The Impossibility of Interface

author: Matthew Fuller

Media Design Research, Piet Zwart Institute, Willem de Kooning Academy Hogeschool Rotterdam

A version of this essay appears in Behind the Blip, Autonomedia, 2003

One of the key interests of the media design programme at the Piet Zwart Institute is how digital media is embedded within, echoes and reinvents things happening in other parts of life and society. This lecture takes one part of computing culture, namely ‘interface’, and examines it in the light of other forms of interface within society. Different grammars of power relation that such interfaces make possible or rely on are also discussed. Such contextualisation, in turn, suggests ways of developing digital media with an awareness of its social ramifications

The graphic user interface allowed computational power to leap from relatively isolated positions - for example, from expertly attended payroll-management machines used in the cores of banks - into new forms of relation with people and other processes. It allowed computer users to develop a visual and spatial sense of what they were doing. What are the terms of this interrelation, and what do software interfaces have in common with other forms of interface? A working definition is provided by the renowned media designer Brenda Laurel. "An interface is a contact surface. It reflects the physical properties of the interactors, the functions to be performed, and the balance of power and control."

Using this definition to develop the above questions, I want to run a consideration of interface through a speculative typology of modes of interface which goes as follows:

• Interface as distributed throughout and indivisible from the system of which it is part.
• Interface as monitoring and control of a reductive, indexical map of separate elements that can be changed from state to state, but not altered.
• Interface as an associational structure which allows a user to manipulate, alter, destroy, and multiply processes and objects from which it is independent.

Software: Information and Matter

Is what is suggested here, simply a ladder of abstraction? More a recognition of processes becoming internal to the computer. What is meant by this is an extension of what Katherine Hayles suggests when she names virtuality as "The cultural perception that material objects are interpenetrated by information patterns".

We must note that the recognition and interpretation of such patterns is itself the performance of a conceptual device - 'information' - that has certain effects, tending towards the manufacture of a caesura between matter and information. This is always only the effects of an interpretative operation on matter, rather than something 'given freely'. Matter is understood to be always informational, but never pure. If matter is never simply informational we must also flip the pairing. If we are to understand informational patterns as always having a materiality then the interpenetration of informational patterns by other informational patterns is what is discussed here as software. Such informational patterns are always understood to be embedded and manifested in materiality, the architecture of a computer for instance.

Metaphor

Firstly, we need to briefly touch upon metaphor. Whilst, (as in the case of animation or multimedia applications using the model of a 'set', with 'lighting', or the 'darkroom' metaphor of PhotoShop, etc.) they remain as imaginal regimes structuring particular kinds of software, the assumption here is that metaphors will be gradually erased from mass use interfaces. Metaphors, imaginally mapping out what functional capacity a device has by reference to a pre-existing apparatus. However, as functionality outstrips the explanatory capacity of reference to previous media forms, this initial usefulness will diminish. The uneven spread of metaphor-reliant constructs within most applications would tend to support this. Software will need to be seen to do what it does, not do what something else does.

In this respect two uses of the magnifying glass as metaphor are interesting. In Mac OS's Sherlock, a search tool, clicking on a button with a picture of a magnifying glass actuates a search for a string of characters typed into an adjacent field. In a program such as PhotoShop or Quark Xpress the magnifying glass works to perform a jump-cut which magnifies or de-magnifies a selected element of the 'canvas' or 'page' (via either a marquee or simple click whose degree of magnification or demagnification may be set in advance via a tool settings dialogue window).

In the case of Sherlock, on the other hand, what results from the use of the magnifying glass is a list of files in which the particular string occurs. Instead of a mimicking of the capacity of sight, or of the action of lenses, what emerges is a database form: the index. (Such an asymmetry between metaphor and function is incidentally illustrative of the medial incompatibility that faces the construction of image databases.) Perhaps the designers of the interface mistook what they must have taken to be the cuteness of the name of the device for its actual function, or were channelled into keeping a metaphor long after its initial imaginal power had been shed. Perhaps a 'realistic' image of a syringe or a violin, Sherlock Holmes' actual technics of associationality, were beyond them?

