[talks]

Notes on Habitat-scale Robotics and its Constraints

[08.2009]

Permalink

Our deep systemic crises invite three interrelated and only apparently opposing responses: modernism, inertia and fundamentalism; fight, hide, and flight, accordingly.        

As part of a broader design policy project that explores the terms of an alternative modernism suited to careering headlong into the challenges of our own planetary accident, armed with comprehensive, caffeinated social programs of computationally-driven systemic redesign-by-subtraction, I hope to outline one truly plausible response to the question that attends these proceedings with us: "why, especially now,  do we need robotic architecture at all?"

Below are four points of further inquiry that will, I hope, steer in some way the goals of others' design projects.


1. The Two Planetary Computers

First things first. The question of robotic architecture is for this workshop couched in the context of the real and proposed ubiquity of computational media and intelligence. But when we discuss ubiquity of computation, what exactly do we mean? Is it only the computational activation of objects, or the scattering like pollen of processors onto landscapes? It has been a point of conjecture from Lucretius, to Descartes and from the Church-Turing thesis and Wolfram that the physical world is itself already computational in nature. But even if we are not hard computational ontologists (and I am not making that claim) it may be a durably useful perspective to imagine that what we are currently designing is not the first planetary computer, but the second; and further that the ultimate purpose of the second is to co-evolve dialogically (and dialectically?) with the first, providing new means to govern fragile worldly systems with lightness and care.  This image of UbiComp is not as the installation of an intelligence-layer on a mute inert earth, but as the provisional modification of sensation, calculation and communication within ecologies that are already far more complex and finely tuned that this clumsy new machine will ever be.  Accordingly the assignment for habitat-scale robotics is first to absorb the permanently overwhelming depth of worldly demands made on it and to better mediate them. Only then, I would add, can the large scale geo-engineering, proposed as necessary in order to redesign climate change by redesigning the climate itself, have a chance to do more good than harm.


2. The Complete Prosthesis: Severance, Recognition, Intentionality
        
Robotics has a specific position in this, and it brings with it a particular set of terms and limits. While any machine is to some extent prosthetic, robots can replace entire bodies. They prostheticize not only a hand, as a hammer would, or fingers as would a loom, but can assume the place of whole persons or entire animals, or crowds or flocks. Thereby robots take on more identarian complexity, and invite the redesign of humanity and animality through them as models.  As such, robotics-as-invention is quite old. Golem was made of clay, but animated by the incantation of ritual linguistic code, a sort of proto-software. Mary Shelley's Frankenstein's monster was an assemblage of cadaver component debris brought to life and to consciousness by electrical current. It is hard not to read the latter figure as a therapeutic response to the uncanny liveliness of industrial machines and how they made human bodies appear newly machinic in their reflection. The interchangeability of human, robotic and quasi-robotic labor become such a commonplace of the twentieth century, that it is really only now that its too-obvious-to-be-articulated logic can be outlined in retrospect (for example see Bruno Latour's essay "Sociology of the Door" which describes in detail what happens when a doorman is replaced with an automatic door-opening and closing mechanism.)
          
Of course in his first employment of the word "robot," Karel Capek derives it from the Czech word for forced laborer or serf; a slave body, more or less, a completely amputated avatar that retains or extends the profile of its commander. In the USA, where slavery built an empire, the robot cannot not take on a native politics of both fear and wish-fulfillment. In the post-WW2 boom years, the trope of robot as servant (as opposed perhaps to the more contemporary robot-as-brick) brought the marketing hype of both hedonistic push-button interiors, quadrophonic stereos rising out of waterbeds on the one hand, and the kindly chaperone automaton on the other. The former is a magical expression of desire and prowess now able to command worldly objects with its voice, while the latter is the interjection of pure, inhuman rationality into situations maintaining autonomic systems and scolding from a knowing distance. Playboy vs. Alphaville. A cursory scan through post-War science fiction and technology fairs is full of both: Id robotics and Super-Ego robotics. To some extent these figures linger and will a while longer, ... but my real interest is elsewhere.

Robotics is now more bottom-up than top-down, more bug than brain. This traces the broader movement from centralized technological systems, what Lewis Mumford called "megamachines," into decentralized, atomistic, quasi-autonomous self-organizing computational components. The shift from centralized top-down authorities to fragmented network robotics also parallels a transformation in the topology of political authority in general. For good and ill, in the privatization of economies designable intelligence explodes from the center into a million peripheral nodes each moving and evolving without a single common plan, command or control system (only libertarian ideological codes). Doesn't the story of central planning overwhelmed by random bottom-up speculation frame one aspect of roboticist culture's research epistemology as well? Today however as computing moves (back) to vast, ambient cloud platforms, does that distributed autonomy now abdicate or dissolve into a higher metarobotics? Is this how the Singularity stands for robotics' own Theological Turn? (For a different "cultural" register, recall that psychoanalytic theory has made much of the uncanny animation of quasi-human machines as part of an experience of haunting.  Robotics in any form is at first a supernatural proposition: the recognition of that numinous agency in robotic animation is a precondition of the later more pedestrian experience of anthropomorphic recognition and prosthesis.)


