Our knowledge of how cities bring together different social classes is still limited. Previous studies have mainly focused on quantifying residential segregation, mostly over well-defined social groups (e.g., race). Little is known about the impact of mobility and human communications on urban social integration. The dynamics of spatial and social-network segregations and individual variations across these two dimensions are largely untapped. In this work, we introduce a computational framework ¾ by coupling large-scale information on human mobility, social-network connections, and people’s socio-economic status (SES) ¾ to enable a new way of understanding spatio-temporal and social-network segregation in cities. The framework can be used to depict segregation dynamics down to the individual level, and also provide aggregate measures at the scale of places and cities, and their evolution over time.

Teaming up with the MIT Sensenable City Lab, we conduct a case study in Singapore, by leveraging housing price and income data, along with a large mobile phone dataset that captures the mobility and cellphone communications of massive urban populations (project website: http://senseable.mit.edu/singapore-calling). We find that segregation is more pronounced among relatively wealthier classes, a finding that holds in both physical and social-network space. We also highlight the joint effect of distance decay and ‘homophily’ as forces in shaping human communication across socioeconomic classes. The time-resolved analysis reveals the changing landscape of urban segregation and the time-varying roles of places. Segregations in physical and social-network space are weakly correlated at the individual level but highly correlated at the group level. Various social and policy implications are discussed. We hope this framework provides a new perspective to understanding neighborhoods, human dynamics, and their linkage with socioeconomic environments of cities.

During the long boom years in London it was possible for many to overlook the fact that an increasing proportion of properties—and new build flats in particular—were falling further and further below the Nationally Described Space Standard (NDSS). However, the argument has been made that young people in particular were happy to trade housing space for the leisure and employment opportunities on their doorsteps. The Covid-19 pandemic, and ensuing lockdowns and economic turmoil, have exposed the shallowness of this ‘bargain’: homeworking and living in a 30m² space—smaller than most budget hotel rooms—simply isn’t tenable on a long-term basis.

The increasing densification and shrinking of housing in the capital has then reinforced existing socio-spatial divisions between those who have proved to be well-off enough to have a garden space and extra room able to be converted quickly to a home office, and those living cheek-by-jowl in over-crowded housing with only poor-quality (or no) external spaces for recreational use. Making matters worse, the Government has expanded Permitted Development Rights (PDRs) to allow the quick conversion of office and some types of industrial units to residential without the need for consent from the council and, consequently, the ability to enforce even minimal standards. In 2020, following the collapse of many retail chains, this was extended to make it easier for developers to convert former shop units to housing.

Because many of these conversions are being done without consent they are, consequently, very difficult to track through any of the normal reporting channels: planning applications tracked by the Greater London Authority, council-level approvals, or other regulatory processes. However, the requirement for sellers and landlords to produce an Energy Performance Certificate (EPC) at the unit address level holds out the promised a unique insight at scale into the internal size and configuration of properties in London; it can also, in principle, be matched to other sources such as the Land Registry’s Price Paid Data (PPD). For properties for sale, the requirement for an EPC came into effect in 2007, and for properties ‘to let’ the process began with new tenancies in 2008. Significantly, initial EPCs were valid for 10 years, meaning that the first round of certificates began to expire in 2017, triggering the need for new ones to be issued as properties were sold or re-let, though some lease extensions were allowed to continue without updates. In this paper we present an initial quantitative analysis of the geography of micro-homes in London, working through 2.8 million data points to identify important trends, recent developments, and important implications for the future in a Covid-19 context.

Predicting neighborhood change is often a prelude to policy implementation. But if some group of people approves of a policy they will probably support it, whereas if they do not approve they will probably sabotage it. Hence policy implementation might fail whenever there is no anticipation of people's assessment of different, alternative policies.

Enter the 'Planticipate' app. For many years it has employed machine-learning methods to forecast how different sorts of people will rate different policies. It scientifically improves socially-sensitive policy implementation within any situation comprising a goal plus up to five alternative policies for achieving that goal.

