The cyberstructure

Drawing on medium theory, I now want to explore the idea that to understand patterns in culture we are forced to look inside the structures of the machines - namely the notion that medial changes create epistemic shifts.

Attention to the materiality of software requires a form of reading/ writing of these depths through attentiveness to the affordances of code. By attending to the ontological layers of software, that is the underlying structure and construction, we gather an insight into the substructure and machinery of software. But there is also a juridical and political economic moment here, where wealth and law are used to enforce certain control technologies, such as DRM, and the impossibility of opening them through legislation that criminalizes such access. Software is increasingly mediated not only by its surface or interface, but also by the underlying mechanisms which are not just difficult but also criminal to access.

The digital clearly has a functional dimension, in that it runs processes. But what is also radical about the digital is there is no real separation between data and execution. For the computer and the programmer, all are data flows. In contrast to a factory, where one might use leather and other tools that will allow the production of commodities such as shoes, leather is not used to reshape the tools themselves directly. In contrast, anything structured within code and software can be transformed. So the digital is not only changing the way things are classified and the way in which things and objects are recognized by the system, but also it changes what they are and how they can be used - that is, software acts upon software. This feedforward and feedback mechanism is an important part of software development implementation and creates rapid stages of innovation in computational systems.

We can think about shifts in mechanism, by exploring this process of digital transformation of the basic categories by which a system, process or object is understood, at the early and often public moments in the 'softwarization' process - before the technologies are completely formalized.

The traditional web site is static, but the Internet specializes in flowing, changing information. The “velocity of information” is important—not just the facts but their rate and direction of flow. Today's typical website is like a stained glass window, many small panels leaded together. There is no good way to change stained glass, and no one expects it to change. So it's not surprising that the Internet is now being overtaken by a different kind of cyberstructure. (Gelernter 2010)

This new 'cyberstructure' is the stream. This is also a move to transforming media content into 'apps' or media-rich software experiences and the reconstruction of a new gestural and contextualized interactional layer of computation.

These streaming systems also encourage a different means for recon­figuring the labour processes itself. Computational systems make possible a streaming form of labour, what we might call: microlabour. That is, tiny moments of people's labour computationally monitored and paid for in a process that can be seen as a massive intensification of the pin-making factory structure described by Adam Smith.

These are all examples of intense microlabour practices that the new computational technologies make possible. Workers' labour power is literally incorporated and mediated through the software. For example, the word processor christened, Soylent, described as a ‘Word Processor with a Crowd Inside', pays its ‘workers' via the Amazon Mechanical Turk system to undertake word-processing functions (Bernstein et al. 2010: 1). So, the software contains a function that automatically summarizes a document, not by algorithmic compression, but by literally farming the sentences out to millions of human workers around the globe paid at piecework rates for editing. The programmers wanted to be ‘slightly generous while matching going rates on Mechanical Turk, [so] we paid $0.08 per Find, $0.05 per Fix, and $0.04 per Verify' resulting in a piece work ‘average paragraph cost [of] $1.41' (Bernstein et al.

Were we instead to use a $0.01 pay rate for these tasks, the process would cost $0.30 per paragraph. Our experience is that paying less slows down the later parts of the process, but it does not impact quality [19]—it would be viable for shortening paragraphs under a loose deadline. (Bernstein et al. 2010: 7)

This notion of not only aggregating human beings through software, but also treating them as components or objects of a computational system is indica­tive of the kind of thinking that is prevalent in computational design. Production or consumption is mediated by the creation of code objects to represent activi­ties in everyday life and translate them internally into a form the computer can understand. In many ways this is a discretization of human activity, but it is also the dehumanization of people through a computation layer used to mediate the use of social labour more generally. This also demonstrates how the user is configured through code objects as producer, consumer, worker or audience, a new kind of multiple subject position that is disciplined through computational interfaces and algorithmic control technologies. But it also serves to show how the interface reifies the social labour undertaken behind the surface, such that the machinery may be literally millions of humans 'computing' the needs to the software, all without the user being aware of it. In this case it is not that the machinery represents what Marx called 'dead labour', but in fact that it mediates living labour invisibly into the machinery of computa­tion. Indeed, this is an example of where computation serves to hide social labour such that workers are hidden 'behind web forms and APIs [which] helps employers see themselves as builders of innovative technologies, rather than employers unconcerned with working conditions' (Irani and Silberman 2013).

