The team captured important patterns that revealed how powerful information — especially visual information — can be. It also shows a potential way that planners can visually grasp citizen sentiment, how quickly issues can take off, and how quickly sentiments change.

At the Columbia University Spatial Information Design Lab, researchers in 2006 created Million Dollar Blocks, a visualization tool that uses data from the criminal justice system to demonstrate how incarceration displaces people and affects neighborhoods. In New York City, decades of gentrification, "wars on crime," and various policy interventions have had harsher effects on poor, minority communities. Further, the researchers showed in their visualizations that in many large U.S. cities, more than $1 million can be spent to incarcerate individuals who hail from the same single city block — so-called Million Dollar Blocks (tinyurl.com/o5dffyh).

Researchers infer from the visualizations that the incarceration of these community members has significant impacts on education, housing, health, and families — impacts that will, in turn, require urban planning and public policy attention.

Just last year, the Arizona State University Decision Theater Network (of which coauthor Kevin Desouza was interim director) analyzed more than 124 million tweets related to the FIFA World Cup, which took place in Brazil from June 12 through July 13. By using social media, geotagging, and visualization technologies, DTN gained richer insights about the event and the physical space. Using sentiment-analysis tools, tweets were analyzed and translated into high-definition visualizations that reflected tweet volume and user sentiment, as well as where and when it occurred.

Two analyses illustrate this. The Twittersphere reacted strongly to an incident in which Uruguay's Luis Suarez bit Italian defender Giorgio Chiellini on June 25. The next day, he was suspended. Negative sentiment continued for days after.

The team also looked at Twitter volume and sentiment during a 24-hour period surrounding the semifinal match between Netherlands and Argentina on July 9. Pregame sentiment is a bit difficult to explain without some context (which we won't go into here), but during the game, shifts in the Twitter data are clearly understood. Tweet volume spiked in response to consecutive yellow cards, generating negative sentiment scores (represented in red), while Klaas-Jan Huntelaar's goal about an hour later generated an outpouring of positive sentiment (green).

For planners, exercises such as these, which extract and analyze sentiments, can be valuable for understanding residents' overall perception of projects as well as their attitudes toward various related interventions.

Data analysis and engagement

Planners commonly use surveys to collect information. Traditionally, a group will collect survey data, analyze it, and present the information verbally and in report format. Other stakeholders have little input. This creates space for limited engagement, incomplete analysis, and poor actionable information.

Visualization offers an opportunity to engage more with data. The Arizona State University Decision Theater worked with the Raza Development Fund, an organization that serves the Latino community and low-income families in Phoenix through the investment of capital and financing solutions. To find ways to improve services, Raza used two door-to-door surveys for two South Phoenix groups — 1,000 families and 80 business owners — and fed the data into a variety of commercial and in-house software programs to generate a visual representation of responses.

From there, high-definition visualization dashboards both offered insights into respondents' answers and sparked more questions. Another important outcome: Community leaders walked away with more than a summary report. They now had a tool for inspecting the data themselves, drilling down on key elements of the survey, and even doing on-the-fly analyses.

Collaboration throughout — involving representatives from Raza, the DTN team, and South Phoenix communities — helped the partners to define the project and scope, as well as make decisions based on the results shown on the dashboard.

Several interesting correlations between the data and individuals familiar with the South Phoenix neighborhoods were revealed. The question, "What is missing in your community that can improve quality of life?" yielded various responses, but most people called for more streetlights. That came as a surprise to community leaders, who expected people would say "more police" because of the high crime in the area. Discussions with residents revealed that streetlights were a bigger need, in part because their absence made cab drivers feel unsafe, causing them to avoid the neighborhoods.

Drilling down, they found correlations between the amount of time someone has lived in the area — and their specific location — to their response. Largely, residents of 20-plus years were likely to call for more streetlights. Among the people expressing a need for more police, many lived near intersections where crime traditionally happens.

Further, the business assessments revealed the impact of a major state law: 2010's controversial anti-immigration law known as SB 1070, which requires police to determine the immigration status of someone if there is a "reasonable suspicion" they are in the U.S. illegally. Overwhelmingly, business owners said that SB 1070 had negative impacts on their business. Interestingly, those that claimed no impact had fewer employees than business owners who noted a negative impact.

The visualization of urban datasets also proved very helpful in terms of citizen engagement and transparency in New York. In 2013, the New York City Department of City Planning released MapPLUTO (Property Land Use Tax Lot Output), a massive detailed spreadsheet of every piece of property in the city. This data — including the unique tax lots, number of buildings on the lot, square footage, and other tax information — from the country's largest city offered several opportunities for research and investigation into practices and policies.

The chief science officer of the cloud-based mapping platform CartoDB, Andrew Hill, created several visualizations, such as the number of residential units per tax lot, the age of buildings by location, the largest private landowner in Manhattan (groups related to Columbia University), and other useful and interesting data visualizations that clue citizens into the city's policies and land ownership (tinyurl.com/ndabgdr).

