Urbagram

A short History of Intersections

Anil Bawa-Cavia — Mon, 19 Sep 2011 20:23:24 Z

Mapping Centrality

The Egyptian symbol for ‘city’, Richard Sennett points out in Conscience of the Eye (p. 46), consisted of a cross bounded by a circle, providing one of the earliest signs of the urban; intersection and enclosure — grid and boundary — come together in this hieroglyph. The notion of the urban as an intersection offers an insight into one of the longest-running themes in urbanism – centrality. Whilst Egyptians made a point of right-angled street intersections, the Roman grid plan (Hippodamus) was designed to explicitly generate centers at different scales, with the earliest phase of settlement locating itself at the intersection of two prominent axial streets, the cardo and the decumanus. To this day cores, hubs, and centers form ever-present concepts in urban discourse; all relate or allude to a set of slippery spatial phenomena we call ‘centrality’.

Images of centrality · Egyptian hieroglyph for ‘city’, 3110-2884 BC (left) and (right) Chicago sociologist Ernest Jeffers’ original concentric zone model of the city, used during his lectures in the 1920s. Photo: metroblossom.
[ source ]

The first formal representation of urban space in modern urbanist discourse was the Chicago School’s concentric model. This updated the Egyptian hieroglyph to the modern metropolis, presenting a symmetrical radial growth emanating from a single Central Business District (CBD), surrounded by successive bands of residential and retail development, an affirmation of centrality as the dominant spatial relation in all matters urban; all space in the urban realm to be defined with respect to a core and a periphery.

In Writings on Cities Lefebvre calls for “a renewed urban society, a renovated centrality”. His essays use varying notions of the concept, providing 35 instances of centrality that present it as intrinsic to the urban, an “essential quality” of its spatial manifestation. “This is the sense of urban centrality, of differences assembled through unity” he states, before moving on to his wish for a “reconstruction of centrality destroyed by a strategy of segregation”. Whilst Lefebvre fractures, diffuses and scatters the concept throughout his texts, in more recent years it has resurfaced in far more precise and quantitative terms than he may have encouraged.

Central Place Theory · Christaller’s model produces a hierarchy of increasingly spaced centers at different scales.
Walter Christaller (1933)
[ more ]

Centrality in modern spatial analysis is strongly tied to graph theoretic measures of nodal relations in topological space. Let us define spatial networks as weighted graphs that use a Cartesian distance or time-based cost function for edge weights. Confining ourselves to such spatial networks we can define closeness centrality as a measure of a node’s closeness in space/time to all the other nodes in a network, and betweenness centrality as a node’s appearance on shortest paths between other nodes, information centrality (Crucitti, 2005) as the impact of node removal on the communicational efficiency of a network, and straightness centrality (Crucitti et al, 2006) as a measure of route efficiency between any two nodes, describing the relation of a straight euclidean path to the actual path on the network. Many other centrality measures can be formulated in this manner. These multiple inter-related centrality phenomena in graphs imply a more complicated tangle of concepts now orbiting the term centrality, focusing on centrality as a function of connectedness.

Polycenters · Dendrogram showing hierarchical composition of urban centers as a function of public transport flow data from London’s Oyster database.
Roth et Al, 2010
[ source ]

Whilst these forms of centrality are based on dyad relations in a network and can be computed on a node by node basis, they exist in relation to other macroscopic concepts in spatial analysis, namely centricity and centralisation. Centricity can be considered a concept constructed out of a simplifying dualism, monocentrism-polycentrism. This is an axis, or spectrum, describing a multi-scalar phenomenon (local, urban, regional, national, global) which is both functional and morphological (Green, 2007). Centers of activity, centers of population, centers of wealth: Centers are always both relational and functional in nature. Many measures of polycentricity, such as the statistical distribution of sizes of urban settlements, which is often used at the regional scale (Meijers, 2006), focus on just one component of polycentricity, the functional, with no respect for the other, namely the morphology describing the distribution of the centers in space. Centers, to be defined as such, must of course possess a tangible quality, a centrality, which permits them to be distinguished.

Street Centralities · Centrality in a network of self-organised streets in Cairo, showing (a) closeness, (b) betweenness, (c) straightness, (d) information centralities.
Crucitti, Latora, Porta (2006)
[ source ]

Centralisation, on the other hand, is a measure of spatial concentration, again both in functional and morphological terms. Distributions of activity, population or infrastructure can be examined, their statistical dispersion often giving us insights into how centralised-decentralised a space may or may not be from the point of view of a certain urban activity. Graph theory has its own contributions here: The dispersion of graph theoretic centralities, degree distributions or node assortativity can give insight into the level of centralisation in a spatial system.

Urban modeling has consistently been drawn to centers. Christaller’s Central Place Theory provides a model that produces a hierarchy of increasingly dispersed centers, whilst Wilson’s BLV model, taking ecological dynamics from Lotka-Volterra and entropy maximisation via Boltzmann’s formulation of the microcanonical ensemble, models the slow-scale, complex spatial evolution of centers (e.g. retail). In both these, multiplier, or feedback effects, are present in processes that dynamically produce emergent centers.

Centrality, then, emerges as a strong recurring concept in urban analysis by virtue of its status as a multiple; its ability to not be drawn in to a reductive dualism. There is no one centrality, but one or more centralities, at any given point, in any given spatial system, at any given time.