So, for example, as old media forms, like TV, film and newsprint, are digitized, there is experimentation by designers and programmers on the best form to present media content to the user, and also the most profitable way that a subject position can be constructed such that its practices in relation to the interface are literally inscribed in algorithms. Traditional media are softwarized in this process and the way the content used is mediated through a software interface. When transformed into software, first, a new media object is subject to algorithmic manipulation. One thinks here of the so-called 'casual gaming' systems that are designed to not only present a non­linear entertainment system, but also use gamification techniques to create an addictive environment for users. Game mechanics, such as badges and levels, are used to manipulate the users and serve to maximize profit for the companies using these techniques. In short, media becomes programmable. Secondly, streaming media are any media built around a continuous data flow, and this will likely be the paradigmatic form of media for the future. This means that media will increasingly be subject to regular and repeating computation. Again Gelernter usefully explains,

If we think of time as orthogonal to space, a stream-based, time-based Cybersphere is the traditional Internet flipped on its side in digital space­time. The traditional web-shaped Internet consists (in effect) of many flat panels chaotically connected. Instead of flat sites, where information is arranged in space, we want deep sites that are slices of time. When we look at such a site onscreen, it's natural to imagine the past extending into (or beyond) the screen, and the future extending forward in front of the screen; the future flows towards the screen, into the screen and then deeper into the space beyond the screen. (Gelernter 2010)

To further explore this notion of real-time streams, I turn to Twitter, as the current instantiation of the real-time streaming message bus. As a form of computational media that is highly social, it presents an interesting case study in relation to our public/private experiences of communication through a computational platform. Here though, links need to be made between the aesthetic, the phenomenological and the digital. Indeed, the real-time stream is fundamentally a reconfiguration of temporality, a new construction of experience, which is structured around a desire for 'nowness'. But Twitter messages are never marked as 'now', instead they lie already in the past, whether one second ago, one hour or one day. Twitter is nonetheless oriented towards the future, the possibility of something else happening, others in the loop, pure potentiality or unfolding. This is partly constructed through an aesthetic experience of commitment to a radical now, an experience mediated in real time through Twitter's service requiring users to send tweets, twitpics and links that document the experiential aspects of their lives. So at the individuational layer we see profound changes in response to reconstructions of the levels of the interactional and social. So Twitter becomes, like all content-driven platforms, a vicarious experience; both in terms of the person who is live-tweeting his or her experience while, let's say, at an art exhibit, and in terms of the vicarious thrills of the followers - some of them located hundreds or thousands of miles away - who view the work through the eyes of the Twitterer.

So what does it mean when life is vicariously viewed? It's not dissimilar, at least formally, from watching an artwork being remediated through a visual medium, such as television. But the difference is that the television camera focuses attention towards the artwork, rarely able to capture the experiential quality of the presenter, who, using tried and trusted media structures, presents it as either a lecture, a documentary or a narrative. Within real-time streams, however, the artwork and the user merge, the pure singularity of the user - their wobbly half-focused photos, their exclamatory tweets, the geo-location updates - is fused with the artwork they are experiencing. A new kind of algorithmically mediated phenomenological code object. The two are therefore presented as one, less an object of attention rather transformed into a happening, an unfolding or a pure experiential social media event. In some senses then, this begins to highlight connections between the real time as constructed by digital technologies and Adorno's notion of identity thinking.

The other side of the coin, that of individuals using the real-time stream as a site or space for expression, is still nascent. It presents particular challenges to the user, particularly in relation to untethering the experience from the expression of tweeting it, more so if the expression is the collection of tweets themselves. The radical temporality, the short life span of the tweet, also presents interesting new challenges for an expression or intervention that might disappear in the blink of an eye. Additionally, the architecture of Twitter, its retweets, hashtags, URL shorteners, and @mentions, all subtract valuable real estate from the single tweet, which, being only 140 characters long, is already a terse real-time messaging format.

Some artists have already experimented with this medium - Cory Arcangel has appropriated its search function for Great Twitter searches Volume #1 and #2 (Archangel 2012); An Xiao has used it for art projects, like tweeting daily minutiae in Morse code for the Brooklyn Museum's 'Istfans Twitter Art Feed' (NYFA 2012), and many artists from different disciplines use it as a soundboard for their work - these are interesting examples of art in the real-time stream, but perhaps future artistic practice is signposted in the ability of a tweet to remain on the surface of consciousness, to bob and dance across the real­time stream by raising the desire within the user to retweet and @mention it.² This points to aesthetic possibilities using the statistical clouds generated by the real-time streams of the industrial internet - as a form of 'social art' or 'social design' drawing from existing practices of social reading (see Hastac 2013 for a discussion of social reading).