The app relies heavily upon regression statistics to place error margins around its forecast policy scores to plot the presence or otherwise of statistically significant differences between them; it uses Bayesian methods to estimate the reliability of its forecasts so far, and it uses innovative ‘face charts’ to suggest possible reasons for different groups’ different policy assessments. It also incorporates a simulated neural network to make comparable and probably more accurate forecasts, thereby generating more subtle insights into different people's different policy priorities.

Forecasts of policies' desirability can be made on behalf of up to 93 Census-derived categories of people - old, young, male, female, living in Australia, professionals or whatever.

In this paper ‘Planticipate’ will be detailed and demonstrated. It will then be used to estimate probable group attitudes towards several neighborhood-management policies in order to get a better fix on each policy’s likely reception within different parts of the community.

Our app continues to be extremely hungry for more users. This is because its self-improving, machine learning-based structure means that its forecasts become more and more accurate the more that it is used. Symposium attendees are, therefore, encouraged to use this freely available app to help train ‘Planticipate’ and so further boost its accuracy while simultaneously improving their handling of any policy-implementation situations with which they are currently involved.

Urban economists have put forward the idea that cities that are culturally interesting tend to attract “the creative class” and, as a result, end up being economically successful. Yet it is still unclear how economic and cultural dynamics mutually influence each other. For the first time, we operationalize a neighborhood's cultural capital in terms of the cultural interests that pictures geo-referenced in the neighborhood tend to express. This is made possible by the mining of what users of the photo-sharing site of Flickr have posted in the cities of London and New York over 5 years. We find that the combination of cultural capital and economic capital is indeed indicative of neighborhood growth in terms of house prices and improvements of socio-economic conditions. Culture pays, but only up to a point as it comes with one of the most vexing urban challenges: that of gentrification. Interactive maps of the two cities, publication, and datasets are available under http://goodcitylife.org/cultural-analytics/project.php.

Attempting to jumpstart investment in 10 “tipping point” neighborhoods in Detroit, the Strategic Neighborhood Fund (SNF) was created as a collaboration between a local community development funding institution and city government. Soliciting philanthropists, banks, and foundations to invest in real estate and social capital, the fund has attracted much attention yet offers investors little direction of how the money might be invested, and the public minimal information about the government’s intentions. However, there is concern among the neighborhoods that this fund will only further accelerate neighborhood change and lead to the negative effects of displacement from the eventual gentrification of the areas. This research develops a technical process that attempts to predict neighborhood change – as measured by indicators of socioeconomic “wellbeing” – and investigates the ethical challenges inherent in such a process. The technical component utilizes publicly available data to predict changes in socioeconomic status in Detroit neighborhoods from 2012 to 2017 utilizing machine learning techniques and a variety of methods for identifying socioeconomically “ascending” census tracts. The research investigates how these data can shed light on Detroit’s socioeconomic changes since its declaration of municipal bankruptcy, if there is any predictive power to this data, and what the ethical ramifications of such quantitative assessments might be. Can data analysis and algorithms predict neighborhood change – gentrification or decline? Should such processes be utilized in the policymaking realm? This paper also presents an argument against the use of such algorithm alone as a decision-making mechanism, especially without first working within the communities that might be most affected by its implementation in policy or investment decision-making.

Although this research highlights the utility and risk of a machine learning algorithm in a limited scope, it identifies the stakes of what more powerful machinery can do, and lays the groundwork for a discussion of what anticipatory forms of governance might be necessary to ensure that needs of the community are met in the process of urban revitalization. Though beyond the scope of this research, these technologies hold the potential to bypass the democratic process to determine resource allocation and organize the urban landscape in a reflection of the algorithm, replete with the assumptions and biases built into the process. This research identifies these assumptions, biases, and the risks, though a more complete discussion of the risks in relation to democracy and public decision-making is warranted in future work. Methodologies such as this could be used in a variety of government settings, and it is necessary to investigate these processes in each instance to ensure that biases are not engrained in the algorithms, and that the algorithms are not utilized as the primary decision-making tool.