3. Light not Heavy: Programs and Programming  

Going back to Latour's "doorman," it's important for the question of robotic architecture that his scenario is one of human/architectural counter-placement and replacement. One of the important interfaces between environmentally-active software and socially-resilient architectures is the division of "programmatic" labor; or put another way, we now ask software to do things that we used to ask architecture to do. Software programs have, to a greater or lesser degree absorbed architectural program (though the chain of transference certainly flows in the other direction as well).  The evolution of telephony, and the social network access it brings with it, from a fixed architecture-attached appliance to mobile extensions-at-hand is a well-understood example of this, but perhaps every institution has experienced analogous translations.        

Consider the bank. Throughout the 18th and 19th century, bank architecture's typological plan became more and more stable, generalized and specific to the programmatic needs of storing currency and distributing and accepting currency securely and privately to mass markets of customers. The architecture was a machine of this program, linked as it is with the gothic-classical mannerisms in stone and concrete that communicated stability. However, in the 1960's and 70's as information networks began to link banks and financial institutions, money became increasingly virtualized, further removed from referent substances like silver and paper bills, and now circulated as pulses and registers in databases. Perhaps not coincidentally bank architecture shifts at this moment from classical scale to minimall cheap. The bank become more of a node in this network to which customers could come and access virtual funds transubstantiated for them into hard currency. This service however didn't really need the entire architectural apparatus of a bank in most cases, and so the bank's 'currency internalization/externalization program' was miniaturized and synthesized into a visible array of push-button options and installed as ATM's on the side of the bank building.  What had first existed as architectural program now existed as graphical user interface, what was first a plan is now a surface feature. Later of course, the ubiquity of information networks allowed the ATM itself to be disembedded from the building altogether and to float more freely on any screen to which secure information could flow. In successive steps architectural program became software program.        

This shuttling of program from structure to equipment is fundamental, and not something that first appears with computational telephony networks. In the design of any environment there are core choices to be made about what programmatic features are to be born by architectural planes and contours and which by furniture or installed machines. Sitting, standing, shade, sleeping, etc. can all be built-in to the building or they can be included later by the introduction of specialized furniture. Furniture allows for generic structures to be customized according to specific needs, or considered the other way around, architecture is a collective furniture that makes the most dramatic enveloping demands on posture. Between the two is a give and take of programmatic responsibility, and a parallel exchange is at work between hardware/software's role in a space, and the physical space itself. Program is sorted and shared and always shifting. For this very reason, it has been long assumed (ad nauseam) that robotic architecture would allow for more specifically "customizable" habitats. But for it to do so it must prove capable to providing degrees of freedom than furniture cannot and/or capable to providing the same programs with greater mobility and efficiency, as ATM's and online banking did. Until then furniture wins because it is cheaper, lighter and simpler.        

So then how does robotic architecture justify the appearance imagined for it? I strongly argue that it must do so as to reduce the amount of architecture in the world, not increase it.

As we are well aware, our built habitats are critically inefficient in their net usage of energy and resources. As with all finite material cultures, it is imperative to radically increase the usable efficiencies from them, to adapt, recycle and re-use. It is neither necessary or likely even possible to erect architectures over the coming decades to suit the complex programmatic needs of another three billion people. It is simply not an option, even in concert with an even more ambitious program of subtraction and replacement of existing building stock.         

This is where the programmatic meta-flexibilites provided by robotic architecture come into view. Consider by way of one example, how much dead time most architecture endures. Offices, storefronts, entire malls, stadia, etc. are active during the day and dormant at night while in many cases residences are vacant during the day.  John Thackara often uses the example of how schools and their cafeterias are largely vacant overnight in neighborhoods without sufficient public food distribution facilities. It is obvious, and perhaps even inevitable, that multi-use programs will become more commonplace, segmented by temporal switchovers where one occupation replaces another. The introduction of both blunt and nuanced robotics into this context can make possible forms of urban-scale spatial recombinancies otherwise unimaginable or impractical. In this robotics actually is less a displacement of human bodies (not currently in the business of moving partitions and floor plates on demand) and instead puts the general capacity of thermodynamic reformation at the disposal of software-level media. If it is displacing anything it is zoning, jurisdiction and certain legal governances of place, as well as the unsustainable presumption of monoprogrammatic architectures.


4. Adaptive Augmentation, Sensing and Governing 

This path implies for architectural roboticists a brief that is far more specialized than the open-site total project usually presented to students in design studios. It implies that robotic megaprojects (towers in Dubai and OMA Transformers in Seoul, etc.) are exceptions to the emergent rule of robotically-augmented habitats. Other than to flatter the curiosity of the idle rich, the scalable service and product model to cultivate is based less on show business and more on adaptive re-use and strategic retrofitting. Civil engineering is the key discourse here, not the Luxury Arts, and this distinction could not matter more.  Further, much of the important impact of such programmatic metaflexibility would not be for the direct benefit of human inhabitants. Our architectures-for-things already out-innovate our architectures-for-people. The glossies have it all wrong. The most advanced technological environments are in factories, warehouses, ports of call, military installations, server farms, etc. and are spaces for objects, molecular and informational, not people. Here objects and partial-objects are sorted, assembled, disassembled, stored, heated, cooled, sliced, smoothed, and calculated in all manner of ways. Additionally I would posit with confidence that the amount of energy spent for our benefit on the processing and circulation of stuff is comparable to that spent on the maintenance of human climates. When we consider opening the programmatic horizon to truly recombinant flexibilities for the strategic purpose of lessening total architectural overhead, and looking to tactical robotics as a key means to this end, primitive versions of the profiles and procedures of this world can be glimpsed inside of your local FedEx's facility or your regional big pipe datacenter. Imagine these nocturnal programs springing to life inside of dormant storefronts, parking lots, and office buildings and you have the fuzzy beginning of a proper urban design brief.        