See Also: Network, archipelago, Boundary, Entropy

Bifurcation

Anil Bawa-Cavia — Mon, 19 Sep 2011 20:18:08 Z

Bifurcation Diagram · Parameter space showing the disintegration of a system with one stable state, through bifurcations, into a chaotic system · Alexis Wilke & Benoit Ambry
[ source ]

The breakdown of steady state equilibrium, via bifurcations, to a multistable system, and finally to chaos. Complex urban systems exist at the edge of this chaotic state. Whilst cybernetics was only equipped to deal with the dynamics of non-linear systems with single stable states, complexity science explored the breakdown to multistability and beyond.

See Also: Cybernetics, Complexity?, The Living City, Entropy, Superlinear

Entropy

Anil Bawa-Cavia — Mon, 25 Jul 2011 11:13:16 Z

Regional Retail Model (Parameter Surface) · Navigating a grid of potential outcomes in a retail-space evolution model · Joel Dearden & Alan Wilson (2010)
[ source ]

Entropy Models (Alan Wilson, 1970-2011) take as their hypothesis that Boltzmann’s theory of entropy as microcanonical ensemble can be used to model human mobility in urban travel models. In use since 1970 as travel demand models, Wilson’s entropy models have evolved into BLV models which use Lotka-Volterra dynamics to model the evolution of urban space.

In this particular example we see a regional retail model which models both the fast-scale urban travel demand resulting from retail activity, along with the slow-scale evolution of retail space as retail centers grow and diminish. The video depicts the navigation of model outcomes in a grid that represents a parameter surface.

BLV models are part of a shift towards modeling cities as complex systems, providing a probabilistic basis (statistical physics) for determining the state of an urban model under constraints. The complexity paradigm forces planners to accept that multiple equilibria mean a range of possible outcomes can result from even the smallest planning interventions.

Hyperpublic

Anil Bawa-Cavia — Mon, 25 Jul 2011 10:44:56 Z

Hyper-Public: A Symposium on Designing Privacy and Public Space in the Connected World, June 9-10, 2011 the Berkman Center at Harvard.

Full video coverage.

Wireless

Anil Bawa-Cavia — Mon, 25 Jul 2011 10:40:56 Z

Immaterials: Light Painting Wi-Fi · Visualisation of Wi-Fi use in urban space · Timo Arnall (2011)

Immaterials: Light painting WiFi from Timo on Vimeo.

“This project explores the invisible terrain of Wi-Fi networks in urban spaces by light painting signal strength in long-exposure photographs. A four-metre long measuring rod with 80 points of light reveals cross-sections through Wi-Fi networks using a photographic technique called light-painting.”

Geodesic

Anil Bawa-Cavia — Mon, 25 Jul 2011 10:30:42 Z

TfL Journey Planner · Three-language journey planner for London, showing on a CRT monitor the shortest route between any two stations on the transport network · (TfL, 1963)
[ source ]

A geodesic, the shortest path between two points in a mathematical space. Early urban cybernetic interfaces imparted this machine intelligence to citizens via origin-destination matrices placed at public transport nodes, introducing graph theoretic concepts to the urban masses. This TfL Journey Planner (1963) consists of separate origin and destination button matrices with a CRT monitor in the center. The machine communicates the shortest path between any two selected nodes.

A cybernetic form of urban mobility was born, a citizen’s behaviour increasingly expected to approximate the ‘rational agent’ used in urban computer models of the era.

The technology is now delivered online via a web interface and XSLT Trip Request API.

See Also: Cybernetics, Shortest Path, Network

Beta

Anil Bawa-Cavia — Wed, 13 Jul 2011 11:39:44 Z

Cities as Software

Betaville · An open-source multiplayer environment for real cities · BxmC (2010)
[ source ]

Undoubtedly, the city is always in beta. The miscibility of concepts in urban planning and software development has been evident for several decades. This interchange is most pronounced in Christopher Alexander’s A Pattern Language, an unwitting boundary object of a book that resonated with the nascent object-oriented software community of the 1980s – ‘smalltalkers’ looking for new means of building modular systems.

In an age of ubicomp and real-time civic information we can only expect this mingling of paradigms to intensify. There is a feeling that the tangled history of these two fields — peaking in the systems planning heyday of the sixties, with the proliferation of cybernetic control rooms and the chimeric utopia of a perfectly ‘balanced’ social order — is set to enter a new phase in which they assume a more complex relationship.

Let us define software development as a process of structured systems design focused on minimising a range of costs – namely development, maintenance, growth and usability.

Urban planning is inextricably bound to similar costs – constructing, maintaining, using and growing the city are the chief costs to planners and citizens.

A range of strategies are used in software development to minimise these costs, from principles like orthogonality and compactness, to decoupling and cohesion, modularity and scope-creep, to methodologies like use cases and A/B testing. Importantly, any tendency to create a monolithic plan (see: Mega, Beautification) is countered by these principles.

What would an urban scale A/B test look like? I don’t think I’ve ever seen one in action. I’d like to be able to sign up as a TfL beta tester though.

