Cybernetics poses questions to history and historiography. It is consequential then for a book on the history of cybernetics, like this one, to embrace the circuitous method of its object of research. Sentences are repeated. Experiments and pioneering ideas echo each other across the chapters, with long-distance short-circuits and micro-epiphanies for the reader’s joy. A thick connective tissue weaves together perceptual micro-circuits, structures of cognition and political governance, leaping from the synaptic level to territory and population. The leaping itself is confusing at times, and repetitive – feedback loops work differently in writing than in technological communication. The choice of method, more than the theoretical findings, are the expression of a profound reflection on the writing of history and on the relation between reason and rationality.
A reader may have the impression that discoveries and major ideas are distributed across the book so that their respective interpretations can co-constitute each other, apparently without the need of a grounding reference. The then future of cybernetics, which, according to Halpern, corresponds to obsessive real-time data visualization, acts as a pivot point to periodically cycle back to. The book criticizes the fetish for contemporary methods of data display as the fulfilment of a paradigm of control and its aspiration to finally close the gap between memory and interface. Halpern’s central argument is that this gap’s openness guarantees the production of novelty.1 (As this review will discuss more fully below, the question of openness is a fundamental one for this book.) The recurrent reference to a contemporary visualization frenzy retroactively informs each chapter with a familiar rhythm. Many history books walk the trajectory of a current issue backwards to the point at which we begin to historicize. This book, more than others, short-circuits a ‘current’ development with its precursors, cutting across the spirals of a loopy history, rather than following a vector. This first cybernetic method for history writing pairs up with a second method: data inundation. Key historical discoveries float on top of an arsenal of examples, instances and lengthy biographies of designers, psychologists and mathematicians.
Given these qualities, the text is sometimes mesmerizing for the number of reflections it inspires. This review will engage with two macro-topics that the reviewer considers significant for the book, and, in a broader sense, of the utmost importance: ‘historical reference’ and, very briefly at the end, ‘beauty’.
“I follow Deleuze, who asked in his cinema books not what is cinema but what is philosophy after cinema? My question is a derivative. I ask what is it to tell history under the conditions of digital media? The status of historicism is under duress” 2, writes Halpern in the introduction. Cybernetics probes historicism not simply due to its amplification of the non-linearity of histories, but mostly because it establishes a complex relation with ontology, points of reference and memory. Halpern’s research and writing style treat the first issue. The second one is theoretically rich, opening up pressing matters relevant to historical epistemology: the role of historical points of reference; their relations with the boundaries of a system in the passage from closeness to openness; the status of historical reference in post-cybernetic historical discourse (and away from historicism, clearly). Apart from the notion of reference itself, what could be at stake here is an attempt at re-thinking the rapport between recognition and novelty.3 Halpern’s argument for methodical openness plays out as an intense sequence of examples, and the way of the example leaves little room for theorizing. Two questions seem to be missing from the book: is there such a thing as excessive openness? What is the place of historical references and invariances in this discourse? This review does not find answers, but tries to reflect on the reasons why these concerns matter in the context of classical cybernetics. Halpern’s work, maybe less than theoretically satisfying in this regard, still constitutes an excellent companion for this reflection.
In the 1940s and 50s, Wiener was hoping to produce self-referential communication systems un-moored from ontology and taxonomy, with no external reference. For communication to be effective, according to Wiener, there has to be no time-lapse between memory storage, recovery and response and the system ought to be closed in on itself. Really, the aspiration was to absorb all memory faculties in an interactive interface that projects action into the future. Memory serves the purpose of executing functions.
In the constellation of concepts and experiments exposed in Beautiful Data, Weiner’s work is influenced, among others, by Henri Bergson’s notion of memory (to which Weiner dedicates part of the opening chapter of his Cybernetics 4 and Warren McColluch and Walter Pitts’ research on neural nets. The former inspired the idea of a memory without the baggage of the archive. The latter offered a glimpse into what that ‘memory’ would have to look like if technologically produced. Halpern problematizes the use of this model of memory within and for the purposes of a closed system of communication, but it is unclear what happens to memory in an open system.
Many of Bergson’s intuitions were dear to cybernetics. Memory is no longer a storage of the past; perception/action begins with memory in a complex that sees recollecting and recording as co-temporal; it is possible to produce a self-referential system that accesses reality thanks solely to the presence of memory, thought; “action and thought are co-constituted” 5.
Going back to Bergson’s Matter and Memory, we appreciate one of the major discrepancies between Bergson’s and Wiener’s models. While perception and matter are only quantitatively different from each other, for Bergson, memory is qualitatively different. Memory does not have an extensive form; it is the past co-temporary with the present and the difference in nature guarantees for a rich in-between. Referencing Deleuze, Halpern identifies a permanent gap at this point, in which (in Bergson’s processual philosophy) the interaction of recollected memory and perception supposedly produces novelty, the future. Wiener’s trouble with memory was temporal as well as spatial: the memory of computation is mostly storage. Closing the system equals closing the gap between memory and reaction. For this to occur, either the system has already been predicted in its entirety or the individuals participating to it have no memory, their past is indeterminate.
