Missing Intentionality: the Limitations of Social Media Analysis for Participatory Urban Design
OverviewThis case study reflects upon some limitations of Urban Sensing, a research project funded by the European Commission, which explored the potential of social media analysis and data visualization as sources of knowledge for participatory urban design and management.1 The overall idea behind the project was:
• To analyze what city inhabitants and visitors publish on different social media channels (Twitter, Facebook, Foursquare, Flickr);
• To extract indicators on how these people perceive and live in the urban environment;
• To use this knowledge to feed more inclusive urban design processes (e.g., by measuring the real-time reactions of citizens towards new architectural interventions).
Urban Sensing built upon several existing projects, either conducted by research institutions (e.g., CASA at the University College London,2 Spatial Information Design Lab at the Columbia University3 or Senseable City Lab at the Massachusetts Institute of Technology4) or independent designers and design firms (e.g., Christian Nold,5 Art is Open Source6 or Stamen Design7). The trajectory of Urban Sensing was also influenced by the work of scholars coming from different disciplines, from geography, to urban studies, up to computer science (Zook and Graham 2007; Girardin et al. 2008; Kotov, Zhai, and Sproat 2011; Liu et al. 2011; Shi and Barker 2011). In a 24-month period, the team8 behind Urban Sensing created and tested a technological platform, which:
• Gathers data from 4 social media streams: Twitter, Facebook, Foursquare and Flickr;
• Applies multiple strategies (including text mining) to analyze these data and extract indicators related to several areas of interest (such as characterizations and perceived identities of geographic areas or users' feeling toward local policies and urban interventions);
• Visualizes the results, plotting them on a web-based map, like in the Figure 1, which represents the position of geo-located tweets in the city of Milan (Italy) in a two-week period (January 2012). Colors denote the eight most adopted languages while writing the tweets (yellow = Italian, green = English, bright green = Indonesian; pink = Spanish, light pink = French; blue = Dutch; light blue = Portuguese; and red = Japanese).
Figure 1 Screenshot produced during Urban Sensing and showing geo-located tweets in the city of Milan in January 2012 (Lupi et al. 2012)
Urban Sensing and its visualizations can be used by city designers, planners and administrators or accessed by a broader audience interested in urban dynamics. Imagine, for example, that some urban planners were working on a new master plan for the area of Bovisa, a district in Milan where the quite large Polytechnic University is located; they could use the Urban Sensing platform and identify the most crowded areas of this district by tracking the number of photos and contributions originated from or related to specific geographic locations (Flickr, Twitter, Facebook) or the number of check-ins in Foursquare for each venue in a predefined time-lapse. In Figure 2, Urban Sensing platform plots the geographic locations associated to Twitter contributions (blue dots) on a geographic map of Milan, also displaying the time trends (i.e., the number of contributions per day in the time span).
Figure 2 Urban Sensing, visual representation of the area of Bovisa, in Milan (August-September 2013)
Figure 2 shows how, during weekdays, the Bovisa university campus presents spikes of social media activity during lesson hours (9 a.m. to 6 p.m.), whilst the nearby UCI movie theatre and the shopping center are the stage for a high concentration of social media activity from 9 p.m. until midnight. This data can be further analyzed by tracking contributions from single users and—consequently—investigating how the users move across the city. In Figure 3, the blue dots represent the initial position of the users, whilst the green and red dots show the positions of the same users immediately before and after. By connecting the dots, we can clearly trace users’ movements over time and have an idea not only of the most crowded areas of Bovisa, but also of the locations where specific students come from before getting to the Polytechnic campus, and where they go after. In this sense, the analyses elaborated with Urban Sensing highlight some of the students’ patterns of use for this specific area of Milan.
Figure 3 Urban Sensing: Map of the Bovisa area, Milan (August-September 2013).
Urban Sensing aimed at investigating the potential of this and other more complex types of social media analyses and visualizations to support participatory urban design processes. The basic tenet was that through this kind of technological platforms some of the needs and desires of the city inhabitants and visitors could emerge and be heard by urban administrators, designers and planners. In the specific example above mentioned, the findings helped both Milan city administrators and real estate companies in identifying suitable areas to build student housing.
The limitations of Urban Sensing and the problem of missing intentionalityUrban Sensing was a research project also oriented to investigating the limitations of this kind of approaches, such as:
• Not all city inhabitants and visitors have equal access to technologies and skills to post geo-located contributions.9
• The accuracy of geo-located social media analyses is affected by the distribution of free WIFI networks. Especially tourists visiting foreign countries might not have data plans that allow a constant Internet connection. In these cases, they might still travel with their smartphone and use it, for example, to take pictures or notes to be shared at a later stage when they have access to a WIFI network (typically, either a free or public one or the one at their hotel). Obviously, the distribution of these WIFI networks in the city affects the geographic dimension of social media analyses, as a large number of contributions might emerge in areas where accessible WIFI networks are located.
Elsewhere, these limitations of Urban Sensing have been more thoroughly described (Ciuccarelli, Lupi, and Simeone 2014).
The focus here is the problem of the lack of intentionality.
