Gowanus Atlas: Mapping the community-scale context and stories

Zhibang Jiang

Figure 1: Superfund sign of EPA
Figure 2: Gowanus Atlas


The surrounding area of the Gowanus Canal, a 1.8-mile long waterway in southwestern Brooklyn, is a typical example of the struggles around urban renewal and land rezoning. As a superfund (Figure 1) that everyone wants to keep away from, the Gowanus Canal is now considered to be promisingly revitalized through the remediation of a cleanup program invested by the US Environmental Protection Agency (EPA). In this context, Gowanus Atlas (Figure 2) is a non-profit project promoted by Gowanus by Design to tell the stories of the community around this canal through geographic layers. The project seeks an answer to the neighborhood’s demand for “connection, participation, and activism.”1

If you open the webpage now, I wouldn’t be surprised if you’re drawn to its comprehensive cartographic techniques, user-directed curation of map layers, and community-based storytelling. The vision of this project, as the organizers state, is to give people “access to quantitative and qualitative data about the area’s rapidly changing conditions, its challenges, and the community’s future.”1 Unfortunately, the atlas fails to fulfill these goals completely. Thus, this essay covers my four main criticisms towards its unfulfilled potential: (1) the hidden historical context that the project did not provide, (2) the unaddressed user engagement modes, (3) deficiencies as a user-(un)friendly media, and (4) the invisible side of connections between components on the interface. In response to criticisms, I also propose corresponding solutions or prototypes to improve the user experience of this atlas.

Interface Overview and Resolution

Figure 3: The resolution of the interface into components and a speculative relational data structure. The reader may need basic knowledge of database design as a prerequisite to read this figure.

Gowanus Atlas is trying to construct a virtual common space of geography, culture, living environment, infrastructure, and society around the Gowanus Canal. A quick overview shows that the website has a simple structure. As shown on the bottom in Figure 3, it consists of a canal-wide basemap and a navigation bar with three buttons: Stories, Build your own map, and About.

The Stories button enables users to discover 23 representative figures active within this community. When a user chooses a specific figure, i.e., Eymund Diegel here (, number in figure 3), a story panel with a rich-text introduction appears on the left side. And simultaneously, related Map Layers (), i.e., Demographics, Cleanup Plan, and Canal Watershed here, with thematic geo-coded data points cast on the basemap. Thus, the basemap, Map Layers, and the Stories of those figures are organically united – the figures could trigger the Map Layers. 

The Build your own map button is what a user can dive into when creating an atlas with their customized combinations of Map Layers. In the Build your own map dropdown list, all figures are gone, and myriad Map Layers are legibly classified into six categories (): hydrology, culture, environment, built environment, infrastructure, and social landscape. These categories also represent the fields of expertise of those representative figures (); to be more specific, the map layer and figure list follow an identical taxonomic (classificatory) scheme. This is how the Map Layers and the representative figures are interrelated in the interface.

In essence, this interrelationship is driven by the composite primary keys2 in the Category Table of the data model beneath the interface. The Category Table is a junction table. To represent a many-to-many relationship — i.e., a figure can have many map layers, and a map layer can belong to different figures — a junction table must be created to break down this relationship into two one-to-many relationships. However, the potential of the data structure in the database is unfulfilled in the interface. Why? That’s because the interface only reveals one of the two one-to-many relationships. I will discuss this in-depth later in the Invisible Side of Connections: From Database to Interface section.

Hidden Historical Context

As stated above, the Map Layers work together to tell stories across disparate fields. Unfortunately, the atlas has a notable gap in explaining the Gowanus Canal’s historical context for those stories. Although the canal is notoriously known to be very polluted these days, it is important to understand its historical background because “it has been central to the development of the city for centuries.”3 During its long history, “the name ‘Gowanus’ has invoked diverse meanings and definitions, from bountiful farmland to rough-and-tumble industrial thoroughfare.”4 However, Gowanus Atlas lacks an introduction to the canal’s cultural background and evolutionary history: How was the land being shaped by human activities? How has its industrialization, pollution, and commercial development in turn affected human beings? How did this canal degenerate and how will it be revived? 