I suspect that we can suggest that it is those elements within the wider field of software that are working to the strictures of the version economy - in which a release is demanded every six to twelve months - that are going to maintain metaphor, or have any need to do so. Such a repetition of the familiar is made to keep the mass of users faithfully on the upgrade path. (Increasing the trend to homogenisation, such programs also tend to develop interfaces and functionality which replicate those they compete against in order to make it easier for users to transfer their loyalty from one to the other.)

Metaphor takes a known set of properties and behaviours from one domain - optics or fiction in the case of the magnifying glass - and transfers them to that of the computer as an explanatory or structuring device. With every interface metaphor, there is a point at which its explanatory or structure-providing advantage collapses in the face of the capacities of the ‘universal machine’, the computer, and what it connects to. At the same time, as will be seen later in the consideration of games, there are conditions in which it is precisely this artificiality, and in their use as exploratory imaginal devices, that they have their uses.

I Thought I was Seeing Convicts

I Thought I Was Seeing Convicts is a documentary, composed of 'found' material, by the Berlin-based film-maker Harun Farocki. Much of the footage is of the insides of a Californian State Prison, Corcoran. Cut in with images and sound from the place, and with a prison staff training video, are short sequences of computer interfaces.

The footage from Corcoran surrounds a court case from the end of the 1990s in which prison guards were alleged to have set up fights between members of rival gangs within the prison's exercise yards. As with most US prisons, inmates organise themselves into gangs based on outside loyalties and racial and regional groupings. Introducing a prisoner from one gang into a yard being used by members of another produces a conflict over territory. For the guards, this is a reproduction of the gangs' conflict with each other, rather than with the authorities. It also condemns participating inmates to further convictions. This is crucial under the 'three strikes' system which can be leveraged to start a prisoner off with a sentence for a small charge and, via a series of staged incidents, leave them as permanent residents and sources of budget for staff and facilities.

Watching the yard by video and from watchtowers, guards would bet on the outcome of fights they set up in this way. At other times they would 'break up' the resulting fights by shooting the unarmed people involved.

This video installation is crawling with interfaces. What implications does it have for them?

- Interface as distributed throughout and indivisible

The particular interface I am interested in here is that put in place in the architecture of the exercise yards and the cameras and procedures used to control and to execute the men in the yards. The user is a prison officer, or a command structure of prison officers. The shape of the yard is like a slice of cake, a segment. At the apex of the segment is a video camera. The shape of the yard is built precisely to the field of vision of the camera. Watching the feed from the camera is a guard.

If we are to take Laurel's definition of an interface as, "A contact surface. It reflects the physical properties of the interactors, the functions to be performed, and the balance of power and control." We are in a quandary. Where exactly is the interface? In the architecture, the shape of the yard that perfectly matches the area viewed by the camera? In the shape of the camera’s lense, its refractive capacity? The circuits that turn the light from the lense into a series of pixels? The slow scan of the stored video images? The height and relative scalability of the wall? In the colour of the walls? In the minds of the prisoners? In the regulations governing the behaviour of the guards? In the range and calibre of the guns used on the prisoners? In the skills used to stage a fight and select a target? In the relations between prisoners? In the records kept on prisoners and groups of prisoners? In the multiple economic and racist mechanisms in which the prison is embedded and which it, as an industry, depends upon and reproduces? In the legal system? In the system of property? In the idea of an individual subject accorded rights? In the protocols of wounding, honour, loyalty and fighting within the tacitly approved gang system of the prison culture? Of course, to ask such a set of questions is an impossibility, to hold a thousand enquiries at once.

We cannot here consider the interface between guards and prisoners to be solely representational in the way that a standard computer desktop is sometimes understood to be. The interfaces here operate through multiple procedures. Whilst the system they instantiate is fundamentally hierarchical, the interfaces also operate by means of networks of mutually reinforcing patterns, ideology, structure and technologies. What the notion of interface allows us to do here is to analyse how they link, how one process passes from one axiom into another, and how processes are reconfigured, stripped down, simplified or made amorphous from their passage from one medial, architectural, racial, juridical, regime to another.

