Swarm What?

Swarm What?
Benjamin Bratton, 2012

What it is that designers actually mean when we/they talk about swarms, intelligence, and swarm intelligence? Usually we refer to rhythmic emergent processes –waves– but do this referring by sharing demonstrative images of swarm topologies, frozen or in motion. As information visualization becomes an increasingly global language of systems, events and trends, these images of swarm aggregation take on a more public and political assignment. They are not only statistical abstractions of human and non-human societies parametrically assembling through time, they are also self-representations that these societies that take on, as collective representations always have, the agency of a model, of a structural topology against which individual actors measure their progress. And in this, the image is the interface of a systemic recursion between trace and traced, between monad and society. It has agency.

In the course of his revitalization of the work of Gabriel Tarde, Bruno Latour strongly champions a parametricist ontology that shuns the determinant role of Durkheimian social structure. He quotes Tarde, But, no matter how intimate, how harmonious a social group is, never do we see emerging ex abrupto, in the midst of its astonished associates, a collective self, which would be real and not only metaphoric, a sort of marvelous result, of which the associates would be the mere conditions.”

But today, in our wake of data oceans, is this “collective self” another name for the independent topology of the swarm, the assembled, rhythmic aggregation of a multitude of monadic events which takes on an emergent corporeality that cannot be reduced to those events? Tarde continues, “The impersonal, collective character is thus the product rather than the producer of the infinitely numerous individual characters; it is their composite photograph, and must not be taken for their mask.”Today we have supercomputers to employ in the rendering of such “photographs” which can simulate whatever we take swarm intelligence to be at much more granular dimensional and temporal resolutions. In these images, taken as both indexical images of the territories they represent and as icons of new topologies and typologies in their own right, we recognize their agency as recursive interfaces of and for the collectives they signify. As anyone who has worked with data visualization is well aware, any set of good or bad data can be architected in any number of ways. The structure of one revealed diagram presents one set of interrelationships between one set of interrelating variable events. If we are to try to truly agree with Tarde and Latour’s argument that aggregations at scales larger than the individual are the result of the interrelation of intelligent monads, then we would have to recognize that there are multiple aggregations at work at once: just as the aggregate (and the image of the aggregate) cannot (only) be reduced to its components, the individual participates in as many swarms and networks as you might care to identify, quantify, analyze, and render visually (and therefore cannot itself be contained by any one image). But we don”t render all the images. We render some. Because we render some and not others the politics of vision, of the distribution of the visible, aren”t overcome so much as they are synthesized and instrumentalized. Tarde is often quoted as imagining a public that is as fluent in statistics as natural language, and today”s populist information visualization suggests some spectacularized version of this. But even so, indeed because so, the recursive capacity of the image to influence assemblages at a smaller spatial scale is that much more salient, and the “Durkheimian” question about how images operate as collective representations remains.

Why? Because swarms are topological and typological, and in the cumulative meta-aggregation of topologies into a typologies of topologies, they become both political images and political envelopes. They become interfaces.

ii.
How do we see this operating at the level of the city? How does the image of the swarm contribute to the assembly and self-piloting of the urban swarm? As the sprawl of social networks in situ comes to involve computation in its formation, each of our individual participations in these networks is mediated through our ability to visualize and locate ourselves in their midst. For in the landscapes of ubiquitous computation, aren”t we the most interesting of the smart objects? In that groups occupy spaces and do so in patterns that are repetitive and predictable, then even before computation arrives on the scene social networks are situated and spatialized. Elemental networks are sets of shared pathways through cities, spaces and routes and destinations that they it offer. We may have infinite options of for possible routes, but over the course of given time, our paths are actually quite regular, and through this our networks are composed of those who share some part of our route patterns. Can these overlaps become more than invisible trances? How might they become reflexive media? Citysense, a smart phone application, is one means to visualize this. It records your (phone”s) movements throughout the city and stores this information for a brief period of time, and by doing this for many thousands of people simultaneously and cumulatively, it is able to generate cluster maps of who is where in the city, where they came from, and where they are going back to once done. Citysense learns from these territorial patterns of movement, and identifies distinct classes of behaviors in specific contexts, the company calls “tribes.” Currently the app shows you unusually high activity in one part of the city, where other people you may or may not know but who share your habitual paths are headed after dinner. without revealing anyone”s individual identity to the network or to each other. The platform is effective though rudimentary, but what is most provocative is that while social networks have always structured the microtopology of society, and its formal study is at least as old as Sociology itself, and that while such networks were constituted by shared fields of interest, access to spaces, etc, they were for the most part entered into but remained invisible to those who constructed them.

