The Power of Classification. Culture, Context, Command, Control, Communications, Computing
How do things relate to each other? What is essential to one thing in relation to another? How do subjective meaning and generalized or objectified attributions of sense interrelate? What does it mean for our sense of self and how we relate to each other? What is meaning, how does it develop? These questions have puzzled humankind for ages. In an era of digital information, of humanmachine interfaces and robotic data spiders, surfing in the vast space of electronic data all on their own, they become vastly relevant. Proteus, an early sea-god in Greek mythology, can foretell the future. But he answers only to someone who is capable of capturing him and changes his shape to avoid this. More than ever, in a time of information explosion, it is not enough that data is useful, it also needs to be findable and accessible. Even though it can be quite hard to define the difference between similar and dissimilar objects, it is significantly even more difficult for abstractions and ideas. However, finding information is not the opposite of losing it, but an active effort to recognize interconnections in systems of meaning. Searching is an act of imagination, an approximation of expected outcomes, where findings inscribe themselves into the future.
Dragons of Chaos and Social Fiction
Ancient cultures had concepts of an ocean of information and a deep-sea monster dwelling in its dark expanse. The Black Winged Night of the Dragon of Chaos, the “vast and dark void” of Tiamat, from whose dismembered body the cosmos emerged and the world was formed. These demonic creatures mirror anxieties associated with a space of chaotic and unstructured information, untouched by the logocentric rays of solar deities and the light of reason. As we are adrift on the seas of knowledge, navigation is at the root of modern sciences. Cybernos, the steersperson of the seafaring ancient Greeks, maneuvering the nautical routes with the help of bright stars in the sky, lent its name to the science of cybernetics. This discipline of control and feedback, first applied in the field of ballistic course-plotting, stood at the beginning of many present-day information and communication technologies. Trying to produce intelligent maps of the world, these maps often reveal more about their authors than the territory they describe. Classification, elemental in mapping conceptual spaces of knowledge, typically mistakes transient social fictions for real and physical unchangeable facts. It happens time and again, particularly in relation to race, gender, social institutions, and any other domains where there is a vested interest in the making of realities.
Self-fulfilling Voodoo Categories
Names give an advantage to those who know them, the ability to call forth, to evoke or even to command the powers related to a name. There has always been an intimate relation between knowledge and control, and it is a power to have the authority to name something. Problem solving involves a process of naming things and issues and framing the context to attend to them. To bring order into the classes of names and hierarchies of designations is not only a practical or formal scientific issue, but a religious one as well. Categorization is type of cognitive voodoo related to deep-rooted beliefs that the world is/was created by the use of language, by the spelling out of names, and consequently that the universe can be influenced by a correct use of name and order. And it can be – but not quite that simply. Luckily, creating shared worldviews and producing ideological conceptual fields is more complex. Conjurers of classification, trying to force their hubristic will onto the world, may underrate the forces at work in the minds of others. They become victims of wishful thinking regarding the level of agreement that can possibly be achieved. Fortunately, for a start, the world is not necessarily compliant to voodoo categorization. Rigid standardization is actually hard to come by in a world where research is an ongoing process with changing definitions and a constant drift of understanding. It is a meshed up reality of unexpected shifts in perspective and dynamic interrelations with ever varying trajectories of power. Reaching a consensus about a standard first requires an agreement, and that can’t possibly happen where conformity does not exist. Nonetheless, expert “scientific” classification can be used to advance an agenda, to create a reality that in itself forms an effective case for a particular interpretation of reality. Cataloging schemes are hardly the discovery of a true “natural order”, but authored, and the purpose is chosen, not given. Categorization in a field of knowledge doesn’t necessarily document a given reality, but produces knowledge in a particular interpretation of perception. Classification systems are notoriously off track, but evidently good for the game of self-fulfilling projections of ideological power.
