Seeing Differently: Cartography For Subjective Maps Based On Dynamic Urban Data

May 23 2011

This is an excerpt from Introduction and Chapter 1: The Origin and Politics of Subjective Maps my master’s thesis Seeing Differently: Cartography For Subjective Maps Based On Dynamic Urban Data

A Challenge for Planners and Designers to Understand Contemporary Cities

City planners, urban designers and architects all work intensively with maps. Different maps may profoundly affect the outcome of their works. On one hand, their knowledge of the cities with which they are working is largely obtained by reading both topographical and thematic maps. Their mental models are shaped by the information presented by the maps. On the other hand, with form being one of the core conditions as well as deliverables of design, it is expected that in most established design principles and methodologies a considerable amount of decision-making relies on visual reflections, such as scale, proportion and contrast. All of these factors are dependents on the choice of map representations.

Figure 0.1 Figure 0.1 Ichnographic maps replacing oblique projection motivated the study of urban form

Evidence is given by the significant influence of ichnographic city plan. Emerging during the sixteenth century, as opposed to the widely used perspective maps (“birds-eye view”) at the time, ichnographic plans are “presented as if viewed from an infinite number of view points.” (Pinto, 1976: 35). Among them were Leonardo Da Vinci’s Plan of Imola (1502) and Leonardo Bufalini’s Plan of Rome (1551), which were the first city plans based on surveying result and drawn strictly to scale (Harvey, 1980: 155, 167). The techniques and graphical abstractions developed while producing these maps are believed to have stimulated the interest in “the city as an ideal architectural form” in the Renaissance (Pinto, 1976: 50). It is also not a coincidence that no studies of spatial urban form were identified until the nineteenth century , when accurate ichnographical city maps became widely available following Giambattista Nolli’s iconic works of Rome. The term “urban cartography” came into use after World War II to deal with the mapping of formal, historical and functional conditions of cities (Miller, 2000: 7), just when the patterns of town planning experienced a remarkable development.

For the past two decades, the nature of city has been undergoing an enormous shift powered by information technologies. GPS assisted personal mobility changed travel patterns. Cellphone networks and the Internet enabled ubiquitous and constant access to virtual communication and created time-space flexibility regardless of physical presence. An abundant variety of enthusiastic writers tried to embrace the new era by predicting how these technologies would change city planning and the design of space. It was observed that public place had become meaningless to contemporary citizens (Sennett, 1992), and that the reinvention of public spaces and cities were a pressing mission (Mitchell, 1999). Technical writer George Gilder radically stated that, with the continued growth of personal computing and distributed organizations advances, “we are headed for the death of cities… cities are leftover baggage from the industrial era.” (Gilder, 1995) He was proved wrong. Not only cities have survived, but also urbanization on the global scale is growing at a faster pace than ever before . More recent arguments seem to have become pessimistic, claiming that due to the indifference nature of information technology to space, spatial reconfiguration is no longer required (Kellerman, 2009).

However, many of such analyses were based more on subjective judgment rather than solid observation of how urban citizens live their life. Interestingly, map is a rare element in these books about cities. It is understandable because the maps that planners, architects and researchers currently use do not display much of the ongoing changes. The maps still aim to reveal the same static geographic information with similar visual elements that have been used since Nolli’s time. There is little human activity in these maps. As the major means of communication between designers and cities, they fail to convey the information of an important layer. As a result, planners and designers are tending to think and work in the same way as they did in pre-information age.

It is not that there are no ways to examine the new urban phenomena quantitatively. With the increasing deployment of sensors and hand-held electronics in recent years, it is feasible at a relatively low cost and even becoming common practice to monitor transportation status, air quality or people’s locations and communications at a resolution that has never been achieved before. Moreover, the Internet and its open platform technologies are allowing the general public to access massive data with ease. In contrast with the thriving open data movement , is the surprising fact that few of these resources are integrated effectively into the work process of planners or architects.

The real obstacle preventing people from understanding contemporary cities is information overflow. According to International Data Corporation, the new data produced globally each year is growing at an annual rate of 60% (The Economist, 2010). Comparing to the collection of data, it is much more challenging to process and interpret the meaning of them. Most visualizations of urban data we see today are static, highly aggregated overlays on top of a traditional map, which do not reflect the richness, the dynamic and networked characteristics of the data. As the time this thesis is written, there is not yet a comprehensive solution for processing and visualizing massive urban data, or systematic discussion on how these data should be visualized.