Figure 4: Layers of history of the 1st Street Basin, from Glacial Outwash to Canarsee Tribes to Dutch Settlement and to Battle of Brooklyn2
Figure 5: three layers from top to bottom are combined sewer outfalls (CSO), flood plain and historic wetlands. 13 CSOs that drain over 350 million gallons of raw sewage into the canal each year2

Certainly, the atlas makers could say: “we just want to make a map of urban environments without the cultural side.” They may assume New Yorkers who live in or near, or who are familiar with the area are inclined to be familiar with this project. Nevertheless, the lack of context prevents a wider variety of people from being interested in the area and in the project because the probability that most visitors to the Gowanus Atlas website will live near the Gowanus Canal is minuscule. Prospective visitors would ask, “Why should I care about this atlas? Why is it useful? Could I find something related to myself?” Anyone who is unfamiliar with the ecological notoriety or development history of Gowanus will not know why this map is significant. If the atlas makers expect this project to be acceptable and usable for more people, they have to answer those questions and provide more context. 

Figure 6: a prototype to be as a form for carrying context (This is an excerpt from my previous work)

One solution to this would be creating proximity to generate context. This idea is borrowed from the field of journalism: nearby stories are more newsworthy than events that take place far away.5 Despite the fact that some people have no geographical proximity to the atlas, fortunately, proximity is multifaceted. Other dimensions of cultural6, domanial, or temporal proximity, among others, could still be incorporated to compensate for this deficiency. For example, if one person could pick a date in the timeline of their birth year and find out the atlas’s historical datasets of that year, then the temporal proximity would be built. In addition, I made a prototype (figure 6) to demonstrate a form to carry historical proximity. I added a series of hypertext panels, which are add-on layers above the basemap, to enrich the context. The panels are controlled by clicking or scrolling, and the transition (zoom in/out, movement etc.) of the basemap is also driven by panel switches — if you scroll down the panel will slide up and the basemap will move correspondingly. At each step, users could interact with the map to discover the topic described in the panel. This could be a landing page before the exploratory interface, which is a popular technique in interactive articles.7 

Additionally, it’s important to mention that we shouldn’t think that only people who are not familiar with the Gowanus community need proximity. Even though the local people probably have the understanding of the context, analogous techniques also could be embedded to enhance the sense of closeness. For instance, one person puts in the zip code of their home and nearby POIs automatically show up, which stirs up their desire for exploration.

In summation, addressing the state of proximity fosters feelings of familiarity or meaningfulness, and would be a practical way of tackling context deficiency and improving the experience for all types of users.

Unaddressed User Engagement Modes

Figure 7: Schramm’s model of mass communication8

If increasing “proximity” is about making users feel attracted before they use the atlas, then creating superior engagement modes is to make users enjoy using it, benefit from it, and become co-creators to shape it. The good thing is that the audience in exploring Gowanus Atlas is somewhat beyond a passive receiver, decoder, and reader. The interaction process here is similar to Schramm’s model of mass communication9 without the feedback path. The center of this model is the atlas makers who offer “many identical messages” and the audience consists of individual users who decode and interpret the messages respectively — they are empowered to manipulate the “messages”, to an extent, by selecting different layers and combining them into dissimilar atlases. In my opinion, though, the Atlas could offer more options for users to manipulate it and to be manipulated by it.

Figure 8: the four potential modes of user engagement I summarized.10 The USC & UGC modes mainly fit in the Human-To-Atlas (Screen) relationship model, and are very subtle to play roles in the Atlas (Screen)-To-Human one, which is represented by the 2-step color in magnitude.

How users manipulate the atlas

The atlas does have some good explorations in this direction. It takes the form of a user-selected content (USC) mode that allows users to Build your own map, but the shortcoming is that users can only select pre-existing content and adjust the display of a limited number of options. They can’t add their own content or create their own Map Layers. The project’s stated intention is to represent a changing world, but it uses static content. In its current manifestation, the users don’t know when the data were collected and what periods the Map Layers represent. To achieve its goal, it needs to be more dynamic. It would be fascinating to open part of the datasets for residents — and even professional cartographers and planners — to generate their own content (UGC) in a crowdsourcing way. As to the data quality, the map-makers could provide a framework as collection guidelines or release some thematic templates to ensure that the content is relatively structured and follows basic order. Also, the content could be subjected to a peer-review, and then turned into time-stamped data. With crowd-sourced, time-stamped data, the Atlas could live up to its claim to reflect the area’s rapidly changing conditions and the community’s future, creating society-level relationships with a larger spatial-temporal scale and a wider range of beneficiaries.