The “networkee” (the traced) did not perceive the network in which he was moving while he was moving in it. While certainly in some cases, (such as French court society) the architecture of social networks was formally diagrammed and indexed, what is new is that with social graph applications such as Facebook (and whatever Citysense-type apps evolves into) the social network is now also a reflexive medium through which people access, structure and design their social worlds. You manipulate the image of the network so as to participate in the network, and in this networks are not just the index of social interaction, they are the media of interaction, and a form and content of social display and indentification. And in that networks are intrinsically spatial, this reflexivity of the network image makes the the sociology of the city equally recursive in ways it was not before, and binds the redesign of the city to the redesign of that self-reflexivity, those media, those technologies of display, and to how the software that enables it all drives and is driven by the visceral experience of location. We can then see then how the recursive image of networked location can form the basis of emergent imagined communities around which a geo-politics of identification, solidarity and boundary-formation will emerge. It”s not just that the future of governance is networked, its that the future of networks is government.

iii.
How can we understand then the strata of aggregation from a single event to a swarm with independent topology to a city and back again? We need to demonstrate an interfacial scale model which diagrams the accumulation of interfaces toward networks and territories. Let”s continue the general use case sketched above. The mobile handheld frames visual-spatial interfaces on its tactile surface which arrange specific semiotic cues and clues about a given environment, from subtitling layers augmenting the perception of objects to a simple list of all those persons who could be called at any given moment. As the user establishes place she encounters both tangible urban surfaces and their designed interfacial capacities which the mobile arranges as a personal lens on possible engagements with a site. In the interface the chain of signification between screen-icon and world-object is linked so that her manipulation of the iconic image of the object actually causes some transformation in the object (to click on the picture of the coke bottle on your phone causes the machine in front of you to dispense one bottle and charge your account). That chain of signification works both ways. As the worldly object is represented by the icon and the icon represents in advance a worldly event it can command upon activation, the flow of representation and counter-representation, and of instrumentalization and counter-instrumentalization between image and object is looped. That loop creates a peculiar status for the information visualization that is the interface on the mobile. It is a limited and specific map of the world and visualization of what the device can do, but it is a map that not only draws a location but can act upon it and indeed as an interface allows for the location to be overwritten by the user.

The accumulation of these interfacial events at the level of the user”s encounter with urban interfaciality, events upon events and connections upon connections, generates another scale of interconnectivity with its own patterns, densities, rules, temporalities and scale: the network. As the Citysense application exemplifies, the stigmergic currents of clustered groups of people moving though a city wears virtual grooves into its surfaces, and produces nodal sites by the cumulative overlapping and gathering of multiple itineraries around common interfaces. That accumulation requires a sufficiently heterogeneous field of possible interfacial sequences, singularities, centralizations and decentralizations in order for a specific, meaningful network condition to emerge out of other possible patterns. Like patterns on a game board, it is not enough that there is room for many interfacial links but that there is capacity for a multitude of patterns to arise in the addition and complexification of groups upon groups of links. That there would be one pattern and not another is what makes the social reality of one network, its inclusiveness, exclusiveness, duration, centrality, to have significance, especially to itself. Like the interface which is both an image of the world and an instrument within the world, applications like Citysense demonstrate that the network is not only the emergent form of the cybernetic back-and-forth of many interfacial events, it is also a condition that is capable of describing itself as an image, one that is can come to govern and design how it is itself informed. Through this recursion, the image of the network can also then govern down an order of scale and structure specific interfaces so that their use would result in desired outcomes at the level of the network. There is, as for second-order cybernetics, a reflexive cycle of information that can govern the cycle to which it refers and is referred by.