Motivation and Purpose of Study

In the many years as a student in architecture and urban design, I lost count of the city maps I have drawn: plans of my site with its urban context, figure-background studies of urban fabric, diagrams of relationship between landmarks and so on. These were methodologies in which I was trained to work. I found myself repeatedly doing the same thing: trying to conjure something out of the pure geometries of a city plan. In a profession dedicated to engineering space for life, I was constantly frustrated by the absence of “real life” in my maps.

In 2009 I joined Senseable City Lab, MIT and started to work with high quality, in-house urban data of environmental monitoring, mobile communication, energy consumption, transportation and health. It had been exciting, at first, to place the raw data as colored dots and bars moving around the map, which was the most commonly seen visualization of the day. The most comment I got for such visuals was: “This is pretty, but what does it tell me?” That was the time I became fascinated by the question how visualizations can make urban data truly meaningful to the viewer. During the experiments I looked back to what I did before all the numbers and equations and programs, and rediscovered the most powerful element of a map: the topological geometries.

This thesis proposes dynamic subjective maps as the key to bridge the gap between man and massive urban data. It will argue why such maps are effective and describe the basic design principles. It will explore ways to (1) encode meaning through images with latest computation techniques; (2) enable exploration through human-computer interaction; (3) combine cartography with real-time data flow. A selection of case studies will demonstrate the design process in two typical contexts, the implementation workflow as well as their acceptance by the audience. They will be followed by a discussion of reflections and future directions.

By no means are the maps presented in this thesis intending to solve any design problems directly. They are not suggestions of ideal forms, but of representations. These maps show possibilities of depicting cities from perspectives that are rarely seen before, making use of the latest computation technologies. They are expected to inspire planners, architects and the general public to reflect upon their understanding of cities.

I see these projects as a starting point of the adventure into a new world of computer aided cartography. All the code I produced in making these visualizations will be available online. I hope they will provide a useful toolkit for anyone to explore his own data with and will inspire further innovations.

Objectivity vs. Subjectivity in Maps

A map is a provider of geographical information. Comparing to other types of visualizations, maps carry a special sense of authority, expected by their audience to be both accurate and objective. Therefore, objectivity is the goal of cartography. The evolution of cartography is a history of inventing devices that capture more accurate data and methods that make more accurate reproductions of them.

Figure 1.1 Figure 1.1 World Map of Idrisi 1154 AD, Charter Rogeriana / restored and edited by Konrad Miller

During the years when complete survey was impossible, the absence of objective data was compensated by subjective knowledge. Fig 1.1 shows the world map composed by Arabic geographer Muhammad Al-Idrisi, which is considered the most accurate world map in pre-modern times . The map was a compilation of knowledge of Africa, Indian Ocean and the Far East gathered by merchants and explorers. The coastlines were distorted by the travel time and experience it took to sail around them, thus embedded the subjective view of the world by the voyagers.

Today, with satellite photography, GPS locating and geographical information systems (GIS), we are able to acquire and plot geographical data in high fidelity on the global scale. People are now confident to say that the earth is more or less “fully charted”. Maps are generally considered purveyors of objective reality. That, sometimes, makes us overlook the fact that there has always been a place for subjectivity in maps.

Firstly, bias naturally exists in data sources due to the affordability of survey investments or conflicted interests of parties. Warren observed “Blank spots” in maps of an abundance of regions with lower global economic relevance, which appear much blurrier in satellite imagery and lack vector-format descriptions that are necessary for many geographical analyses (Warren, 2010: 20). Statistical data collection like censuses are conducted by arbitrarily picked blocks, which lead to better resolution results for regions with denser population . When it comes to politically disputed boundaries, such as those in Pakistan and Western Sahara, “objective truth” does not exist .

Secondly, it is inevitable to make subjective choices in making maps from geographical data. Take the world map as an example. According to Gauss’s Theorema Egregium, the spherical earth cannot be represented on a plane without distortion. Choosing a projection method usually means choosing an origin and an associated distortion. The most commonly seen world map uses the Mercator cylindrical projection, which preserves directions but inflates the size of regions as their distance from the equator increase. It wins popularity over equal-area projections (Fig 1.2) for its public familiarity and easy-to-understand regular grids, while has been criticized as promoting “serious, erroneous conceptions by severely distorting large sections of the world” . The Mercator map can be widely acceptable because most population of the world lives off the pole area and are indifferent of the distortion happening to the rim of their world . Similar self-centered choices include that European and United States versions of the world map split at the Pacific, while Chinese and Japanese versions split at the Atlantic; Australians make upside-down maps to protest against Northern Hemisphere bias . Being placed in the center or “top” of the world has a significant psychological effect . These are choices that are majorly rooted in culture rather than science.