Urban Sensing does not only collect users’ contributions related to the context of their use and perception of the city but also all kinds of contributions such as private comments or conversations with friends that are completely unrelated to urban issues. In most of the collected contributions, there is no clear intentionality from the users to post a tweet, share a picture on Instagram or check-in at Foursquare as actions to influence urban planning and management processes. Can this lack of intentionality undermine the potential of this kind of platforms as tools for more participatory processes?
Participatory design has been defined as a “process of investigating, understanding, reflecting upon, establishing, developing, and supporting mutual learning between multiple participants” who strongly contribute to the design activities (Simonsen and Robertson 2013, 2). In the specific context of Urban Sensing, this definition highlights the need for the users to become part of the design process (as participants) through a specific act of will. As of now, with the current instance of the platform, the users are somewhat passive and, in the vast majority of cases, do not even know that the platform exists and is monitoring them.
Some authors have warned against those technocratic approaches that praise the potential of urban informatics as a way of monitoring, controlling, and seamlessly operating the city (Mitchell 2005; Foth 2009; Greenfield and Shepard 2007; Ratti and Townsend 2011). As Ratti and Townsend argued: ‘‘Rather than focusing on the installation and control of network hardware, city governments, technology companies and their urban-planning advisers can exploit a more ground-up approach to creating even smarter cities in which people become the agents of change’’ (Ratti and Townsend 2011, 44).
Platforms and approaches such as Urban Sensing can easily become instruments of control and surveillance if the users are not actively involved as participants.
In order to tackle this risk, the users first need to be aware of these platforms and of how they can control their interaction with them (Galloway 2004). Secondly, it is important to set mechanisms of participation that guarantee that all city stakeholders have sustained access to these platforms as tools of expression, investigation and critique. It is only when the city stakeholders (a) are aware of the potential and the limitations of platforms such as Urban Sensing, (b) are in the conditions of actively participating and (c) their agency is framed by a clear intentionality, that they become agents of change and not passive recipients of top-down approaches.
Final remarksIn a way, Urban Sensing showed how social media analyses could support urban design, decision-making and administration, but at present there are still serious shortcomings for these approaches to be used as a tool of collaborative intervention. The lack of intentionality on the users side is one of these limitations and undermines the potential of these approaches in terms of widened participation.
ReferencesCiuccarelli, Paolo, Giorgia Lupi, and Luca Simeone. 2014. Visualizing the Data City - Social Media as a Source of Knowledge for Urban Planning and Management. Milan, Heidelberg, New York, Dordrecht, London: Springer.
Foth, Marcus, ed. 2009. Handbook of Research on Urban Informatics : The Practice and Promise of the Real-Time City. Hershey PA: Information Science Reference.
Galloway, Anne. 2004. “Intimations of Everyday Life: Ubiquitous Computing and the City.” Cultural Studies 18 (2-3): 384–408. doi:10.1080/0950238042000201572.
Girardin, Fabien, Francesco Calabrese, Filippo dal Fiore, Carlo Ratti, and Josep Blat. 2008. “Digital Footprinting: Uncovering Tourists with User-Generated Content.” IEEE Pervasive Computing 7 (4): 36–43.
Greenfield, Adam, and Mark Shepard. 2007. Urban Computing and Its Discontents. New York: The Architectural League of New York.
Kotov, Alexander, ChengXiang Zhai, and Richard Sproat. 2011. “Mining Named Entities with Temporally Correlated Bursts from Multilingual Web News Streams.” In Proceeding of WSDM ’11, ACM International Conference on Web Search and Data Mining, 237–46. New York.
Liu, Xiaohua, Long Jiang, Furu Wei, Ming Zhou, and QuickView Team Microsoft. 2011. “QuickView: Advanced Search of Tweets.” In SIGIR ’11 Proceedings of the 34th International ACM SIGIR Conference on Research and Development in Information Retrieval, 1275–76. New York.
Lupi, Giorgia, Paolo Patelli, Luca Simeone, and Salvatore Iaconesi. 2012. “Maps of Babel. Urban Sensing through User Generated Content.” In Proceedings of Human Cities Symposium. Brussels (BE).
Mitchell, William John. 2005. Placing Words. Cambridge Mass.: MIT Press.
Naor, Moni, and Moti Yung. 1989. “Universal One-Way Hash Functions and Their Cryptographic Applications.” In ACM Symposium on Theory of Computing. New York.
Ratti, Carlo, and Anthony Townsend. 2011. “The Social Nexus.” Scientific American September: 42–48.
Shi, George, and Ken Barker. 2011. “Thematic Data Extraction from Web for GIS and Application.” In IEEE International Conference on Spatial Data Mining and Geographical Knowledge Services, 273–78. Fuzhou, China.
Simonsen, Jesper, and Toni Robertson, eds. 2013. Routledge International Handbook of Participatory Design. New York: Routledge.
Zook, Matthew, and Mark Graham. 2007. “Mapping DigiPlace: Geocoded Internet Data and the Representation of Place.” Environment and Planning B: Planning and Design 34 (3): 466–82.