How the atlas manipulates users

In terms of having the users be manipulated by the atlas, this atlas is missing opportunities to do more. Whether the USC or UGC mode, it could be mainly considered constructing human-to-atlas (screen) relationships. But are these two modes able to build atlas (screen)-to-human relationships? Yes, to some extent, but in a very subtle way. In the USC mode, Gowanus Atlas responds to you in real time and shapes your thinking about the atlas, but not actionable enough. In a hypothetical UGC mode, the content generated by users could affect the next people who see the atlas; however, the transmission chain is too long and slow. Moreover, both of these two modes confine people to the computer, and the user’s activity is actually a face-to-computer experience in isolation. Futuristic interaction modes should not be limited to the relationship between people and computers, but should ultimately benefit the relationship between people and nature, and people and people. Therefore, it is reasonable to introduce a more actionable, immersive, fun and engaging human-atlas (screen)-human experience (figure 8):

  • One interesting possible human-atlas (screen)-human experience could be a 3D street navigation map that shows POIs nearby after the user designates a personalized walking route. When a user specifies a trajectory in Gowanus Atlas as an input, the system would generate a navigation that calculates, maps, and displays POIs in a spatial and real-time way as an output. This leads users to take the datasets from a digital world and bring them into the offline world, enabling them to feel the data in the real world, to find real locations — art galleries, parks, bus stations etc. — and meet real people. This mode uses the atlas (screen) as a bridge, ultimately moving the computing environment out of the screen and transitioning from a WIMP-based user interface to the reality-based interaction.11 This encourages users to focus more on human-to-human communication than human-computer-interaction. User activities are extended from an isolated experience to a social experience.
  • Another scenario would be to construct a collective experience by enabling multiple users to explore this atlas at the same time. That could bring instant feedback for all users and expedite the course of knowledge exchange, which is the shortcoming of the UGC mode I mentioned above. The usual weakness of UGC is that it can take a long time for users to add content, have that content in circulation, and let that content benefit others. A multi-user exploring mode could speed that process up. And in the multi-user interaction process, the human-to-human communication would be enhanced.

Therefore, there are many ways that the atlas makers could have increased user engagement, although the Gowanus Atlas’s existing features are acceptable.

Deficient Functionality: Map as a User-(un)friendly Media

Figure 9: some options are mutually exclusive, for example in this figure Commercial Land Use is selected while five other options are disabled
Figure 10: Elements of user experience12

From this section, let’s draw our attention away from the imaginary topics of what they could do to practical issues of what they should do. As a designer, user experience is also an ineradicable perspective from which I inspect this work. For the atlas, I appreciate the abundance of cartographic techniques that are applied within the space. The combination of points, lines, and surfaces creates informative, visually appealing Map Layers. Even better, the atlas makers were aware of the potential interaction and readability issues that could result from layer mixing and thus built mutually exclusive logic rules (figure 9) among different layers. That said, however, room for improvement exists in nearly all projects, including this one. To better illustrate its deficiencies, I use the Elements of User Experience (figure 10) framework to separate the different layers in Gowanus Atlas. Jesse James Garrett defined five planes of user experience. He argues a designer must “fulfill your strategic objectives while meeting the needs of your users”13 to achieve successful interaction design. Hence, I will cover and explain the Surface and the Scope Plane in this section and the Skeleton Plane in the next section, revolving around the user needs.

Recognition Versus Recall

In the Surface Plane, or appearance of the finished product, many (not all) of the maps in this atlas lack legends. Although users can be informed of the corresponding tooltips through interaction, such as hovering action, the visible state is inconstant, which requires memory and cognition resources to recall rather than recognizing.14 This situation is even worse in scenarios of numerous layers without legends, where too many data points will be mixed up and overlapped to be identified. In addition, legends as redundancy can also reduce users’ resistance to unfamiliar knowledge, improve the precision in communication15, benefit the effective delivery of the message”16, thus amplifying the effectiveness of the geographic visualization and users’ understanding.

Incomplete User Flow: No Saving and Sharing Features

In the Scope Plane, or the functionality the site must include to meet user needs, I don’t see any features that would allow users to save or share the atlas they have created. Users may want to save a specific set of Map Layers for future reference, or exchange what they have learned with other people who are interested in the Gowanus Canal. Without these features, users are unable to make the transition from new facts to actionable insights.