Networks operate in varying degrees of isolation and proximity to other networks of similar or dissimilar scale, intensity and duration. One set of stigmergic flows through a city creates a network but does so within a common territory with other networks, such as a city or a country or a continent, which is both the historical result of layers of network activity over time and the immediate common infrastructure upon which a network can form at a given moment. While there is no intrinsic difference in temporality between interfaces and networks, in that fixed slow interfaces can correspond with fast, mobile networks and vice versa, the function of territory, as infrastructure and landscape, is more likely a slowly transformed background horizon upon which interfacial networks convene. Political jurisdictions operate at the level of geography (which I will address below) but they make claims over territorial expanses of lands, settlements, and bound populations with conceptually regular borders and membranes. Just as the interface and the network are both real physical connections and also layers of connection which produce a recursive governability in their self-imaging, the territory is both these real flora and fauna about which territorial claims are made and mapped and the descriptive layer about which those claims visual their spatial inscriptions (and spatialize their visualizations). Further, just as the pathway of accumulation works from interfaces into networks into territories, the inverse chain of designation is equally pertinent. The design of specific territories, as images and as real places has long been a normal means to govern the legal identity of networks and their interfacial interrelations.

Lastly (for this model of aggregation) the geoscape refers to a accumulated territory of territories; not an empty grid into which territories can be arranged such as a Mercator map, but a space (at once discursive and military) which the inscriptions of geo-graphy fills always just to capacity by the production of their ongoing, irresolvable claims. The geoscape then is the scale existing both at the limit of the terrestrial globe and beyond the physical limit of any one territory that would claim that entire planetary space for its jurisdictional authority. It operates then just beyond the scale of cosmography, because while it works as a meta-territory, as the geography that a cosmology claims, it is always capable of accommodating and locating as many utterly orthogonal jurisdictional claims over its own same or other space as can be possibly produced. This includes any that would confuse its own philosophical aspirations to universality with the unitotality of the planetary sphere as interchangeable territories and geographies. This is precisely what makes the geoscape a particularly useful concept for the design of a cosmopolitical program that is not bound to that founding confusion, especially as planetary computation comes to take on the status as predominant territorial formation and medium.

iv.
In practice the interrelationship between scales, both as real accumulations and as diagrammatic images of those accumulations, is a densely intermingled pattern of infra- and intra-signification. The interfacial image represent an interfacial assemblage, but it can also represent a network, a territory or a geoscape. For example a hand-held mapping application may construct a framed image of very local interfacial objects and surfaces, or the larger scale networks at work and the territories that structure them. The Citysense app as an interface to the urban field draws an image of particular networks among others as they congeal onto particular nodal interfaces within the city among others, but the image is of the coagulation more than of a given interface within it. It also locates that network onto the schematic territory of the city here rendered only as a field of channels for the potential formation of networks. In this the interface not only accumulates into networks and networks into territories, here the interface directly diagrams the network in formation and the territory as its limit. We can expect any recursivity then to be between a decision to be subjectivity located within a field of urban interfaciality based on a sense of coordination with a network (“my people go there so I should too”) and even the positive effects an action might have on the network accumulation itself and its socialized image (“if I go there it will show up on the map of my network”).

We can imagine any number of other examples by which the direct accumulation or image of accumulation at given scale comes to structure (diagram-in-advance) the instrumental appearance of another scale: where, for example, a theological geography appears as an interface through which a ritual map is read and re-read, where a territorial city zoning plan directly structures the possible networks that can operate within certain locations and in turn delimiting which interfaces can appear there, where the flight schedule board at an airport draws organizes each gate as a node within a synchronized global territory of time, of which it is one instance toward which its passages open up for us.