All the Print that Fits, or Not
While the ordering and structuring of knowledge has always been central to the findability of past information, at least since the library of Alexandria, a sorting concept like alphabetic order by author is a much more recent fashion. Today’s dominant library classification system conserves the 19th century worldviews of one Mr. Dewey and his limited grasp of realities beyond a white Protestant US middle class. The inventor of the Decimal Classification System of books was fond of the metaphor of an army to restore order in a chaotic mob of information, to impose a hierarchical structure and to force ideas into military style organization charts. Melvyl Dewey was making bold assertions about the world when, in 1876, he threw all non-Christians into one single category, listed very last in all categories about religion. Designers of the Soviet library’s catalogue system produced similarly strong ideological statements about the world when they established the top category “Works of the classical authors of Marxism-Leninism”. This demonstrates the problem of mapping catalogue systems onto one another, or attempting to match classification schemes – they each portray a different universe. The US Library of Congress classification system had to put a “former” label in front of their Soviet Union category, and still ranks tiny countries like “Austria” or “Switzerland” on the same level of relevance as the continents of Africa and Asia. These funny distortion effects of reality and relevance are also rooted in the need to find physical objects, books or atlases on shelves. When the software of concept and classification meets with the hardware of the tangible, and the immaterial interacts with the physical, it can produce unexpected results. Compared with smart automated search and indexing technologies, the traditional categorization systems loose out in finding things in large digital resources. But obscured by obsolete habits and outdated strategies from earlier efforts to structure knowledge, attempts to categorize information and research resources in the electronic realm can be highly inadequate. Digital information needs no shelf and the question arises whether predefined categorizations are such a good idea after all. A main reason for Google’s success was that there is no virtual shelf, no awkward pre-constructed file system. But with shelf space, even if it distorts the space of knowledge in curious ways, at least it is easy to see if it’s full or empty.
Mentalist Catalogues and Fortune Cookies
Professional catalogue and categorization workers strive to avoid what is contextdependent and temporary at all costs, but always end up in the middle of it. Trying to establish law and order in the information sphere, some warriors of categorization seem oblivious to the nature of transient realities and the fuzzy inflections of meaning. Cataloguers’ interests and requirements necessarily dominate over the more objective need of navigating the complexity of the world. They breed cognitive management technologies blind to cultural and subjective ambiguity and the slipperiness of context-dependent statements. Ideas of an objective ordering of abstract space are based on a religious notion of immaculate purity. They feed on dangerous ideologies of cybernetic control that imagine the manifest world to be reducible to a single viewpoint. Language is a complex temporal and spatial dynamic of signs and representation in relation to signified objects. In general linguistics there are no positive terms, only differences. Those working on categorization and building the ontology of classification systems intended to provide stable continuation over time have to organize the world ahead. Categorizing things in advance means to forecast the future, which is the magical practice of oracles, clairvoyant seers or spiritist mediums. And it is exactly the traps of categorization that mentalists and cold readers exploit in their illusionist stage shows. Organizational schemes deteriorate with time and scale, and the cost of support for highly managed centralized large volumes soon becomes prohibitive. Even though demand-driven systems like Google are fortunate not have to use advance predictions and projections of what one needs to know, the massive scale of data remains a key technology challenge.