Some of these software principles are borrowed from the broader field of systems design and already pervade the design of cities. Take for instance decoupling, a guiding principle in multi-modal transport networks – improving resilience, fault tolerance and reducing maintenance. From the small scale (the parts in a train) to the urban scale (the design of entire networks), this is an engineering principle which shapes our environment.

Design Patterns · Analogy in reverse: Christopher Alexander’s town planning book (left) had a direct influence on design patterns in Object-Oriented software, most notably the ‘Gang of Four’ book (right).
[ source ]

As networked citizens, we all have a legitimate claim to act as designers and developers of our urban realm. Whilst there is general agreement that modern planning tools need to scale for inclusivity, it’s not clear what best practices are required to aid consensus-driven planning to deliver. As such, there is an increasing interest in software development methodologies as urban planning goes collaborative.

Agile methodologies haven’t infiltrated urban planning, a discipline which contains projects of the requisite scope and complexity. Agile software practitioners have understood that projects of this nature demand an incremental, test-driven methodology for cost minimisation.

Iterative and collaborative design are complementary, as at each increment various stakeholders (read: residents, council members) are consulted. These methods do not map easily to the policy maze and institutional hierarchies that currently make up the apparatus of urban planning.

Betaville introduces the concept of version history via a wiki-like interface, but other methodologies worth exploring in planning are decentralised version control (ie. Git) for more complex read/write planning systems, and REST API design for public urban resources.

Cities are highly non-deterministic social systems. At most scales and resolutions, a city is orders of magnitude more complex than any software to date. Any branch of systems design is only applicable within severe limits. Attempts to ‘control’ the organic mechanisms of society seem anachronistic in the same way any attempt to control the whole of the web seems absurd: Because we have a new found appreciation for self-organised order. This shift in planning towards the stimulation of forms of activity that benefit long-term goals (densification, mixed-use zones, public transport uptake etc), rather than prescriptive practices is already widespread.

Perhaps though, software development has something substantial to offer the design of the urban realm in a network society: An appreciation of modes of networked group cooperation in the service of a complex functioning whole.

Issue tracking, infrastructural A/B tests, more accessible public changelogs, comprehensive APIs, and shorter, more inclusive iterations might be logical next steps for cities looking to lead the way towards more participative planning practices.

Essays

Anil Bawa-Cavia — Mon, 11 Jul 2011 12:58:42 Z

Microplexes

This is a transcript of a talk delivered at the Bartlett school of Architecture on March 24th 2010, as part of a series of seminars held by the Center for Advanced Spatial Analysis (CASA) at University College London. This is my first academic seminar, in which I lay out some of my research aims and their conceptual underpinnnings.

The City as Social Archipelago

An analysis of large-scale data from the location-based social network Foursquare, looking at how the urban can be sensed in an age of ubiquitous computing. This work presents an image of the city in terms of social interaction densities.

The Urban Assemblage

Sketching out a valid ontological basis for the ‘city’ as an object of research. A discussion of Manuel De Landa’s assemblage theory in an urban context.

Mapping Centrality: A Short History of Intersection

An overview of the status of ‘centrality’ as a concept in urban discourse, including visualisations, models, graph theory, and its relation to centricity and centralisation.

The Living City

A transcript of a talk delivered at Cognitive Cities, a conference on networked cities held in Berlin on 26th Feb 2011. The talk is organised into projects relating to ‘interactions’, ‘boundaries’, ‘flows’ and ‘evolution’, showcasing various information visualisations, dynamic models and simulations relating to the Living City.

Sensing The Urban

This is my first academic working paper (pdf). I conduct a spatial analysis of data from a location-based (LBS) social network, exploring the distribution of social activity in a city. I conduct a comparative analysis of three cities (London, New York, Paris), examining polycentricity, fragmentation and centralisation. These measures are used to discuss the spatial structure of the cities in question. I identify social hubs and examine the distribution of these “central places”. Finally, I discuss the wider impact of LBS technologies on urban analysis.

Allometry: On Urban Growth & Form

Allometry is the study of the relationship between size and shape, first outlined by Otto Snell in 1892 and Julian Huxley in 1932. In this short essay I discuss the work of Bettencourt & West, examining how urban allometry compares to that of the biological world.

Beta: Cities as Software

On the miscibility of concepts within the fields of software development and urban planning, including design patterns, APIs, version control and beta testing.

Allometry

Anil Bawa-Cavia — Mon, 11 Jul 2011 12:53:54 Z

On Urban Growth & Form

Biological Allometry · Allometric scaling law relating cruising speed to body mass in a range of living organisms · Adrian Bejan (2000)
[ source ]

Allometry is the study of the relationship between size and shape, first outlined by Otto Snell in 1892 and Julian Huxley in 1932. It has been a concern for mathematical biologists since D’Arcy Thompson’s investigations collected in On Growth and Form.

The allometric hypothesis suggests that there are critical ratios between geometric attributes that are fixed by the functioning of the element in question and if the element changes in size, these ratios need to remain fixed for the element to still function. 1

Bettencourt & West’s work shows clear macroscopic correlations for a number of structural, social and economic characteristics of cities (GDP, gasoline stations, electrical consumption), each scaling with population size N as a power law of the form y=N^β. These provide a mathematical basis for Aristotle’s metaphor of the city as a living organism, as biologists have also shown this to be true for metabolic functions in living beings.