Invisible Side of Connections: From Database to Interface

Continuing with the Elements of User Experience framework, there is an issue in the Skeleton Plane – that is, about the interface design of the site. According to Lev Manovich, a researcher in media theory, database is a collection of disorder items with structured data models (here I reverse engineered a relational data model in figure 3); the interface creators could translate the same database, as the material, into disparate interfaces that may or may not fully employ the database models.17 Unfortunately, the makers of Gowanus Atlas, as the interface creators, failed to fully reflect or employ the database’s structure.

Why did they fail? Let’s first move to the Data Model in figure 3. The Junction Table breaks down the many-to-many relationship between the Representative Figure and the Map Layer table into two one-to-many relationships: one figure could have many map layers and one map layer could belong to many figures. Take class enrollment as a similar example, a student can have many classes, and a class can have many students. Therefore, a figure here has many Map Layers, and a Map Layer may belong to multiple figures. This is what we have in the database.

Now, take a look at the interface in figure 3. The fact “the figures could trigger the Map Layers” that I mentioned in the Interface Overview and Resolution section is only one of the two sides of connections in the data model — the other side is implicit and invisible, although the database has the ability to be employed. In other words, the interface only implemented one one-to-many relationship and failed to allow users to trigger the figures by Map Layers. When users are looking at a Map Layer, they miss the opportunity to know who is related to this layer.

Figure 11: the connections from the figure to the Map Layers. This is a prototype based on the existing feature, thus the relationships between figures and map layers are further manifested.
Figure 12: the connections from the Map Layer to figures. This is a prototype of the hypothetical feature, which would employ the other side of connections in the database.

Improving this problem could be made through a visualization component that includes navigation and quick selection features, which would integrate the entire system. With this new component, a user can click on a person’s name (figure 11) to see how many and what Map Layers that figure is connected to; users could also click on a specific Map Layer and see how many and which figures are connected to it (figure 12). The story panel could be retained in the secondary level of this component with breadcrumbs18, constituting a consistent hierarchy within the same component. Only in this way can the Gowanus Atlas foster the building of users’ mental model in understanding the connected nature between the Map Layers and the representative figures.


“Maps have never been about presenting data, but have always been about the story.”19 This essay has examined some strengths and weaknesses of the Gowanus Atlas in storytelling, user engagement, system functionality, and interface design. It also attempts to propose corresponding solutions to facilitate communication and interaction better.


1 Gowanus By Design, “Gowanus Atlas,” https://atlas.gowanusbydesign.org, Accessed March 23, 2021.

2  Very succinctly, a key in a relational database can be understood as an attribute which helps you to identify a row in a table. It allows you to find the relation between two tables. Here, composite keys are created when two primary keys from the Representative Figure and Map Layer tables are present as foreign keys within the Category Table. The foreign keys, here are Figure_ID and Map_layer_ID, are used together to uniquely identify each record in the Category Table.

3 Christopher O. Anderson, The Turning Point: A Focused Design Study for the Gowanus Canal, 2017 ASLA Student Awards, 2018.

4 Joseph Alexiou, Gowanus: Brooklyn’s Curious Canal, 2nd ed. (repr., NYU Press, 2020).

5 Monika Bednarek and Helen Caple, The Discourse of News Values: How News Organizations Create Newsworthiness (repr., Oxford University Press, 2017).

6 Joseph D. Straubhaar, “Beyond Media Imperialism: Asymmetrical Interdependence and Cultural Proximity,” Critical Studies in Mass Communication 8, no. 1 (1991): 39-59, doi:10.1080/15295039109366779.

7 Fred Hohman et al., “Communicating with Interactive Articles,” Distill 5, no. 9 (2020): e28, doi:10.23915/distill.00028.

8 Wilbur Schramm and Donald F Roberts, The Process and Effects of Mass Communication (repr., Urbana, IL: University of Illinois Press, 1977).

9 Schramm and Roberts, Mass Communication.

10 Robert JK Jacob et al., “Reality-Based Interaction: A Framework For Post-WIMP Interfaces,” In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2008, 201-210.

11 Jacob et al., “Reality-Based Interaction,” 201-210.

12 Jesse James Garrett, The Elements of User Experience: User-Centered Design for the Web and Beyond, 2nd ed. (repr., Berkeley, Calif.: New Riders, 2010).