Supply-chains provide another important model: discontiguity. Obviously the robotic architecture with first implication is that of a mutable building-form, a robot that prostheticizes in situ the collective embodiment of a single site and habitat. But beyond this are network conditions, both nodes and edges, than are available to roboticist manipulation, not only automating habitual actions of space users, but calculating, monitoring and governing the spatiotemporal operations of sites that are already ecologically and economically interrelated at a distance. But of course all is not well in this object world, as the regional spread of many of these logistical networks are exemplary of the same unplanned sprawl economics that contribute to the environmental crises at hand. The itineraries of food from continent to continent to megaport to regional sorting center to supermarket a particularly dramatic example of these inefficiencies, and demonstrates the dubiousness long-term robustness of our entire social-structural logistics systems as a whole.          

While such "footprints" are always inevitably contentious calculations, the lightness of robotic architecture and the lightness it enables in architecture we've inherited will enable its emergence, heaviness on both counts will retard it. Usman Haque's Pachube system for the monitoring and communication of all manner of physical and virtual architectural object-actors (humans, weather, digital blobs, etc.) points to an intriguing set of possible practical operations that might make robotic lightness possible because they make it measurable. Unlike most environmental and architectural operating systems, Pachube best works as a multitudinous network operation, linking multiple sites and their durations into plural spatial datasets. In this its computational agenda is modest in comparison with the procedural complexity of the precedent world it measures. What should be of strong of interest to roboticists is the capacity of such monitoring systems to become more than diagrammatic maps of real events, but instruments of control and redesign. As I've described in many other works, interfaces begin as diagrams and dashboards and once infused with sufficient computational capacity and networked instrumentality, those images become image-instruments capable of controlling what they measure, and not just index it. Robotics becomes in this the specific instrument whereby the intelligent monitoring of local and regional activities can become in their rhythmic aggregation, the causal agent of mechanical and programmatic modulation.
          
A preferable future of robotic architecture is a one of prudent content management, in this case of the existing stock of architectural partitions and surfaces. But, alternatively, it is also very possible that they added energy costs necessary to manufacture and power broadly-used robotically-governed temporal programs will always and forever outweigh the incremental efficiencies realized by flexible architectural forms. It is this equation that robotic architecture must solve if it is to realize itself at any historically appreciable scale.


First Author

Benjamin H. Bratton
Associate Professor, University of California, San Diego
Director, Design Policy Program, CALIT2
HYPERLINK "mailto:benjamin@bratton.info" benjamin@bratton.info
 

Citations

Bratton, Benjamin, "The Turing City?" Mediascapes symposium, the Southern California Institute of Architecture. June 11, 2009.

Latour, Bruno, "Where are the Missing Masses?: The Sociology of the Door" 1992.  HYPERLINK "http://www.bruno-latour.fr/articles/article/050.html" http://www.bruno-latour.fr/articles/article/050.html

Mumford, Lewis, Technics and Civilization, Harvest Books. 1963.


Footnotes

For a more complete survey of this see my paper "The Turing City"  from Mediascapes conference, SCI_Arc. 2009. online at bratton.info.  At CALIT2 where my Design Policy Program is headquarterd, we have designed several optiputer systems. An optiputer is in essence a Von Neumann machine depunctualized and spread across the world, each component connected by direct fiber optic channels --hence the optical in optiputer. For example, a server farm may be housed in San Diego, plugging away at petabytes of data stored in Tokyo, drawing input from sensor sources deep undersea in the South Pacific, all displayed on half-gigapixel display systems in Switzerland. Whereas the internet may be understood as a planetary ecology of information flows, it is not (yet) itself a computer, or more specifically its core computational effects occur at specific server/client nodes. The optiputer model is a single computer, that is global in scale.

This phrase "The Theological Turn" references the renewed interest in the works of Emmanuel Levinas, Paul Ric?ur, Jean-Luc Marion, etc. in light of the centrality of political theology in cosmopolitical discourses.

See, for example, Laurence A. Rickels, The Vampire Lectures, University of Minnesota Press. 1999.

See John  Thackara, In the Bubble: Designing in a Complex World. MIT Press. 2005.

To keep up to date on the clumsy childhood of the latter project, see  HYPERLINK "http://www.prada-transformer.com" http://www.prada-transformer.com/

For a short list of Pachube-based applications see,  http://apps.pachube.com/

Tags: culture industry, theory, branding

Published: 08.10.2009

Permalink