Figure 1.2 Figure 1.2 Gall-Peters, Lambert azimuthal equal-area and Mollweide projection of the world. All maps are Eurocentric.

Graphical simplifications and adjustments must be made by the mapmaker to visually serve its readers better. These decisions sometimes override the faithfulness to geographical reality. The term of “cartography generalization” addresses the need to adapt representation to the scale and display medium of the map. Symbols may be drawn in a scale size much bigger than the features they represent. Only a selective, incomplete set of fact is shown to avoid burying critical information in overwhelming details. It is the cartographer’s judgments what to include or exclude and whether to displace objects from ground truth to make place for annotations. In more radical cases such as subway maps, scale varies and directions contort, as geographical features give way to the demand of emphasizing the topological hierarchy (Fig 1.3). Such subjectively revised maps have a strong impact on commuters’ cognition of a city by shaping their mental model of distance and relative position between places.

Figure 1.3 Figure 1.3 Tube Map of London: geographically correct map vs. standard map

Because there are no standards or boundaries for the subjective choices described above, maps can be psychologically misleading due to ill design or deliberate manipulation. As Monmonier stated in his book How to Lie with Maps, “A single map is but one of an indefinitely large number of maps that might be produced for the same situation or from the same data.” (Monmonier, 2007: 2) The impression by a thematic map can be distinctively different when a set of features is omitted, or the breakpoint between categories is moved, or a new color palette is applied. In Fig 1.4, the US 2008 presidential election votes by county, red (Republican) dominates blue (Democrat), while the election result shows otherwise. It fails to allow for the fact that the population of the red counties is significantly lower than that of the blue ones (Newman, 2008). In political propaganda and commercial advertisement, mapmakers constantly select out contradictory information and use colors, symbols and size distortion in projections to dramatize or conceal spatial patterns.

Figure 1.4 Figure 1.4 A misleading map: US 2008 presidential election votes by Mark Newman

Subjective Maps: from Distortion to Story Telling

Objective map is a myth. Cartography critics consider subjectivity in maps an everlasting enemy to fight off. Books of mapping techniques evaluate all projection methods by how much distortion they produce, based on the belief that the more facts a map preserves the better. Yet maps serve purposes beyond geographical reality. J. B. Harvey and David Woodward formulated a general definition for maps in the History of Cartography: “Maps are graphic representations that facilitate a spatial understanding of things, concepts, conditions, processes or events in the human world”, and the way maps “store, communicate and promote spatial understanding” also qualify them in the literary realm (Harvey and Woodward, 1987: xvi). In a sense, subjectivity is the factor that carries the true meaning of a map and distinguishes one map from another.

Figure 1.6

Figure 1.5 Richard of Haldingham or Sleaford, Hereford Cathedral map. Lincoln (England), c. 1285

Figure 1.6 A View of the World from Ninth Avenue by Saul Steinberg

Mappaemundi, or medieval maps were once simply regarded wrong representations of the world (Fig 1.5). Found portolan charts show that geography knowledge of the time was more than that was displayed in these maps (Edson, 1997: 14). Instead of geographical truth, the medieval maps were set to symbolically depict the ideal order of God’s creations and the religious history, so as to lead the reader into a “spiritual journey”. And few maps go more extreme in terms of subjectivity than Saul Steinberg’s view of the States (Fig 1.6). It vividly illustrates how distance affects a place’s relevance to the observer, and contains a subtle self-mockery at the coastal inhabitants’ pride.

The strength of such maps comes from the inclusion of a perspective. By bringing the observer’s location, time and cultural context into the maps, they create a narrative not just about what the world is, but also how the world is perceived. Such narrative has been mostly excluded as contemporary mapmaking techniques such as GIS simply suggest the same standard map layout. The value of these maps lie not only in the amount of facts they provide, but also in the snapshot of the observer’s mindset within that specific context and revealing the thinking process behind it. Hence the maps stop struggling to mirror the objective world and turn to focus on the human beings who live in it.

In this thesis, subjective maps refer to the maps that represent spatial features in a way that is dependent on the observer’s location, time, connection or understanding of the space. Subjectivity in such maps, from data selection to layout design, is pushed to an extreme intentionally to reflect the context of the perceiver. Unlike in topographical maps, distortion in subjective maps is not an unavoidable compromise, but the instrument to formulate a story.