In the passages on neural nets, Halpern mentions, amongst others, two significant quotes. In 1943, McColluch and Pitts declared the incompleteness of knowledge proper to neural nets to pertain to us all. They used these words: “Thus our knowledge of the world, including ourselves, is incomplete as to space and indefinite as to time. This ignorance, implicit in all our brains, is the counterpart of the abstraction which renders our knowledge useful.” 6. Two years later, in 1948, McColluch characterizes neural nets as a “device which could perform the kind of functions which a brain must perform if it is to go wrong and have a psychosis.” 7. Psychosis here is meant in the clinical sense. As Halpern summarizes, the two scientists had already noted that:
Neural nets are determinate in terms of the future, [they are] indeterminate in terms of the past […] from within a net (or network) the boundary between perception and cognition, the separation between interiority and exteriority, and the organization of causal time are all indifferentiable.8
Thus, a neural net projects action and knowledge into the future, but, in these terms, is incapable of recognizing temporal and spatial boundaries.
In the absence of external references, it becomes exceedingly easy to manipulate an individual equipped with a limited amount of knowledge. Not knowing is, indeed, what makes knowledge useful. For this reason, Halpern forcefully argues for the preservation of the gap between archive and interface and warns against technologies of data display and real-time visualization. These compress the breathing space between memory and reaction, aiming at an efficient control of behaviours. Openness, the text argues, guarantees room for invention, creativity and, possibly, political action.
However, while we are talking about openness, we also reflect on the conditions of possibility of sealing the memory-interface gap. Contemporary systems of real-time visualization, that supposedly show all information needed to operate decisions, may be an attempt at controlling behaviours, by feeding back a reality that has already been stored. Yet, any visualizing system carries the possibilities for its own improvement, revealing the borders of a gap that is still open. ‘Real-time’ names a desire for immediacy of information transmission, rather than being an accurate description of a visualization process. The model of neural nets does not seem to be scalable to a human brain that, we can quite confidently say, is not (necessarily) psychotic. The autonomous abstractive capacities of human vision 9 do not elide the faculty to reflect. The first tests on neural nets were telling in many respects. For instance, they showed that the mathematical logic of Turing machines is inadequate, when it comes to formalizing a human brain. Second, and more interestingly, that the logic of computation produces a different type of ‘thought’.
Engaging with the aforementioned systems of display, one can differentiate them from systems of knowledge, not that systems of knowledge are by any means complete, but for the different value they place on not knowing. A system of knowledge values what is not known as indispensable for more thinking, rather than as ‘useful’. This type of engagement cannot not start from pivoting on an external element, perhaps some sort of reference. Reference captures something more precise than simple openness, which does not constitute a suitable solution.
An example from the book, that of Songdo (South Korea), can help here. Songdo is the setting of the book’s prologue and one of the largest private estates on earth. The developer managing this ultimate smart city has run into a number of decision-making complications.
Engineers openly confess to never speaking to developers or urban planners, and admit that the city could take any form (circles, spires, anything really the surface does not reflect the infrastructure). At the same time, the developers are being forced to admit that their standard strategies are self-destructive. Banking on real estate while selling bandwidth, it’s unclear what is actually more valuable or what is actually being purchased in such developments.10
The open gap between form and function could create space for imagination and alternative histories to come, says Halpern – perhaps too much imagination, the only parameter being optimizing an abstract market value. Songdo could be anything, including nothing, the emptiness of stored data that seems unmoored from reality. The incapacity to operate distinctions, starting from determining what is being sold, translates into the inability to take decisions and choose a form. Songdo is a flat, amorphous surface, says Halpern. Its lack of nudges and differences, a sign of vast potentialities, may be precisely the reason why it cannibalizes itself – the developer has been losing money since the project began.
Other examples from Beautiful Data show a similar pattern. Nicholas Negroponte’s MediaLab specialised in producing demos of projects with no precise end-point. A successful demo is one that leads to another demo by failing, in a chain of fantastic purpose-less inventions. This modus operandi prizes creativity and then swirls out of control, not simply losing touch with reality. The effect is similar to looking at a well-camouflaged animal. Impossible to locate, it is spatially indeterminate. Its external boundaries cannot be told apart from the surrounding environment and its internal components are individually indistinguishable – real estate and bandwidth are reversible, one stands for the other. There cannot be a set point, from which one begins to build or count, informing with directive energy a previously indeterminate shape. Our capacity to make distinctions, appreciate depth and follow a thought through evaporates in a continuum that is illegible and ultimately irrelevant. If we set indeterminacy as an end in itself, we might find ourselves with little political agency within specific contexts or, more simply, immobilized.