A number of infrastructural features – such as road surface area and gas stations – scale with an exponent β< 1, indicating economies of scale, whilst the majority of wealth indicators – such as average wage, or number of patents – scale with β> 1, indicating increasing returns at larger sizes. There is evidence here that larger cities are more structurally efficient than smaller ones – a trend also echoed in organisms. These findings are complementary to observations of geometric scaling and fractal structure in cities (Batty & Longley, 1974)

Claims of a unified theory or a solution to the city are misleading, as many of these relationships are in effect universal allometric scaling laws evident in urban settlements. These are no more ‘solutions’ to the range of phenomena exhibited by cities than the allometric relation of metabolism to body mass – Kleiber’s Law – provides an understanding of the inner workings of a living organism.

Correlation in data-driven research such as this clearly does not in itself provide a theory. Association of these characteristics is weakly argued when a “universal social dynamic” is posited, “inextricably linking” the correlated phenomena in a “dynamical network” 2. Replacing simplified notions of supply and demand (or of causation) with a complex network model seems both timely and useful, but mere correlations of variables with size don’t prove this to be a more faithful conceptual model for a city than any other.

Volumetric Uniformity · Building heights in Andy Burnham’s master plan of Chicago · Jules Guerin (1909)
[ source ]

In referring to the relation between the size and shape of things, allometry must in an urban context be considered with respect to the energy capacity of the human body – or the enhanced body, if we consider various prostheses in the form of vehicles.

A famous case study in the allometry of buildings is Baron Haussmann’s Paris. Building heights were set during the re-development to a maximum eaves height of 17.5m, or 5-6 storeys, creating a homogenous landscape which is still visible today. This was considered at the time a comfortable distance a human would climb unaided multiple times a day. Berlin settled on 22 meters for its Mietskaserne (tenements) in 1853, with this figure shaped by the concerns of the Fire Department of the time.

Haussmann Reneged · Discontinuities in building heights along the modern Parisian skyline. · David Forster (2010)
[ source ]

In modern urban development, the advent of the automobile, rail and subways has produced a large sprawl of low-level (2-3 storey) housing stock, dramatically enlarging the surface area of the urban landscape while lowering its density; a trend seen in most Western cities. This is in sharp contrast to the multi-storey apartments that pre-date the industrial revolution and still inhabit the compact cores of most European settlements.

Meanwhile, the introduction of megastructures and skyscrapers, enabled by a range of technologies (most notably the triad of elevators, escalators and air-conditioning), has produced new allometric relations in buildings. The pace of technological innovation means that the processes that might describe an urban ‘metabolism’ must be considered to be evolving from decade to decade. For example, since the introduction of skyscrapers, rank-size building height distributions also produce clear scaling laws. This phenomenon, as well as geometric scaling in cities are discussed further in (Batty et al, 2007).

Beijing Megastructure · OMA’s CCTV in Beijing · Alebi (2010)
[ source ]

These shifts in construction and transportation produce new surface to volume ratios in our growing cities, which in turn have repercussions for the energy consumption of urban regions. If we take as a simple example an ideal sphere, heat dissipation is proportional to its surface to volume ratio, which scales inversely with size, resulting in more efficient retention of heat at larger scales. The same kind of relations, when aggregated from the building to urban scale, suggest that discontinuities in building size and density significantly alter the energy footprint of our cities.

At the urban scale, transportation, along with the spatial distribution of land-use, will continue to be key factors governing allometric relationships in cities. A continuing research challenge exists in taking historical data for particular urban cases to show how a range of variables from land-use ratios to the range of infrastructural features explored by Bettencourt & West change during the long-term growth that typifies the evolution of an urban settlement.

Registry

Anil Bawa-Cavia — Thu, 07 Jul 2011 11:55:03 Z

Creating a global registry of built works:

http://www.aaronland.info/weblog/2011/05/17/things/#bwr
http://buildingequalsyes.spum.org/

Superlinear

Anil Bawa-Cavia — Thu, 07 Jul 2011 11:52:09 Z

Geoffrey West’s talk for Edge, Why Cities Keep Growing, Corporations and People Always Die, and Life Gets Faster, provides a nice overview on universal scaling in urban systems, part of his attempts to produce a “quantitative predictive, mathematizible kind of science” for cities.

The talk outlines West’s interests in parallels between biological, urban and business growth. His much reported claims regard the scaling (as a power law) of diverse urban metrics with population size. Super-linear scaling of productivity markers for cities suggest increasing returns, whilst sub-linear scaling of infrastructural costs associated with cities suggest economies of scale, all in all providing an optimistic view of cities’ role in human civilisation.

Plug-in

Anil Bawa-Cavia — Mon, 11 Apr 2011 12:29:55 Z

Plug-in City
Archigram (1961-64) · Computer-controlled city with removable elements plugged into a ‘megastructure’ service framework.
[ source ]

Habitat Expo ’67
Moshe Safdie (1967) · Generative modular housing complex designed to maximise light to habitation units.
[ more ]

Cubic Houses
Piet Blom (1984) · Modular tilted cubic housing complex, Rotterdam.
[ more ]

See Also: Mega, Emergence, Yona Friedman

Assemblage

Anil Bawa-Cavia — Mon, 11 Apr 2011 12:20:45 Z

Assemblage theory provides us with a bottom-up ontological framework for analysing social complexity 1. The social assemblage is distinct from a Hegelian ‘organic totality’ in that the relations of its component parts are not logically fixed – as in the traditional view of organs in a body. By contrast, assemblage theory asserts that the same component can play differing roles in diverse assemblages. To consider biological organisms as assemblages then is to contend that relations between organs are only contingently obligatory as a result of co-evolution.