Kohei Sugiura’s travel-time map of Japan (Fig 1.7) plotted the shortest travel times from Tokyo to various destinations around Japan using airplanes and trains . Coordinates for the various areas were determined by time using concentric circles centering on Tokyo and their topographical directions from Tokyo. The map contains a rich story about the structure of national transportation networks and the economic relevance of different regions. It is very impressive how distortion of geometries alone can express so much information. The visual impact is much more immediate and fierce than that of symbols and colors on a traditional map. This example also proves that subjective maps can base upon concrete and objective data. The subjectivity is embedded in the narrative rather than the content. Therefore, subjective maps should not be considered as just artworks, but also a serious method of data visualization.

Figure 1.7 Figure 1.7 Travel-time map of Japan by Kohei Sugiura

Subjective maps can play an important role in visualizing urban data. One of the biggest challenges in this area is information overflow: there are too many sets of data but not enough encoding methods or map space; too many data in each set with equal priority. Imagine plotting the full timetable of trains on the map of Japan: visualization of massive urban data can easily hide useful information for the viewer in a fog of details. Subjective maps naturally carry out the idea of context-aware reduction – content is selected and represented in response to the viewer’s situation and demand, content that is related to there and then. It relieves information overflow by introducing the human factor into cartography. The human factor creates a story as well as a purpose for the maps, which is essential for any map to deliver a meaning.

A map of one perspective is generally considered partial and incomplete. However, most of the maps in history were universal due to the limitation of their media and the remarkable amount of efforts needed to produce them. On internet-powered digital medium such as PC and cellphone, maps can now respond to user interaction, keep its content up to date and generate new looks on the fly. Empowered by such technologies, subjective maps become dynamic, hence are able to compete with traditional maps on completeness and usefulness.

The Politics of Subjective Mapmaking

Maps are no doubt powerful in shaping public opinions, therefore have become rhetorical devices and instruments of persuasion. It is a justifiable fear for subjective mapmaking in general: a fear for being manipulated and controlled.

It has been commonly warned that map can be used for empowerment and control. Tyner (1974) defined “persuasive cartography” with three types of maps: maps for political propaganda, journalism and advertisement. During WWII and the Cold War, maps with exaggerated design and fraudulent selection of information were used heavily to sell ideology to the public (Monmonier, 1996: 94). In 2009, Verizon released an ad campaign against AT&T’s 3G coverage, which dramatically contradicts AT&T’s own map (Fig 1.8). The two maps were based on the same data but using different data generalization methods . Many subjective choices in mapmaking were deliberately hidden behind the veil, taking advantage of the fact that the public rarely question maps of their authorities.

Figure 1.8 Figure 1.8 AT&T 3G coverage maps in Verizon’s advertisement and their own

Another concern lies in the unavoidable subjective interpretations of “one standard map”. As maps are intensively used for decision-making, politicians, designers and the general public tend to pay more attention to the areas with greater visual impact. Regions with measurement of absent or smaller value are marked with less significant colors and symbols and even left blank. Warren (2010: 23) questioned whether “mapmaking is a kind of cartographic harvesting of the most vulnerable places and people on the planet. ”

One approach to fight this cartography monopoly is promoting community participation. Mapmaking is no longer exclusive to the elites. With emerging tools and data resources such as OpenStreetMap and Google Maps Application Programmer’s Interface (API), anyone with a home computer can make a map. Meanwhile, critics worry that “cartography is being undisciplined” to the unlicensed population (Crampton and Krygier, 2006). Grassroot mapmakers produce maps that look “good” or “right”, unaware of bringing their unconscious prejudice into the maps. The maps coming from bottom-up with a voice may not be expected any more neutral than the ones for top-down.

Dynamic subjective maps in this thesis could bring a fresh view to tackle the problem. On one hand, subjective maps stand against the single-pole view of the world. There is no standard perspective in such maps, so every observer can have his own map. By adapting the map to the context of the user, information important for decision-making is emphasized instead of hidden by features that are irrelevant but dominating in volume. On the other hand, interpretation of subjective maps does not easily fall into stereotypes because of their unique spatial arrangement. They stimulate the users to question the logic of distortion, the quality of thematic overlays and the meaning of symbols.

It is true that maps with more subjectivity leads to misinterpretation more easily. That is why, it is especially important for designers of subjective maps to work in line with cartography ethics, including the careful selection of data, the neutral design of representations and the openness about the process. This will be further discussed in the following chapters.

data visualization writing