One of the problems we have when thinking about the new real-time media streams, like Twitter and Facebook, is that our metaphors are usually based on old media forms (e.g. television or radio, or even, in some descriptions a cocktail party). For example, thinking about Twitter as television means that there is a tendency to think in terms of channels, programming, a passive audience and interactivity limited to selecting the channel to watch. It is no surprise thatYouTube borrowed this 'comfortable' metaphor to explain to users how to use it. However, these metaphors can also stifle how we understand a new media platform, and perhaps even shape the way it is further used. Here, I want to argue that if we are to understand the medium of Twitter, then we should try to look at the properties and affordances that are particular to it as it currently stands: (1) it is a real-time medium; (2) information flows through it, if you miss it then the information passes by (although through data resellers you can purchase a Twitter archive and replay it); (3) it is based around an interest graph (things people find interesting rather than friends or colleagues) (Barnett 2010); (4) it is surprisingly flexible, if terse, and even with the 140 character restriction, the content of a tweet support urls, links to any other form of media (e.g. web/photos/video), etc.; (5) the entire tweet (as a JSON data entity) can carry location data, user data and other contextual data that makes it extremely powerful; (6) it has little active memory as a system, until 2012 you could only retrieve the most recent 3,200 tweets per person (Haines 2011). Twitter has recently allowed data reseller partners, like Datasift, to access Twitter's Tweet archives, however, Twitter also passes its entire archive of public Tweets over to the Library of Congress³ for storage and research after 6 months for internal library use, for non-commercial research, public display by the library itself and preservation (Twitter 2010)⁴; (7) Twitter has a client/server structure rather than a peer-to-peer architecture, so should Twitter go down, then so would the ability to share on the network.⁵

These qualities seem to suggest a medium, or platform, that is more similar in some respects to a newswire system or news agency data feed, such as provided by Reuters or Bloomberg. These are extremely fast data transmission systems usually built around a proprietary data protocol generally for transmitting real-time news, data and financial market information, for example, Thomson Reuters uses RDF (Reuters Data Feed). These data feeds themselves are extremely low latency, in Reuters' case financial, text and numeric data took less than an average of half a millisecond to move through the system in 2005, and would be markedly faster today (Quinton 2005). We could therefore think of these streaming systems as, in effect, the mature mass-data versions of the system of original stock tickers introduced in the late nineteenth century, invented by Edward A. Calahan in 1867, an engineer associated with the American Telegraph Company.⁶ This was,

a printing telegraph with two independent type-wheels, placed under a glass bell jar (to keep off dust) and powered by a battery. The wheels were mounted face-to-face on two shafts and revolved under the action of an electromagnet. The first wheel had the letters of the alphabet on it; the second wheel had figures, fractions and some letters. The inked wheels printed on a paper tape divided into two strips: the security's name was printed on the upper strip and the price quote on the lower one, beneath the name. (Preda 2005: 755)

The stock ticker produced a constant flow of information that was sent by telegraph around the country about news and price movements on the stock exchange. The telegraph, of course, was also used to send personal messages as well as stock information. Indeed, Schott (2009) speculated that tweeting using the Telegraphic codes from 1891 might return within the terse format of a Tweet, he explains,

The 140-character limit of Twitter posts was guided by the 160-character limit established by the developers of SMS. However, there is nothing new about new technology imposing restrictions on articulation. During the late 19th-century telegraphy boom, some carriers charged extra for words longer than 15 characters and for messages longer than 10 words. Thus, the cheapest telegram was often limited to 150 characters. (Schott 2009)

With the growth of stock exchanges, the stock ticker companies developed a client/server structure with price changes being transmitted back to the centre as a hub, which were then resent, through spokes, to the provincial exchanges. For the users of the ticker service, unlike the telegraph, the stock ticker printed its message out as a continual unfolding stream of information on paper which could be read by anyone,

The flow of price variations visualized the results of ongoing conversational exchanges, and disassociated their results from the individual authority of the participants in those conversations. At the same time, the flow linked the results to each other, made the ties that bound them visible as the tape unfolded, and made the market in its turn visible as an abstract, faceless, yet very lively whole. All the felicity conditions that made the speech act valid (intonation, attitude, look, wording, pitch of voice, and so on) were blanked out. Authority and credibility was transferred from the broker's person to the machine. The flow of figures and letters on the ticker tape became an appresentation.... In other words, perception (of price rhythms) and representation (of floor transactions) fused together. (Preda 2005: 763)