Consider two examples of this more directional model of interfacial accumulation. Flight Patterns by Aaron Koblin is a now canonical work of information visualization within new media art. The screen begins with an empty black field. Then indistinct white lines begin to appear and draw themselves into criss-crossing clusters. Soon the lines become denser and denser and the clusters they form more and more regular. In an instant we recognize a larger familiar pattern: the national borderscape of the contiguous United States. The piece is a temporal map of airline flight routes into and out of the U.S.A. in a given day and works to to invert our more traditional view in which the national territory appears first as the structuring condition, cities as a network of destinations and only then as the active motion of flights between them. Here the churning flow of mobility through the individual interfaces of airport gateways accumulates into regional networks which in turn come to reveal a larger territory only through their productive structuring of its formal emergence.

The model of multiple small interfacial events aggregating into regular networks, themselves aggregating into territorial structure deciphers informational mobility as much as physical mobility. NYTE (New York Talk Exchange) is a project by the SENSEable City Lab at M.I.T. that draws upon data gleaned from the pathways of another planetary geography: calls made on the AT&T network from New York City to the rest of the world. A given call originates on a single handset, an interface presenting itself as a standard keypad for calling or a programmed list of contacts. Transnational migration flows into New York”s ethnic enclaves allows for existing social networks to be implanted partially in-tact within a new country, while telephonic networks allow other existing social network connections overseas to be maintained without direct physical proximity, and even new similarly remote networks built upon old. As voice data is segmented into discrete packets and sent over multiple bandwidth infrastructures before being reassembled in the handset of its intended receiver in Sydney, Brasilia, Shanghai, or Bangalore, each arc from point A to point B cannot be directly traced as in Flight Patterns. Another map is required. In NYTE we see the five boroughs striated into a regular grid of call-origination clusters, and here two directions of an interfacial aggregation model are apparent. First an artificial image of the territorial city is drawn as a given field in which smaller sub-sets of clustered units of network activities are located, in which, in turn, individual uses of handset interfaces are indexed. Graphically, the first structures the second and the second the third. However what is mapped is not the way the worldly geography structures the possibility of calls made, but also the opposite of this. The continuous addition of call upon call (of interfacial event upon interfacial event) produces regular patterns of network connection between a particular enclave in the city and the other end of the network edge in a remote metropole. In the image here, these are rendered in the color coded bars, and in the explosion of the single cluster demonstrated here, we can see that between one quarter and one third of all international called made from this location are made to Seoul, Toronto and Porto. These are the networks that emerge from continuous connection (and which likely precede those connections through other network modes such as kinship and economic collaboration). Taken together, the SENSEable City map also shows how the accumulation of all these networks resolves into a specific global geography in which the entire planet can be remapped according to the density of its telephonic proximity to New York (not shown): again interfaces into networks into territories.

V.
Latour concludes, “What Tarde could not have anticipated, however, are the added bonuses of the digital world that now provides and embodiment for this theory, at last: the notion of navigation where we are able to physically (well, virtually) navigate on our screens from the individual data points to the aggregates and back. In other words, the aggregate has lost its privilege it has maintained for one century.” Yes and no. I would argue (and have) that the appearance of planetary scale digital infrastructure and the media by which information flows are rendered from the unit level to the aggregate level are not just contemporary, they are in essence mutually constitutive. The former works to transform centralized structures in the image of the flows they govern (a new regime of the edge ruling the node) while the later transforms the structural image into a recombinant interface that can always be reformed in the image of a new analytical visualization (the datum in conspiracy with its abstractions). The former is a shift in urban processes and the later is a shift in how we know things about those processes, but both a means by which we directly participate in those processes. Where I part ways with Latour (so far as I understand our positions) is that where he sees in these trends a fatal demotion in the status of the structural image, I do not. Quite the opposite. As familiar geographic diagrams of social processes (in Tarde”s expanded sense that “everything is a society”) become fragmented, multiplied, replicated and as the discontiguity of the networks they image is exacerbated by globalization, there is, as I see it, an increase in need and desire for images, maps, utopias, political theologies, governmentalities, perceiving rhetorical diagrams that can account for all positions among positions. This is both a cosmopolitan urge and a fundamentalist urge at once. But if we wish to consider how images of the aggregate are socialized among multiple networks, it is imperative to understand them not only as multitudinous indexes at hand but as effervescent icons at play.

Source