Assemblage Theory & Social Complexity · Manuel De Landa (2006)
[ more ]

Each assemblage is an emergent entity which can combine with others to produce ever larger assemblages; both assemblages and their component parts are thus characterised by reciprocal relations of exteriority 5. Small market towns may synthesise into regional or national markets, individuals may form social networks which in turn coalesce into organisations. But each individual or town may be involved in a wide range of assemblages, taking up differing roles in each.

The components of an assemblage can play roles which lie on a spectrum from material to expressive . All physical materials can be expressive: Take for example the characteristic spectroscopic fingerprints of different materials, as used by astronomers to identify the makeup of the universe. These expressions, much like a human fingerprint, are not functional in themselves. In an urban context one such example of expressivity without function is a characteristic skyline formed by buildings. Deleuze regards DNA as a special instance in history in which physical expressivity turned functional. Genes and words – given their functional capacities – are in his view specialised expressive entities 4. De Landa maintains however that language does not play a constituent role in social assemblages, but that it is one amongst several primary expressive materials 1.

To extend assemblage theory to the urban — the urban assemblage — is to regard the urban as the site of certain densities and concentrations of material processes which are necessarily spatial. We can conceive of these as intensities of flows along networks operating at different scales. The components of these assemblages are not only bodies, but labour, resources (food), money, buildings, infrastructure, language and a range of other materials, including those constitutive of climates.

One can choose the individual as the smallest component in this ontology, but here a crucial distinction must be made with traditional sociology. The individual mind is considered a physical assemblage. As a material component like any other, it is not the core element in social assemblages. This allows us to free assemblages from their representation as a collection of subjects, moving us towards an object-oriented ontological framework 6. In this framework subjectivity itself emerges as an assemblage, it has no privileged position. This puts assemblage theory’s treatment of the individual in close proximity to Latour’s account of actors as quasi-objects in actor-network theory, and its treatment of reality in proximity to that of speculative realism.

This ontology is opposed to any form of taxonomic essentialism, which entails a supposition that categories can refer to things in the real world, and in doing so moves us towards the notion of a mind-independent reality. As De Landa puts it,

“The ontological status of any assemblage, inorganic, organic or social, is that of a unique, singular, historically contingent, individual… Much as biological species are not general categories of which animal and plant organisms are members but larger-scale individual entities of which organisms are component parts, so social assemblages should be given the ontological status of individual entities: individual networks and coalitions; individual organisations and governments; individual cities and nation states. This ontological manoeuvre allows to assert that these individual entities have an objective existence independently of our minds.” 2

A Thousand Years of Non-Linear History · Manuel De Landa (2002)
[ more ]

This approach can be seen as a form of materialism in which matter 7 and energy take precedence over discourse and ideology. Human culture can thus be interpreted and analysed as a history of material flows – linguistic, economic, genomic – forming a kind of geological strata over time, observable as material histories subject to morphogenetic processes 3. From this perspective the social has a material being all of its own. Just as De Landa, very much a realist, admits ‘objects’ their own being in-themselves, so too he asserts that cities possess a city-being, a life of their own quite apart from the human mind.

The dynamics and structural properties of social assemblages – their growth, topologies, characteristic intensities, relations of exteriority, and interactions, can thus become the object of social science. Assemblages, unlike totalities, are analysable because they can be reconstructed from their constituent parts. In order to do this one must understand the relations that characterise the assemblage in terms of its emergence as a result of the dynamic interactions of its components. This does not entail the reduction of a totality, as to explain “the mechanisms of emergence does not explain emergence away…” (Mario Bunge).

This stance is not without its problems. If we talk about congestion, commuting or segregation, as characteristic phenomena of city-beings, then we may go about attempting to show these phenomena using agent-based models. Very quickly, however, one needs to delve into human psychology to refine these models. These phenomena aren’t produced by perfectly rational automatons. Modeling agents with an increasingly sophisticated capacity for spatial cognition, for example, becomes important. Our analysis is back to hypothesising the internal processes of the human mind in an urban context, far removed from the materialist spirit of assemblages.

In terms of methodology, network theory provides us with a representational toolkit for examining urban assemblages, whilst complexity science gives us a mathematical framework for dealing with emergence. Computer models can help us to understand how synthetic processes involving differing components can produce assemblages with differing characteristic identities. These tools allow us to pursue a social science methodology based on assemblage theory.

See Also: Emergence, Network, Morphogenesis

Lost in Space

Anil Bawa-Cavia — Fri, 01 Apr 2011 10:30:57 Z

Chromaroma · The gamification of urban space: Chromaroma is a team game based on data generated by smartcards on the London transport network · Chromaroma (2010)
[ source ]

Lost in Space, a spatial research conference organised by University College London, bringing together architects, designers, neuroscientists and computer scientists to talk all things spatial.