This ‘blanking out' of the speech act is, of course, very familiar to users of Twitter. This medium, due to its reliance on 140 character messages, ensures a terse text that tends towards a factual and news-like message. It is interesting to note that a great majority of tweets are found to be only around 40 characters in length (Ayman 2010, cf Hubspot 2009). It is, perhaps, little wonder that the news organizations have adapted to using it so easily.⁷ Indeed, users have also taken to Twitter and Facebook in such a remarkable way, even more so with the growth of mobile app versions of the services. The similarity of ticker users becoming entranced by the information on the stock tickers to contemporary users of Twitter and Facebook, who also stop and stare at their screens and seemingly fall into a trance, is striking. In his reminiscences, Richard D. Wyckoff (1934a: 37), a stock operator and pioneer of financial chart analysis, wrote that ‘in 1905 friends of his could sit and watch the tape for an hour and a half without any interruption' (Preda 2005: 766). Wyckoff had trained himself to sit still at the ticker so that he could watch the ticker tape for up to an hour. He also remembered how in 1907, James R. Keene, the financial speculator, fell into what he called a ‘ticker trance',

I used to stand facing him, my left elbow on his ticker while talking to him. He would hold the tape in his left hand and his eye-glasses in his right as he listened to me, then on went the glasses astride his nose as he bent close to the tape in a scrutiny of the length that had passed meanwhile. I might be talking at the moment his eye began to pick up the tape again, but until he finished he was a person in a trance. If, reading the tape, he observed something that stimulated his mental machinery, I might go on talking indefinitely; he wouldn't get a word of it.... He appeared to absorb a certain length of tape, and to devote to its analysis a specified interval, measured by paces. Sometimes he returned to the ribbon for another examination, followed by more pacing. (Preda 2005: 766)

The similarity of users absorbed in their mobile devices watching Twitter is striking and this ‘Twitter trance', or attentiveness to the temporality of Twitter, disciplines the user to keeping ahead of the stream and by being ‘in the stream'. The user has to develop a temporal awareness fitted and co-constructed by this technical device (or agencement), whether delivered via the browser or an app on a mobile phone. Again, the similarities to the stock ticker are remarkable.⁸ In our increasingly financialized world this democratization of stream-based data flows as a disciplinary practice and its connection to financial flows is instructive, indeed,

The stress on observation and attentiveness fitted in very well with the overall discourse of the “science of financial markets',' so popular in late 19th century. It required of investors precisely those qualities preached by manuals: attention, vigilance and constant observation of financial transactions and of price variations. For the investor, it is only reasonable to follow the market movements and to try to be efficient. (Preda 2005: 773)

The temporality presented by Twitter, in particular, is a constant real-time social media flow, which is modulated by the activity of the users on the stream. Time is referred to in the past tense, with even the most recent tweet having happened a few seconds ago, and marked as such to the user. Each tweet is placed within its temporal box, neatly segmented in a structured textual format by the other tweets in the constant flow of time represented by technical time (see the raw Twitter public timeline here as JSON data).⁹ As Preda (2005) described with the stock ticker,

a ragged temporal structure was replaced by a smooth one, with the conse­quence that price variations became visualizations of market transactions and objects of symbolic interpretation. The ticker made market exchanges visible as they happened, disentangled them from local conversations, and transformed them into something that is both abstract and visible in several forms to everybody at once. They are visible in the flow of names and prices on the paper strip, but also in the financial charts, which are nowadays also produced in real time. The quality of price data changed: instead of multiple, discontinuous, heterogeneous and unsystematically recorded prices, we now have single, continuous, homogeneous, nearly real-time price variations. (Preda 2005: 776)

Indeed, we could argue that the quality of media conversations is changing: instead of multiple, discontinuous, heterogeneous and unsystematic conversations, we now have single, continuous, homogeneous, nearly real­time updates of news, stories, lives, events and activities, all streamed through a common format that is distributed in real time around the world. But it is important to be attentive to the cadences of streams as clearly the pace of flow change dynamically over time in response to external events and activities. This, I think, helps us to think about the way in which a seemingly limited platform of data transmission has become a mass social media and in doing so is preparing/teaching users to cope with real-time streams of information, a key requirement for the kinds of services and technologies that are currently being developed in Silicon Valley. Thus, perhaps we should think of Twitter less as a radio, or as a television, but rather as a ticker, albeit one that contains information other than stock prices (although it does that too). Twitter as a ticker also reinforces its sense of temporality as a constant set of discrete ‘ticks' that move dynamically around the world, and are connected to the activities and tweets of the users that use the service.¹⁰ We might also note that the interactional layer of Twitter, as commodity surface, remains in the form of a flowing stream of information, which, while it may possibly move fast in and out of itself, hides the underlying mechanisms, which are constantly shifting at the levels of codal, logical and physical layers.