13 Garrett, Elements of User Experience.

14 Jakob Nielsen, “Enhancing The Explanatory Power Of Usability Heuristics,” In Proceedings of The SIGCHI Conference on Human Factors in Computing Systems, 1994, 152-158.

15 C. E. Shannon, “A Mathematical Theory Of Communication,” Bell System Technical Journal 27, no. 3 (1948): 379-423, doi:10.1002/j.1538-7305.1948.tb01338.x.

16 Michelle A. Borkin et al., “Beyond Memorability: Visualization Recognition and Recall,” IEEE Transactions on Visualization and Computer Graphics 22, no. 1 (2016): 519-528, doi:10.1109/tvcg.2015.2467732.

17 Lev Manovich, “Database as Symbolic Form,” Convergence: The International Journal of Research into New Media Technologies 5, no. 2 (1999): 80-99, doi:10.1177/135485659900500206.

18 The breadcrumb is a secondary navigation pattern that helps a user understand the hierarchy among levels and navigate back through them. (Source: IBM Carbon Design System)

19 Mathieu Guglielmino, “The Rhetoric behind the Coronavirus Propaganda Maps,” Nightingale, The Journal of the Data Visualization Society, 2021, https://medium.com/nightingale/the-rhetoric-behind-the-coronavirus-propaganda-maps-3cd6ec84aa63


Alexiou, Joseph. Gowanus: Brooklyn’s Curious Canal. 2nd ed. Reprint, NYU Press, 2020.

Anderson, Christopher O. The Turning Point: A Focused Design Study for the Gowanus Canal. 2017 ASLA Student Awards, 2018.

Bednarek, Monika, and Helen Caple. The Discourse of News Values: How News Organizations Create Newsworthiness. Reprint, Oxford University Press, 2017.

Borkin, Michelle A., Zoya Bylinskii, Nam Wook Kim, Constance May Bainbridge, Chelsea S. Yeh, Daniel Borkin, Hanspeter Pfister, and Aude Oliva. “Beyond Memorability: Visualization Recognition and Recall.” IEEE Transactions on Visualization and Computer Graphics 22, no. 1 (2016): 519-528. doi:10.1109/tvcg.2015.2467732.

Garrett, Jesse James. The Elements of User Experience: User-Centered Design for The Web and Beyond. 2nd ed. Reprint, Berkeley, Calif.: New Riders, 2010.

Gowanus By Design, “Gowanus Atlas,” https://atlas.gowanusbydesign.org, Accessed March 23, 2021.

Guglielmino, Mathieu. “The Rhetoric behind the Coronavirus Propaganda Maps.” Nightingale, The Journal of The Data Visualization Society, 2021. https://medium.com/nightingale/the-rhetoric-behind-the-coronavirus-propaganda-maps-3cd6ec84aa63.

Hohman, Fred, Matthew Conlen, Jeffrey Heer, and Duen Chau. “Communicating with Interactive Articles.” Distill 5, no. 9 (2020): e28. doi:10.23915/distill.00028.

Jacob, Robert JK, Audrey Girouard, Leanne M. Hirshfield, Michael S. Horn, Orit Shaer, Erin Treacy Solovey, and Jamie Zigelbaum. “Reality-Based Interaction: A Framework for Post-WIMP Interfaces”. In Proceedings of The SIGCHI Conference on Human Factors in Computing Systems, 2008, 201-210.

Manovich, Lev. “Database as Symbolic Form.” Convergence: The International Journal of Research into New Media Technologies 5, no. 2 (1999): 80-99. doi:10.1177/135485659900500206.

Nielsen, Jakob. “Enhancing the Explanatory Power of Usability Heuristics.” In Proceedings Of The SIGCHI Conference On Human Factors In Computing Systems, 1994, 152-158.

Schramm, Wilbur, and Donald F Roberts. The Process and Effects of Mass Communication. Reprint, Urbana, IL: University of Illinois Press, 1977.

Shannon, C. E. “A Mathematical Theory of Communication.” Bell System Technical Journal 27, no. 3 (1948): 379-423. doi:10.1002/j.1538-7305.1948.tb01338.x.

Straubhaar, Joseph D. “Beyond Media Imperialism: Asymmetrical Interdependence and Cultural Proximity.” Critical Studies in Mass Communication 8, no. 1 (1991): 39-59. doi:10.1080/15295039109366779.

Leave a Reply

Your email address will not be published. Required fields are marked *