My personal highlights were design researcher Regina Peldszus’ presentation on the design of space stations, and neuroscientist Freyja Olafsdottir on how we might be constructing future trajectories during sleep, implying our mobility is aided by hippocampal brain activity during periods of rest. Fascinating stuff bordering on sci-fi.

Slides from my own talk are now available for download as a pdf:

Placerank: Remarks on Pervasive Cartography (pdf, 7.9MB)

You can find more details over at Lost in Space.

The Living City

Anil Bawa-Cavia — Tue, 29 Mar 2011 21:59:39 Z

This is a transcript of a talk delivered at Cognitive Cities, a conference on the future of cities held in Berlin on 26th Feb 2011.

Hello. I’m going to show you some research projects by myself and other colleagues at CASA, which concern what I call the living city.

What do I mean by ‘the living city’?

Invisible Cities · Bending Bookshelf (2010)
[ source ]

Cities are made up of physical networks of infrastructure, from buildings to roads and subway lines. But they are also made up of ﬂows of people, vehicles, information and goods. It’s these cities of ﬂow and networks of interaction which I call the living city. The living city is hard to spot, as it doesn’t tend to manifest itself in a tangible way. We move around cities. We interact with each other. But these actions leave little physical residue. The living city defines a multiplicity of invisible cities.

Desire Path · @iirraa (2010)
[ source ]

The living city is unplanned. In fact, we can define it precisely as those elements of urbanity that are unplanned. The so-called desire path here is the living city in conﬂict with the planned pavement. It’s a rare case of the living city leaving an obvious trace. For the most part, we need to pick up trails of digital data in order to understand the spaces produced by the dynamic interactions of many individuals with each other and with the city infrastructure. As a researcher, one of my tasks is to make this living city visible through data.

I’m going to talk about the living city today in terms of interactions, boundaries, flows and evolution.

Interactions

In this ﬁrst project I’ve looked at the social life of cities. The city of social interactions.

Social New York · Depiction of New York City in terms of social interaction density · Anil Bawa-Cavia (2010)
[ large ]

What I’ve done is produce bottom-up images of cities by mapping millions of events on Foursquare, a location based social network. This is an image of New York City constructed from 600,000 GPS events on the Foursquare network. Each event is a user checking-in to a social venue (a bar, restaurant, art gallery, nightclub, etc) on their mobile phone, in order to let their friends know where they are. Each dot represents a walkable cell of 400×400 meters and its size is proportional to Foursquare activity over a year. By this process, recognisable social hubs emerge. You can see Midtown, Central Park, Lower East Side, Williamsburg on here.

Social Paris · Depiction of Paris in terms of social interaction density · Anil Bawa-Cavia (2010)
[ large ]

Here’s Paris. Again, you can see areas like Le Marais and Jardin du Luxembourg pop out. In these activity ﬁngerprints, as I call them, we see a new image of the city emerge in terms of the density of social interactions. As smartphones begin to penetrate to the level of mobile phones, namely 90% in the UK, these could be considered quite comprehensive snapshots of activity in cities. This is just an aggregate snapshot: We’re only just beginning to look at the spatio-temporal aspect of this data, the dynamics of this kind of social activity.

Social London · Depiction of London in terms of social interaction density · Anil Bawa-Cavia (2010)
[ large ]

This ﬁngerprint of London is a little more dispersed than the others, with areas like Shoreditch and Regents Park, Soho and London Fields, all showing up as centers of activity.

The ﬁngerprints show how the city is actually used, not how it was planned or built. This kind of activity data builds on land use data from the land registry, which is all urban researchers have had to go on in the past.

Urban Activity Fingerprints · Visualisation of three western metropolises in terms of social interaction density · Anil Bawa-Cavia (2010)
[ large ]

By using embodied measures, in this case the 400m cell, which is a comfortable 7 minute walk for most people, we can compare these social activity ﬁngerprints across cities. I focus on the human body because it’s a constant across cities and comparative analysis needs such constants. The capacity of our bodies to move through space is fairly constant across cities, whereas most other aspects of cities are variable across different urbanisations.

I’ve produced a range of spatial analysis of this activity, exploring issues of interest to urbanism, such as polycentricity, centralisation and agglomeration. If we analyse spatial dispersion for example, London’s ﬁngerprint shows a greater evidence of polycentricity, and Paris’ is the most compact and walkable. If we look at fragmentation, as I do in this image.

Fragmentation Studies
Relative fragmentation of social activity on the Foursquare social network across New York, London & Paris · Anil Bawa-Cavia (2010)
[ large ]

London and New York show far higher degrees of splintering in their social activity than Paris, which breaks down to far fewer, larger clusters of activity. This analysis is produced using DBScan, a spatial clustering algorithm. Each polygon is bounded by a cluster of venues, and its brightness is proportional to activity at those venues. Again the same embodied measure is used to deﬁne threshold distance in this fragmentation analysis, because the city feels fragmented to the individual if one can’t walk between locations – our image of the city is splintered by unwalkable distances.

