This presentation takes a reflective look on the past decade-plus of thinking about, and working with crowd-sourced geo-data in geography and urban studies including our own and others contributions during that period. In doing so, we highlight four distinctive time periods starting around 2004. The first period, Web 2.0 Catalyst (2004-2009), represents the initial wave of interest in new internet applications that broaden the ability to create online content (e.g. Facebook, Twitter, Google Maps). During this period many urbanists extended earlier critical and theoretical arguments made about the mass adoption of the internet during the 1990s. After this initial critical engagement with crowd-sourced data the amount and breadth of geotagged digital content rapidly increases and the once simply hypothetical becomes practice. This helps launch a second period, Mapping Bonanza (2010-2014), characterized by overall excitement and hype about the possibilities of big geodata. Both popular and academic work during this period often engages these new data sources primarily by ‘simply’ mapping them as points and clusters on the landscape. This somewhat naive approach quickly drew criticism, some of which we summarized in a 2013 paper titled ‘Beyond the Geotag’, in which we outline ways these data can be applied in more nuanced ways. These sets of approaches represent a third time period, Beyond the Hype (2015-2019) that focuses on specific affordances of big data to enable relational, scalar, longitudinal and multi-dataset analyses. With the increased attention to and awareness of the privacy and ethics implications of using big geodata, we finally look ahead to the next phase of research, Back to Future (2020-?). As we argue, future work should draw more explicitly on the foundational goals (and lessons learned) of critical social science research. To help shape this next phase, we draw on our reflections to identify 7 ‘Back to …’ propositions.

Gentrification and neighborhood change are on the rise in the United States (US). In the 1990s only 9 percent of low-income census tracts in the top 50 US cities experienced economic transformation, while in the 2000s that figure jumped to 20 percent (Maciag 2015). With this noticeable uptick in the economic transformation of low-income neighborhoods, some scholars, such as Manuel Aalbers, suggest we have entered into a new phase of gentrification, the “fifth wave” (Aalbers in press). Whether the fifth wave of gentrification is qualitatively different from prior gentrification periods is debatable.

In this paper, I systematically investigate fifth wave gentrification drivers and consequences. To do this I first reconceptualize gentrification by integrating different definitions of the term since its introduction in 1964 by Ruth Glass (Glass 1964). One key definitional issue is whether residential displacement should be disentangled from the gentrification process (Brown 2017; Freeman 2005). I focus on this concern and discuss other forms of displacement, such as cultural and political, often associated with gentrification (Hyra 2017; Zukin 2010).

After this conceptual review and refinement, I analyze different gentrification drivers and consequences in the US in the fifth wave period (between 2000 and 2016), with special attention given to the Great Recession (2007-2009) and post-recession recovery (2010-2016) time frames, compared to prior gentrification wave periods (Hackworth and Smith 2001; Lees, Slater, and Wyly 2008). I highlight new gentrification drivers such as the mortgage market crash fallout and its relationship to the rise in the demand for rental units in low-income communities (Aalbers 2016; Sassen 2014). In terms of new consequences, I contribute to health implication debates for low-income residents who are able to stay in place during the process of gentrification (Hyra et al. 2019; Gibbons and Barton 2016). Data for the analysis are drawn from prior articles and reports, first-hand observations, and interviews with stakeholders in Washington, DC’s Shaw/U Street and Anacostia neighborhoods, as well as descriptive analyses of US Census statistics of property value, education levels, and racial demographic information, and Home Mortgage Disclosure Act lending and Housing Vacancies and Homeownership data.

This theoretical paper contributes several insights to the gentrification literature and the conference theme. First, I present a conceptual refinement of gentrification. Second, I enhance the conversation on whether we have entered into a fifth wave of redevelopment in the US, and elevate new drivers and consequences of neighborhood change in this contemporary redevelopment period. Third, I conclude by lofting a series of unresolved gentrification issues both theoretically and methodically, and propose several analytical, design, data collection, and analysis tools to deepen future studies of neighborhood change. I specifically focus on the challenges and opportunities of using “big data” and “machine learning” to better understand neighborhood change, gentrification, and multiple forms of displacement in the post-Great Recession period and beyond (Chapple and Zuk 2016; Reades, De Souza, and Hubbard 2018; Taylor, Poorthuisb, and Zook 2015).

Social behaviors emerge from the exchange of information among individuals—constrained by and reciprocally influencing the structure of information flows. The Internet radically trans- formed communication by democratizing broadcast capabilities and enabling easy and border- less formation of new acquaintances. However, actual information flows are heterogeneous and confined to self-organized echo-chambers. Of central importance to the future of society is understanding how existing physical segregation affects online social fragmentation [1]. We show that online interactions are segregated by income just as physical interactions are, and that physical separation reflects polarized behaviors beyond culture or politics. Our analysis is consistent with theoretical concepts suggesting polarization is associated to social exposure that reinforces within group homogenization and between group differentiation, and they together promote social fragmentation in mirrored physical and virtual spaces.

In order to analyze patterns of segregation, we built suburban networks of social interactions for multiple American cities. Nodes represent neighborhoods and edges indicate whether people visit or communicate with people who live in other neighborhoods. Edges are the social bridges that connect neighborhoods and transfer information across the city. Their structure shows the segregated interactions taking place in cities. The systematic breakpoints in social communication affect the spread of information and, since we learn from imitation, may also promote divergence of behaviors. The differentiation and polarization of behaviors due to the segregation of interactions is reflected in collective interests and topics of conversation on social media. We applied a topic model clustering algorithm to the matrix of neighborhoods’ hashtags. The topic model re- veals groups of hashtags that co-occur and can characterize topics of conversations. Topics usage across neighborhoods is positively and negatively correlated with the principal hashtag component and neighborhood median income, respectively.

The structure of the networks and hashtag space indicates that the poor live in isolated, peripheral and segregated neighborhoods, not only in the physical space but also online. If we consider that access to information is crucial for finding opportunities for economic and non-economic activities, the geometry of these social networks manifests the asymmetric nature of poor/rich relationships that are still present in the online space.

References

[1] Alfredo J. Morales, Xiaowen Dong, Yaneer Bar-Yam, and Alex Sandy Pentland. Segregation and polarization in urban areas. Royal Society Open Science, 6(10):190573, 2019.