The title calls attention to the word, but the book does not include a definition of ‘beauty’. Halpern quotes a variety of aesthetic strategies for the display of messages. The paradigm of ‘interactivity’ seems to be the invariance, around which they all organize.
“How did data become “beautiful?” 11 is a driving question for the book. Therefore, there is a processual dimension to beauty alongside the interactive one. The more explicit passage on the aestheticization of information is in the prologue. Data is not naturally beautiful; it becomes so through crafting. Another term in the same semantic neighbourhood, elegance, shows a similar behaviour. Elegance has a number of definitions in computation, but, from the more classical to the more conceptual 12, they all imply intentional parameters and various sophisticated procedures.
The genealogy of visualization tools plays a part in the beautification process. Visualization tools seem to naturally come with aesthetic inheritances and, when deployed for conveying information, the distinction between aesthetics and epistemology becomes all the more problematic. This crafting goes hand in hand with the use we make of data, in fact, in many respects, the beautiful, in visualization, is what is useful 13. The aspiration of breaking down data into small and smaller details, to then recombine them in newer perspectives or snippets of movements, draws attention towards a scale-dependent definition of beauty. Because visualization is allegedly scale-free it can produce a fascination for the screening of the maximally small and the maximally large. We seem enchanted by the capacity to scale down and scan the smallest minutiae, and then go all the way up and admire the macro-tendencies of swaths of entries. Curiosity towards unusual dimensions is normal. These distant perspectives inspire a wish to know and find out and research. Yet, this passion for easily scaling movements signals a fetish for visualization itself.
Interaction, the sense of action implied in verbs rather than in adjectives, process, and more precisely, crafting with the aim of amplifying capacities are points of convergence for the notion of beauty in data display. Halpern seems to suggest that the capacity to make us ‘see’ characterizes what is beautiful, in the current regime of visualization. This preference for vision and clarity has the taste of enlightenment values of knowledge, but it mistakes rationality for reason. Visualizing something does not assure its visibility 14. Data becomes beautiful when it affords a ‘better’ vision or perspective that makes us forget the parameters according to which we establish the comparative ‘better’. This is also the point, in which the boundaries between epistemology and aesthetics begin to fail. Halpern’s work is receptive of this crossover and not uncritical. Looking at the current landscape of media theory, there is a tendency for attention to shift from ‘engineering’ and ‘programming’ to ‘craft’ and design’. It may be the by-product of this murkiness, risking a simple conflation. We can read Halpern’s book also as a history of design that asks, in an oblique way, what is the troubled relation between aesthetics and epistemology.
Bergson, Henri. Matter and Memory. Translated by Nancy Margaret Paula and W. Scott Palmer. New York: Zone Books, 1988.
Fuller, Matthew, ed. Software Studies, a Lexicon. Cambridge: MIT Press, 2008.
Halpern, Orit. Beautiful Data. Durham: Duke University Press, 2014.
McCulloch, Warren S. and Walter Pitts. “A Logical Calculus of the the Ideas Immanent in Nervous Activity.” Bulletin of Mathematical Biophysics 5 (1943): 115-133.
Wiener, Norbert. Cybernetics: or Control and Communication in the Animal and the Machine. Cambridge: MIT University Press, 1948.
- In this context, Gilles Deleuze is Halpern’s chief reference, with his philosophical challenge to the ‘image of thought’ in order tloopyhich we cansimply, immobiliubled relation between aesthetics and epistemologyemnt a histrory sonal dimensionao conceive pure novelty. Halpern reads Deleuze’s work on Bergson: novelty happens thanks to the movement of contraction and interaction between perception and memory. ↩
- Orit Halpern, Beautiful Data (Durham: Duke University Press, 2014), 20 ↩
- The question of recognition and novelty is taken from Deleuze (Difference and Repetition). Reflecting on the relation between recognition – drawing upon previous experience and knowledge – and thought – qua novelty experienced due to an encounter – is relevant in the context of history, and a cybernetic one at that, especially in light of the individual dimension that ‘novelty’ may assume in Deleuze. ↩
- Norbert Wiener, Cybernetics: or Control and Communication in the Animal and the Machine (Cambridge: MIT University Press, 1948) ↩
- Halpern, Beautiful Data, 55 ↩
- Warren S. McCulloch and Walter Pitts, “A Logical Calculus of the Ideas Immanent in Nervous Activity”, Bulletin of Mathematical Biophysics 5 (1943): 129-131 ↩
- As quoted in Halpern, Beautiful Data, 145 ↩
- Halpern, Beautiful Data, 156 ↩
- Halpern, Beautiful Data, 62-64 ↩
- Halpern, Beautiful Data, 32 ↩
- Halpern, Beautiful Data, 5 ↩
- Matthew Fuller, ed., Software Studies, a Lexicon (Cambridge: MIT Press, 2008), 88-91 ↩
- Halpern, Beautiful Data, 5 ↩
- Halpern, Beautiful Data, 38 ↩