It is a particular pathway through these that the guards were able to 'hack' in order to realise a level of brutality that could not be officially countenanced, or even acknowledged as having happened.

The events at Corcoran stage, in a particularly twisted manner, the transition that the philosopher Gilles Deleuze notes when he talks about a shift from a disciplinary society which operates by compartmentalising life, by ordering and naming, to a society of control - in which behaviour is continually modulated, using as often as not the device of free will. Examples might be electronically triggered ‘congestion charging’ for road traffic, as recently introduced in London, or the crude lie-detector mechanisms being used by some insurance company call-centres. In a sense, what is discussed here in terms of interface is how, even in the archetypal domain of discipline, the prison, these two modes of power can combine or alternate. Under discipline, such abuse of the prisoners is ‘unfeasible’, under control, manifest in the prisoner’s choice to enter the yard, it is a ‘natural’ result of their behaviour which ‘calls for’ disciplinary sanction. Flicking the switches between these modes allows the guards the maximum repertoire of cruelty.

To describe such a structure as a prison as a series of interfaces risks flattering the libido of the prison, it becomes something separable, the discharge of a series of pre-ordained functions. A rational, managerial process. Discipline as a 'mold' allows the interface to remain something discrete, neutral. Control as a process of constant 'modulation' is that neutrality gone mobile, that soaks into everything, but that chops and swerves, demanding constant renewal of adherence to codes and processes. This is what the guards with their walls, cameras and guns were able to manipulate, the interface between discipline, in its predictability, and the vaporous insinuations of control.

At the same time, it forces us to reject any accounts of interface as being solely or primarily to do with representation, about the manipulation of elements that are separate from the structure of the surface layer. Interfaces code in advance how and when something occurs, but cannot necessarily determine it (being in the case of computers always in negotiation with another part of the system, CPU resources for instance) or its uses. The vectors that connect one thing to another, an instruction to an object, a node to another, a layer to a filter, are always political at the same time as they are technical and aesthetic.

At their Palo Alto Research Centre, Xerox once staged a well-known experiment with the tagging and positioning of workers (in a type two interface). This enabled co-workers and managers, as they moved around the building in the course of a day, to constantly find the physical location, activity and nearest phone or pager number of others. The tracking system also carried data on their availability (or not) for communication for various levels in the hierarchy. As media designers are generally amongst the class of worker that is allowed to experiment upon itself, such a system was soon dropped.

- Interface as monitoring and control of a reductive, indexical map of separate elements

Alongside the footage of the prison, Farocki cuts in, (on the two screens the interface plays on) material from other kinds of control systems. One piece of software is a schematic rendering of a building's floorplan. It allows an operator to switch lights on and off, to open and lock doors by clicking on symbols embedded in the schematic. Other footage shows a stock recognition system built into some kind of surveillance device. The main window on the screen is video feed from a camera following someone in a supermarket. As they pick up and examine articles from the shelves, a series of pop-up menus allows the operator to specify and record the items that the target touches.

Richard Sennnett describes an interface of this sort in a Boston bakery. A 'user-friendly' interface had been installed at the front end of a baking system to allow workers to control bread production without the need to deal directly with the raw materials, to make judgements about the readiness or consistency of dough, to shape loaves and so on. The machinery allows flexible specialisation rather than mass production. The work is done by a transient force of part-timers or temporary workers supervised by one foreman. It is deskilled, operating the system demands pressing buttons on a screen, and by: "Monitoring the entire process via on-screen icons which depict, for instance, images of bread colour derived from data about the temperature and baking times of the ovens; few bakers actually see the loaves of bread that they make. Their working screens are organised in the familiar Windows way; in one, icons for many more different kinds of bread appear than had been prepared in the past - Russian, Italian, French loaves all possible by touching the screen"

The interface, however, does not match up to the baking process every time. Much of the product is wasted when the actual temperature, the rising of the loaves, or some other factor, fails to match the representation on screen. There is a skip outside full of burnt loaves, victims of automated friendly fire. Workers controlling the process via the interface have no need for an understanding of how to bake bread. The process is illegible to them.