Social Agglomerations · Depiction of New York City and Paris as agglomerations of social interaction, using data from the Foursquare network · Anil Bawa-Cavia (2010)
[ large ]

Here we see each social venue in a city as a translucent dot, so every restaurant, bar, nightclub and art gallery in the dataset is represented here, with dot diameter proportional to activity at that location. The brightness shows us activity intensity. You can see a heavy concentration of activity in New York around Manhattan, and the decentralised proﬁle of Paris, in which activity tends to settle along axial and radial lines, often corresponding to avenues and boulevards in the city.

We can use this research to start quantifying qualitative terms such as ‘sprawl’ or ‘segregation’, in terms of human interactions. New York is characterised by vast areas with very low background activity, and we can begin to look into correlations between activity levels and land-use for example. A lot of these low activity areas are known as relatively single-use residential which are typically called ‘sprawl’.

On the Parisian ﬁngerprint we see aspects of segregation, as activity falls away beyond the péripheriqué ring road, meaning these smartphone users are socialising in under a quarter of the unité urbaine that deﬁnes Paris.

(You can see more of this project on the Archipelago page).

Boundaries

As social beings, we are predisposed to group ourselves into communities and even cliques. We can detect these invisible boundaries in the living city using communications data.

Redrawing the map of Great Britain · Community detection in a network of 6 billion phone calls · MIT Senseable City Lab (2010)
[ source ]

This map of the UK is composed of regions produced solely using phone call data. Billions of phone calls are analysed in order to draw these new boundaries based on communication patterns. In some cases they correspond to topographic or political boundaries but in some cases they don’t. The analysis of this network of interactions was a collaboration between my colleague Jon Reades and the MIT Senseable City Lab.

Here’s a short video from the research group.

Redrawing the map of Great Britain · Community detection in a network of 6 billion phone calls · MIT Senseable City Lab (2010)
[ source ]

The team used community detection methods from network science to partition all the callers into communities based on the phone call network. This analysis can also be produced at the urban scale. [ more here ]

Flows

Increasingly we leave traces of movement, from GPS enabled devices and RFID transport cards, or ‘wired’ infrastructure.

London Bus Flowprint · Visualisation of flows on the London bus network · Anil Bawa-Cavia (2010)
[ source ]

I’ve been producing what I call ‘ﬂowprints’ of cities. These visualise the city of ﬂows. This is a ﬂowprint of the London Bus network, which contains about 30,000 bus stops. Each dot is a bus adhering to one of 700 routes on the network. I’ve used simulated ﬂows in this ﬂowprint, incrementally adding more and more buses evenly to each route, in order to show certain properties of the network structure.

I keep adding buses till we reach about 8000, which is the actual size of the London bus ﬂeet. The brighter areas are where routes overlap or converge. These tend to be bus depots around the outskirts of the city, or the core of the network, which you can see in the middle. This is the historical center of the network, around Victoria, and a great example of what complexity scientists call ‘path dependence’: the capacity for small decisions or events to take a complex system in a completely different direction, constraining and dictating its future growth, disproportionately influencing its future. Because of location choices made 150 years ago in 1860, the network evolved around Victoria Station as its center, and to this day it contains the highest convergence of routes on the network.

These ﬂowprints are like ‘macroscopes’ for cities. They allow me to look at the dynamics of the whole city functioning at once whilst preserving a high level of detail.

Oyster Flowprints · Depiction of flows on the London Underground network during a typical weekday (top) and weekend (bottom) · Anil Bawa-Cavia (2011)
[ source ]

We can use Oyster data to produce ﬂowprints based on real ﬂows of passengers on the London Transport network. Here we see the London Underground network. Along the top you see a typical weekday and along the bottom a weekend. The graph shows activity on the network, and you can see how weekday ﬂows are characterised by the double-humped dynamic produced by commuting. The ﬁrst hump, which peaks at about 8:40AM is far steeper than the second one, which peaks just after 6pm. This implies that Londoners all go to work at about the same time, but come home at a range of times between 17:30 and 19:00PM, presumably based on whether they go out for a pint after work, which Londoners are renowned for doing.

Along the bottom you see a Saturday which shows much lower use, but rises slowly till a peak around 6pm.

By visualising the ﬂowprints throughout the day we can see them expand during rush hours, long tendrils reaching far up into London’s suburbs (or metroland), and then contract during the day, in which only really the central portion of the network is in use.

Oyster Flowprint · Visualisation of trips using London’s RFID Oyster Card on the London Underground network · Anil Bawa-Cavia (2010)
[ source ]

Here you can see the london underground ﬂowprint in action. Each trail is produced by an individual passenger touching in and out using their Oyster card. We construct a route for each journey based on a simple shortest path algorithm. This is necessary because we only have origin and destination data from the system.

Here we see the expansion of the ﬂowprint as we reach morning rush hour, with these long tendrils heading northwards, the metropolitan, victoria, northern and central lines. London functions as a network of suburban villages and this underground network was designed to bring people into central london over a long distance.

We see the pulse of the city in this expansion/contraction movement. These diurnal patterns are the strongest signatures of the living city, in that they apply to most large cities in the world. We are analysing large-scale Oyster data sets (200M+ trips) at the moment, to help the transport authority understand how large groups of people react to disruption events. There are on average 200 simultaneous disruptions to the London transport network, so it’s always in an imperfect state.