Modern Ghost Logic
For complex information systems, it is essential that machines not only respond and interact with humans, but with each other as well. What are they talking about behind our backs? Semantic computer networks run on ontologisms and first order logic reducing logical inference down to simple rules. Syllogisms are a form of logical argument described by Aristotle as “… certain things being stated, something other follows of necessity from their being so.” The classic example is: Humans are mortal. Greeks are human. Therefore, Greeks are mortal. This kind of Cartesian logic not only sounds stiff and technical, but based on absurd absolutes invariably leads to ridiculous conclusions. Clay Shirky gives the following example: If – Count Dracula is a Vampire + Count Dracula lives in Transylvania + Transylvania is a region of Romania + Vampires are not real – then the only logical conclusion from such a set of statements is that Romania isn’t real. Sometimes the move from the logical to the silly is closer than it appears. Computers are very well adjusted to syllogisms, but the world can’t be reduced to unambiguous statements that can be effortlessly recombined. At the dawn of the 20th century, Sherlock Holmes significantly propagated the suggestion that brilliantly smart people arrive at unavoidable conclusions by connecting antecedent facts: “When you have eliminated the impossible, whatever remains, however improbable, must be the truth.” Holmes’ inventor Arthur Conan Doyle not only popularized the value of deductive reasoning, but was also deeply engaged in fairy photography, conversations with ghosts and a range of other spooky entities. Doyle was an enthusiastic follower of late 19th century Spiritualism, a complex socio-cultural adaptation to the advancements in science and new technologies. Holmes is part of this response to a beginning modernity, where the irrational and the rational meet at the dawn of mass societies. The cocaine-driven pseudo-rationalism of logical deduction was a hysterical reaction to an ambiguous world filled with libidinal ectoplasm and an explosion of things from industrial-scale production machines. Painting pictures of a simplified world in a narrowed down logic is soothing and somewhat comforting. Unfortunately, the unnerving daily reality – from ancient times to high tech societies and urban angst – mostly involves incomplete, inconclusive or uncertain and highly context-dependent information. Machines are good at logical reasoning that works well in places like index tables, and this is where computer assisted automation is largely useful and effective. Humans handle information based on “feelings”, popular heuristics, intuition, paranoia, wild speculations, and peer pressure or group dynamics. People rely on imitation, tradition or repetition and many other methods, but rarely on syllogistic reasoning or deductive logic.
Do Ideas Dream of Electric Sheep?
In its philosophical origins, the term ontology means the study of entities and their relations in a systematic account of existence. This tradition, less concerned with what is than with what is possible, asks, “What exists?” The object is a purely speculative purpose, not to facilitate action but to advance understanding. Ontological implications of categorization are highly problematic and ideas of “natural” classification betray essentialism beneath an epistemological cover. Ontology, which is about making clear and explicit statements about entities in a particular domain, has variable definitions in itself. Computer sciences and the field of digital knowledge management/classification have now taken this word “ontology”, and have applied it to the problem of machine intelligible information and an explicit specification of a conceptualization. Their organizing of collections of entities, things or concepts into related groups and hierarchical trees is based on such categorization and classification. Philosophers like to accuse each other of category errors, a semantic or ontological error. As in “colorless green ideas sleep furiously”; properties are ascribed to things that supposedly could not have that property. Similarly, it is seen as a mistake to conceptualize the mind as an object of immaterial substance, because it appears meaningless for a dynamic set of dispositions and capacity. Unfortunately there is no agreement on how to actually identify category mistakes. Ontologisms call for domains with central and legitimate authority, a confined stable corpus with limited entities and clear edges suitable for formal categories. Strictly regulated realms of juridical systems are just such an example. They require participants to be highly trained cataloguers, along with high levels of expertise and user coordination and authoritative reference sources for decision making. Closed ontological domains of hierarchical nature are typical for religious systems. It is not only psychotic personalities, institutionalized or not, who try to bring order into the world with endless and bizarre lists or Byzantine systems of classification in private cosmologies. These are also the base for a critical psychological discipline of controlled paranoia, Cabbalist analytical methodologies, and other gymnastics of the mind. Furthermore, there are long traditions specializing in experimentally induced deliriums of interpretation and artificially produced individual delusions of reference. These include techniques for the cathartic shattering of categories through the paradox and the breakup of false identifications in the perplexities of Zen riddles.