Whilst the interface is extremely easy to use, its relation to the procedure it manages is flawed. The monitoring and control of the baking process is not deep or detailed enough to allow this. Additionally, the 'sociotechnical' and economic arrangements of the bakery rely on having a workforce with no actual baking skills - this would increase costs.

Sennett describes 'alienation' experienced by an older generation of bakers as 'the unhappy disassociated consciousness which reveals, however, things as they are and where a person stands.' By contrast, "Flexibility creates distinctions between surface and depth; those who are flexibility's less powerful subjects are forced to remain on the surface."

The closing and shutting of doors, switching lights on and off, shown in Farocki's film is very similar. Elsewhere, security guards working twelve-hour shifts on minimum wages, watching monitors, wiping bar codes, checking faces have the same relationship to the processes they control, and those that shape their jobs, as the push-button bakers. Like call centre workers, working with a series of written scripts for their speech, menus, forms and databases, plus a time-per-call quota, the task is to be an interface. As with Laurel's description of interface, this is a surface that is thick, that bristles with connections, blockages and channels, variable speeds, timeframes and routines.

Performing a surface, manufacturing an alienated interface, double consciousness, is an essential capacity of life in the 'mode of information'. In order to get space to breathe, we must create faces, bubbles, blips, strings of delay, backchannels, work-arounds in the domain of work, even if only in the name of efficiency. This, rather than submerge into the well-bloodied trap of 'realising our full potential' in the terms of a rigged productivity. Therefore, whenever an interface promises to make something 'clear' or speaks of allowing something to work in just the simplest way possible it must first of all be assaulted with questions, yawns and scripts rather than rewarded with immediate identification.

It is this gap between a model of function and its actuation that in some cases describes a degree of freedom, and that in others puts into place a paralysing incapacity to act - it would not, for instance, be over-welcome in the steering system of a plane. The politics of the indexical map is produced precisely in its power to command an isomorphic relation between itself and what it controls. It must not, it can not, slip...

- Interface as independent associational structure

Only at this point does an interface explicitly correspond to the logical processes of a computer: to write, store, delete, read, and to calculate on the basis of these functions. It must be added that the interface of the third kind is never, of necessity, permanently 'independent' of those elements that it provides an associational mechanism for. It may, given an appropriate level of access, be rewritten.

In order to develop a provisional understanding of this type of interface, I want to approach it through videogames. Steven Poole's book, Trigger Happy provides a good starting point for understanding the dynamics of interface and what might be a kinaesthetics of information. One of the key themes in the book is how interaction is enhanced by the internal data-structures of the game, the models it has to govern the motion and behaviour of objects, vehicles, bodies within the game, and how they behave on screen in relation to the user-inputs
"…writing a racing game for instance… …the car can be defined as a certain mass resting, through a suspension system, on four wheels, which have a certain frictional relationship with the road. From this very simple mathematical definition, it turns out that that 'realistic’ car behaviour, such as oversteer and understeer, load-shifting and tilting come for free."

What Trigger Happy shows here is that the simple application of Newtonian physics, acceleration, gravity, inertia and so on rather than a 'full' model of the material world creates a small set of axioms which combine to give a rich array of play interactions. Indeed Newton gets dropped swiftly enough when a little mutation makes for more exciting game-play or opportunity for a cartoon-style plasmatic world. Just as the 'realistic' movement and handling of a car is gained by the mutual interplay of various mathematical models of certain aspects of the mechanics of a moving car, in other games very simple behavioural codes - such as when to bob up and down from behind a crate during a firefight (House of the Dead or any other amount of shooters), or when to flee a room when a leader figure is shot down (Halo) - are also derived 'for free' by the interplay of simple software ‘agents’ governing dominant or recessive simulations of 'emotion' and behaviour in non-player characters.

Much of this useful book is an extended consideration of how the axioms of games contribute to a multiple self-enframing that is either compelling, because of the richness of variations to be brought about by the combination of their various types of capacity, or enjoyable for simplicity coupled with combinatorial unpredictability.