Real-Time Cycle Hire Map · Interactive web-based London Cycle Hire usage map · Ollie O’Brien (2010)
[ source ]

We’ve been working with ﬂow data from the London Cycle Hire scheme, known to Londoners as the Boris Bikes. This is a screenshot of a real-time bike hire usage map available online. We have these interactive maps for about thirty cities with hire schemes. You can select any bike station to see its recent usage history and whether there’s any bikes available right now. In this visual, the red nodes are bike stands full of bikes, and the blue nodes are empty, so this is taken from the end of the day, after commuters have cycled back out.

This map was produced by Ollie O’Brien at CASA, using openlayers and openstreetmap, two open web mapping technologies.

Boris Bikes · Flow network for London’s Cycle Hire scheme, known as the ‘Boris Bikes’ · Anil Bawa-Cavia (2010)
[ large ]

If we examine the underlying ﬂow network for the cycle scheme, we can see where the ﬂows are concentrated. The edge weights of this graph relate to the volume of ﬂow between two bike stands.

The two strongest nodes, which also act as hubs with high numbers of connections, are King’s Cross in the North and The South Bank in the center. These aren’t big destinations in themselves, they are transport hubs, interchanges. This implies that people are using the bikes as part of a multi-modal trip involving either rail or tube, which is encouraging as this could help to ‘distribute’ the rush hour peak we saw in the earlier Oyster data amongst different modes.

Activity is heavily concentrated in the East of London, and this is reﬂected in the bike stand provisioning, which slightly favours this part of the City. West London is generally speaking more affluent, with more car owners and better subway provisioning. The cycle network is designed as a kind of mesh, a point-to-point topology that doesn’t privilege any one node.

The quadrangle cluster in the West is Hyde Park, and these are the most popular bike stands on weekends.

We can use the usage data to understand precisely the kind of trips people take on bikes. It turns out the modal trip frequency occurs at 800m or about a 10 minute bike ride. It seems distances lower than 600m or so are largely deemed walkable, which backs up ﬁnding from other research, which takes 400m to be a comfortable walking distance in an urban context.

Evolution

The living city is increasingly a phenomenon we can study in near real-time. At CASA we also model the slow scale evolution of urban form. This occurs over much longer time scales; months, years or even decades.

Slow-scale retail space model · Results grid for a model exploring the slow-scale evolution of urban retail spaces · Joel Dearden & Alan Wilson (2010)
[ source ]

In this model by Alan Wilson and Joel Dearden, the slow scale evolution of retail space in cities is simulated. I won’t go into the technical details of the model, but the grid you see in the upper left is a set of outputs of the model. In each cell is a different equilibrium outcome showing the distribution of retail areas, and their relative size, for an urban region. This is a land-use transportation model and this surface of cells represents what physicists call a ‘phase space’ of possible outcomes.

Data from the near real-time living city can be imported directly into these models as empirical functional relationships. For example, how far people are prepared to travel on different modes of transport. We can observe these functions directly from the real-time data and bring them into this much slower scale model.

Slow-scale retail space model · Results grid for a model exploring the slow-scale evolution of urban retail spaces · Joel Dearden & Alan Wilson (2010)
[ source ]

You can see here the height of each bar is the size of each retail space, and each cell is a model run. These are high streets or shopping centers – any cluster of shops. They are all competing for the custom of millions of residents. Some emerge over time and others die out. The complexity of the model means there are many equilibria and small parameter changes or planning interventions can tip the system to a different one.

Evolution of Urban Form (Parameter Surface) · Navigating a grid of potential outcomes in a retail-space evolution model · Joel Dearden & Alan Wilson (2010)
[ source ]

The tool allows planners to test planning interventions, and to see how stimulating retail ﬂoor space in one location might affect other locations.

Cities are complex systems full of interconnectedness, and outcomes to even small scale interventions are unpredictable. This is why the model produces a range of possible outcomes on this kind of surface. It makes planners sensitive to the kind of complexity they are dealing with, while at the same time providing a toolkit for simulating decision making.

This is still work in progress, but shows how data from the living city might inform slower scale models. The more we know about how people actually use cities, the more we can reduce this parameter surface considerably, which will aid planners.

Bifurcation Diagram · Parameter space graph showing the disintegration of a system with one stable state, through bifurcations, into a chaotic system · Alexis Wilke & Benoit Ambry
[ source ]

So to wrap up. This is a bifurcation diagram showing the disintegration of a system from one stable equilibrium, into 2, 4, then 8 equilibria and further forking into what we have on the right extreme of the graph, which we call chaos. Cities exist at the edge of this chaos.

The living city is an entity unto itself, not just simply an aggregation of people. It’s complex and highly unpredictable. But it can be made tangible using visualisation, and it is knowable through the kind of analytical research I’ve shown you today. It’s knowable in so far as the emergent processes that typify its behaviour across different time-scales can be understood, which they can. Processes such as congestion, on a small time scale, and gentrification and segregation, on a longer time scale. Processes like fragmentation, and agglomeration, which I am trying to engage with in my research.

The promise of large-scale real-time urban data is that the infrastructural city can become more responsive to the living city and vice versa. Perhaps the planned and the unplanned city can come closer together not only through better informed citizens, but also through a better informed, and more reflexive infrastructure.

As a result our cities could become more usable spaces, and I think citizens, designers, engineers, planners and researchers all have their contribution to make in achieving this.

Thank you.