Modeling and simulation

Beyond visualization, modeling and simulation provide ways to test how a variable will behave alongside other variables, without having to test it in real life. When complex data and information are hard to process, users may find it difficult to gauge subtle combinations of decisions. This is particularly troublesome when dealing with large-scale, complex problems that affect individuals, institutions, and systems, and it can often have significant short- and long-term consequences.

A tool planners and others know well illustrates this completely. The game SimCity lets players build a city by simulating the actions of an urban planner and developing residential, commercial, and industrial zones. The player (who serves as the mayor) starts with a blank map and must expand the city within the budget provided, as well as install a government, implement services for citizens (schools, police, fire departments, parks, and health), and make tough decisions on financial trade-offs. Planners and their local government colleagues must constantly sift through these complex issues, but few are armed with such a robust way to simulate the impacts of their decision making.

In "What SimCity Teaches Us About Real Cities of the Future" (tinyurl.com/b9rl8fc), Adam Sneed recalls important decisions he made when playing the game. To build revenue he could dedicate himself to industry, which he knew would bring in money but would do environmental damage. Or he could diversify by investing in commerce, education, and tourism, an approach that would take considerably more time.

As in real life, when his imaginary city hit hard times, it was hard to change course because earlier decisions were entrenched. Likewise, he got a lesson in resiliency. When the city's key industry nosedived, it plunged the whole place into turmoil. (For a real-life example, we need only to turn to single-industry towns like Detroit.)

For planners, achieving a goal while satisfying multiple wants, needs, and desires — as well as competing objectives — from multiple stakeholders is a challenge. Simulating multiple scenarios and analyzing the various trade-offs can help, but too often, these exercises are done by experts behind closed doors. Today, we have tools and environments that let us bring stakeholders into this process, allowing for increased engagement and transparency.

A great example of this is the Arizona Budget Analysis Tool, an interactive model that demonstrated how different decisions would affect the overall budget and the budgets of various state agencies. AzBAT (www.dtn.asu.edu) allowed decision makers to engage multiple stakeholders with competing needs. Planners frequently find themselves in similar situations, particularly where financing is concerned, and modeling and simulation can provide the context and visible evidence necessary to see the workability — and consequences — of a particular approach.

Researchers at the Digital Arts, Leadership & Innovation Lab in Computer Science at Dartmouth College are developing the Collaborative Urban Planning Sandbox (tinyurl.com/qzgfydo) to provide intuitive modeling for both technical and nontechnical users to publicly collaborate and deliberate.

CUPS is a traveling interactive planning system that provides users with physical blocks on a special table that represents different structures such as schools, residential buildings, and agricultural buildings. Users — including policy makers, professionals, and community members — place and move blocks at will and a 3-D visualization reflects the changes, and offers design tips and guidance as needed. The moves generate immediate feedback about the potential impacts from their design changes. Through CUPS, groups who previously couldn't or wouldn't engage together in the design process have greater opportunities to collaborate on issues that can be trying and contentious.

Similarly, a collaboration was undertaken to conduct a multi-stakeholder modeling of energy reform in Mexico. Mexico has decided to open up and auction off its oil and natural gas reserves to foreign investors and competitors for the first time in 77 years.

To optimize investments, DTN developed a holistic model that helps visualize various investment scenarios. Using DTN's Complex Systems Framework — a proprietary software that enables statistical models to be visualized in single or multiscreen environments — an interactive seven-screen model was created. The model allows simulation of various energy investment options (deep-water drilling, for example) and displays their impact on socioeconomic variables such as GDP, job creation, and tax revenue.

Representatives of the federal Secretaría de Energía de México used the tool in early 2015 to explore different scenarios — what happens if they modify the investment amounts for exploration and production, for instance? Energy planners quickly grasped which options would be best for fostering job creation, lowering carbon emissions, or achieving some other result, and Mexican officials are now empowered to make data-driven decisions.

Technologies are opening up new avenues for us to engage with stakeholders and immerse them in all aspects of the planning process. By allowing diverse stakeholders to come together to resolve differences, analyze trade-offs, and simulate both the intended and the unintended consequences of decisions, emerging modeling, visualization, and simulation tools make decision making stronger, more transparent, and timely.

Kevin C. Desouza is an ASU Foundation professor of Public Affairs, the associate dean of research at ASU's College of Public Service and Community Solutions, and a senior research fellow of the Decision Theater Network. Kendra L. Smith is a postdoctoral fellow at the College of Public Service & Community Solutions and a research fellow at the Center for Urban Innovation at ASU.

Resources

Tableau (visual analytics): www.tableau.com

Complex Systems Framework (open-source decision technology): www.complexsystemsframework.com

Gephi (interactive visualization): gephi.github.io

NetLogo (programmable modeling environment): http://ccl.northwestern.edu/netlogo

Repast (open-source modeling and simulation platforms): repast.sourceforge.net

Vensim (simulation software): vensim.com