The sweetest pleasure in many games is their coherent range of dynamic properties and the interplay between them. This is something relatively independent of processor power - think of Pong or Tetris - and extends to the ergonomic and audio qualities of a game as well as the visual. Current games platforms allow thousands of tiny variables to be computed on the fly according to complicated interactions of multiply intersecting numerical models. Early videogames, and those still played on a GameBoy or on emulators such as MAME, kept such computations down to a minimum. What is important for Poole is that they are an always consistent set of properties. Thus, what he calls their 'realism' is not predicated on being 'authentic' but on internal consistency. Thus the axioms can be codes, rather than what we understand as natural 'laws', whether these correspond to those experienced outside of the screen or are of a completely novel kind made possible by computerised media is immaterial.

Such insight into games, compare interestingly with one of the classic rubrics of user-centred design, found exemplarily in the work of interface psychologist Donald Norman. The focus should be on 'Interacting with the task, not with the computer’. In a videogame, the task is precisely to perform the interaction with the computer, for as long as it remains pleasurable or compulsive. This is the doubling of the focus of interaction, where one interacts and works within patterns of information, folded into further patterns, embedded in matter referred to above.

Potentially at least, the standard grammatical positions of subject and object, which work-oriented usability theory is predicated upon, become caught up in arrays of dynamics that are processual rather than fixed and to do with the interplay of freedoms and constraints rather than outcomes, it is easy to imagine that we leave matter behind.

However, a reading of video games shows that the best of them - and there are of course plenty which simply translate the tiresomeness of work - still have a concept of task, which is now this very processuality, built in. There, the pleasures of the game become those of multiple virtualities. This is a virtuality as becoming and as involvement, a link between Hayles' use of the term, and that of Deleuze, who says, "What we call virtual is not something that lacks reality but something that is engaged in a process of actualisation following the plane that gives it its particular reality."

Such pleasure is found in how games borrow or, by virtue of interaction, differ from the codes of other formations. Film - in terms of the phenomenological and informational capacities of different 'camera' angles, pace, lighting and framing conventions in various kinds of game; simple narrative modes, such as quests, folk tales, whodunnits, multi-choice gamebooks and how they differ from them in terms of the linear irreversibility of the sequences and the granularity of action; the potential mix of visual codes, photorealism, Albertian perspective, animé style deformation, etc.; how stylisation, reaction, timing and patterning are transducted from different sports, plots, weapons and vehicles, and bring with them gender codes, bodily capacities, different kinds of action amplification and kinetic fluidity, one-game-only finger skills built into muscle-memory, informational stamina, how the interplay of different codes can be thought and sensed into. This 'realism' is also what is gross about many games, gross in the refusal to stage these conventions as axioms, as synthetic.

Now, however, we can return to Norman's formulation of a good interface. The task to be operated on, is now to be operated in: patterns of information operating on other patterns of information. The task is to reformulate the task, to be coupled to that process, to be absorbed. At the same time, the position of the interface designer needs to be blown open. Norman is right when he comments that 'every interface designer is also a system designer', that 'nothing can be designed in isolation'. However, there is within interface design a tendency to close down what counts as an element within a system. (It is an insight, but not sufficient, to establish that a designer of a computer system needs to make sure that it is capable of being cabled up to the other electronic media systems the user might have.) We need an understanding of systems that never stop in their unravelling and invention of new connections.

Neither, however, should connections, combinations of things - things are always combinations - be welcomed simply on the basis of their hybridity. Paul N. Edwards describes a presentation of a computer game with an output beyond peripherals that might 'intuitively' need to be connected:

'As we rode the eye of the bomb to the white flash of impact, we experienced at once the elation of technological power, the impotence and voyeurism of the passive TV audience, and the blurring of the boundaries between 'intelligent' weapon and political will.'

Edwards is recalling Norman Schwarzkopf or some other walking abscess playing video with the bombing of Baghdad in a 1991 press conference. Although this sentence might overestimate the case for the existence of political 'will' as a form of individuated intentionality rather than the playing out of drives and capacities built into various mutually stimulating formations and dynamics in forms of governance and power, it does show that the interplay of certain medial, computational, erotic, political and material arrangements are inherently part of a 'system' that is ostensibly only technical. How such combinations are forged, how they proceed, their constitutive axioms and protocols need to be taken up in a consideration of interface.

The massive gallimaufry that is Microsoft Word has been discussed elsewhere, but it is perhaps useful to compare this form of program, the applications that fill the hard drives of most computers, with the kind of cultures of interface that Poole discusses in 'Trigger Happy'. One distinction is between the class of games known as simulations and those devoted to play.

"Simulation… …promotes in certain genres (driving, flight, games) the primacy of supposed 'realism' over instant fun. A true videogame deliberately simplifies any given situation (imaginary or real) down to its essential, kinetic parts"

Such applications asimulate, for example, a writing machine, a perfection for a certain sort of textual production, modelled by systems designers whose conception of writing is formed by the boundaries of the workplace. This is not to say that they form a coherent whole that is embodied in even one place of writing, indeed some of their elements are of fundamental operative and conceptual difference, but that the composite writing-system they embody is that composed in the conjunction of the series 'writing' with the series 'work'. (And - although no special attempt is made to exclude them as potential customers - not with, say, 'generative', 'automatic', etc.) It is the clutter of 'realistic' models of writing or other processes, which at the same time hide their acculturation, which Word and related programs insist upon that makes them close to simulation games.

At the same time most of computing rests upon simulation at the most fundamental level - for instance, the simulation of CMYK on RGB screens. What is suggested here is that those simulations are not universally navigable in the same way and, as a result, need to be made in a way that is interrogable. Here we can take something up from Poole's implicit suggestion that a more enthralling game is one which immediately connects the user at a deeper level with the underlying axioms of the game. Software that reveals its processes as it enacts them.

The one inside the other

The three types of interface suggested here can clearly also be seen to operate the one inside the other at different moments. When you use a WYSIWYG (What You See Is What You Get) interface in a Desktop Publishing program, you operate on an interface of the second kind, using the conventions of the third kind. There is a reference to an object, the document.

The process is also reversed in programs such as those which allow the automated monitoring of employees by keystroke surveillance or web-logging - an example of the second type - built into word-processing or web-access software - an example of the third. One must be clear that this is not simply something reserved for work subordinates (although it is more likely to be applied the more mechanical the demands of work are) but something unleashed at the outset by automation. At another location, at another speed, within the layers of capital's transduction of skill, one could imagine a program for building up data for the use of expert trading systems operating by secretly monitoring the buying and selling patterns of stock market traders.

The scale or detailing of references to an object is another axis of difference. One distinction between the first and third mode of interface is that once an interface becomes entirely digital, there is no room for a vague action. Comparing the use of computers by artists and designers, one research group notes that there is a reluctance to use computers in the initial stages of a work. Computers do not usually provide for good initial sketches. Binary logic ties any interface action into a yes or no, and any mark into data which is as fixed as any other element. Equally, every object or element of data in a file, once it has been saved from the buffer, has the same status. No matter how many layers are assigned to it, the undo function operates in linear time.

In the theory of information and control systems, primarily using interfaces of the second kind, there is a distinction between first and second order control. In the first case the control of a 'steady-state' of a process is given over entirely to a mechanism - a thermostat for instance. Second-order control is where the operator provides an overview of these processes, interprets and recognises patterns and is able to react rapidly to them.

One model of the 'human data processor' echoes this distinction in the way in which it produces a division of labour between mental processes, a "Co-operation between a high-capacity parallel processing system, which functions subconsciously, and a sequential conscious processor of limited capacity. The sub-conscious processor takes care of routine tasks, and only in unfamiliar environments and tasks is there a need for higher level control of the processing by the versatile, but slow sequential processor.”

Thus, the models of discipline and control are recapitulated at the structural level of the interface and of the subject modelled by it. However, as we have seen, once control folds in upon control, messes with its too easy seriality, opens it to inspection, blocks, breaks, scratches, streams, there is perhaps the opportunity for something else to emerge.

Design: there are people connected to this computer. It will shut down in _10_ minutes

Typical in user-centred design research is the statement to test users trying out a prototype of a new piece of software, "I'm testing the product; I'm not testing you”.
Typical of statements for the back blurb of third-party application manuals is, "Need to learn Blah Blah Version Blech Fast? This is the book for you."
Whilst the usability researcher depends for the effectiveness of their research on generating an environment of trust between themselves and the test user, once the software is inserted into a work environment, the user will specifically be tested against how they match up against the software. The cruelty in this relation is a direct inverse to the institutional niceness of the usability test. The more that is invested in making the software user friendly, the more employer, co-workers and technical staff are justified in demeaning the worker who has not internalised its regime. Research based on making software achieve fitness to desired task is as equally about enhancing productivity and regimes of work as it is about the pleasurableness of a tool fitted perfectly to its purpose.

It is not, therefore, simply a question of expertly making a tool most fit for its purpose. Rather, that we might begin by tracking the various mutations of the axioms, drives and enframements which a work brings together and to find ways of staging that combination without blocking out the conceptual and material potential for recombination in different forms, or any access to them by others.

What follows is to understand that 'Output', can only ever be a provisional term. For the contexts in which most software is used - work - the question, 'what does the user need to get out of this?', as if they were an agent that only comes into composition with the software of their own volition is infinitely loaded. Following this, 'Whole tasks' that can be measured and designed through at every stage to increase or improve their performability, pleasure, coherence, comprehensibility and function only ever exist temporarily. Even when the apparatus for their execution is designed, manufactured and distributed in the millions, the task whose execution they embody, may change or disappear. The devices will be erased or recombined, part used, bastardised. It is ensuring functional openness to this bastardisation which needs to be a primary task of software interface design.

There is no calculus of becoming

What remains though, if we are to take such a discussion of interface seriously, is to ask how this process, rather than a 'position', of interface design can itself be opened up?

The poet Jackson Mac Low had it this way: the poet is, "Pre-eminently the maker of the plot, the framework - not necessarily of everything that takes place within that framework ... creates a situation wherein he invites other persons and the world in general to be co-creators with him! He does not wish to be a dictator but a loyal co-initiator within the free society of equals which he hopes his work will bring about."

Mac Low, in an almost classical statement of the 'open work' policy of some of the later Twentieth Century avant garde, clearly knew what was at stake in the capacity of poetry to generate actual freedoms of perception and language, and - relevant to computational culture - the non-determinate processes of chance often built into his work. At the same time, the claims for the poet as a stable position, as master of signification, the situationist, which might be read into such a passage need themselves to be thrown over to the same processes.

Franco Berardi, ‘Bifo’, in an early text on computer-based media states, 'Whenever a social universe forms adequate to the technological and communications potential of actual social brainpower, we call it Renaissance."

There is something necessarily circular about Bifo's formulation here in that it describes a process that is continually catching up and changing the entities that compose it. When the poetics of creation in Mac Low's terms expand to generate an unfolding of their full social, aesthetic, technical, capacity for resonance, such a moment creates an explosion. A thousand impossible questions are asked. Such an explosion can take place at many levels of scale: the molar politics of class or individual; a flowering coalition of insight, rhythm and sensation familiar, but not often enough, to culture; it can take place in software and by means of software or in those moments where absolutely nothing is left untouched.

In most cases it is, along with the workers in the bakery described by Richard Sennett, entirely sensible to answer interface with utter indifference. There is so much boredom, structural cruelty and stupidity governing the physical properties of the interactors, the functions to be performed, and the balances of power and control they perform and embody. The slow deliberate violence of the state burying its hatred for life inside the body of a living prisoner, a relation embodied in the apparatus directly surrounding them, the buildings in which they are fixed, and the society that provides their cloak is one example of interface, is one which can only be met adequately with its destruction.

But, in other moments and dynamics of interface, it is this unleashing of patterns of potentiality and innovation, that Berardi calls a renaissance, at other times, comes as revolution, others, as the simply careful and persistent attention to opening up a particular range of possibilities by virtue of 'recombinant intelligence' - intelligence which realises its own multiple virtuality - that is what is at stake in what has momentarily settled up as interface. And it is by what it combines with, where it goes, what it makes happen, that makes interface at once so profoundly social and so crucially rewarding of inventive creation.