Blind Taxonomies and Orders of the Imagination
Mnemonic devices and memory hooks do not work because they are objective. On the contrary, the ancient Ars Memoria applies narrative structuring of the imagination and a visual anchoring of information to the geometry of thought. Typologies and taxonomies do not make assertions that can be judged true or false, but rather they are tools for the organization and stabilizing of thoughts about a shared reality. Geekish dreams of an “Ontology of Everything” aside, most proponents of semantic webs do seem sufficiently aware that building a top-down ontology or taxonomy that works for everyone and describes everything is not an option. They assert that this research enables local communities of interest to create their own ontology – and is not pursued for the enforcement of a New World Order of authoritarian classification. The reification of typologies is not unusual, but building taxonomies in the naïve belief that they represent the hierarchical structure of reality can be considered as rather unenlightened. In the real world of vast domains of proliferating entities that overlap in multiple ways, unstable and without clear boundaries, ontological structuring does not work well in broad access for non-expert users. The desired level of consistency in a normative classification setting influences the balance between complexity, simplification and scale. Either there is broad agreement in a narrow band of users, or slight agreements in broad groups.
Stereotypes and the Exploitation of Subjectivity
The distributed use of tags in flat hierarchies enables a new heterogeneity with large amounts of specific intelligence that improves organizational value with scale and time. Triggered by users with similarities in classifying, collaborative tagging may disproportionately reinforce each other’s views, predisposition and foregone conclusions. Specialist blindness, enthusiastic favoritism, tribal fads, gangland attitudes or stereotype prejudices can develop a strong dynamic of skewing issues based on a questionable validity of judgment. But if tagging remains transparent, it allows preserving individual, conflicting or even heretical viewpoints without having to force them into the straitjacket of temporal mainstream opinions. It accommodates statistical distributions where infrequent events frequently make up a majority. The total volume of the long tail of events with low popularity can exceed that of those with high popularity, and Internet ventures have leveraged this for their business. This outreach to the obscure and far out, disconnecting the service model from the peak-idiocy blockbuster demand curve, has certainly made media programming somewhat more intelligent. It has also contributed to the commercial exploitation of cultural niche markets and marks the transition from the traditional disciplinarian modes of preconfigured categories towards the new societies of control; from educational indoctrination to the fluid mining of cognitive response and reaction flows in opinion poll perception management. The so called “Web 2.0 interfaces” enable the commodification of subjectivity where social networks are exploited and then licensed back to the user.
Digital Eyes and the Hidden Gods
“The perfect search engine would be like the mind of God,” Google co-founder Sergey Brin once said. Accordingly, digital search engines aim for maximum reach and maximum recall. In the beginning is the search term, but Google wants to process all the information in the world and “understand exactly what you mean”. Without users, there is no mind of God. Google has now become the mainstream oracle of choice, the waves of zeitgeist queries breaking on the rocks of solidified identities. A recent USC study by the “Center for the Digital Future” found that a majority believes that “most or all of the information produced by search engines is reliable and accurate”. Increasingly, people rely on online resources for their routine intake of daily news instead of traditional print newspaper. Clearly, the issue of ranking and the intrinsic ordering of information and the underlying measuring and organization system prove to be a factor in forming worldviews. Imagination, shaped by the information we consume, in turn predetermines what we are looking for. Google News is by now a classic example of online news aggregation. Google’s ranking logics and indexing methods result in exclusion, and their news service hardly qualifies for a pluralism of viewpoints. Search results based on skewed but hidden mechanics of classification lack in inclusiveness, fairness and scope of representation. Their ranking practices are a trade secret, and when alternative information suddenly vanishes from top search results, there is simply no way of finding out why. It is a decentralized system, where all users become transparent to deep marketing data-mining practices and motivational research, but with an impenetrable center where at the core remains a hidden god. While a corporation may run its server systems on open source software, any political or economic influence remains hidden and the key providers of search and retrieval technology are completely unaccountable to questions of censorship and manipulation. In an age where access to information is largely controlled by a few companies, this is one of the most problematic aspects of the search engines.
Playgrounds for Spooks
Search, data mining and information retrieval technologies are in high demand by state or business intelligence agencies, and spooks are on the board of all major commercial operations. These technologies are indispensable in security operations, risk management and for Command, Control, Communications, Computing and Intelligence (C4I) systems. They can be used for humanitarian aims or rescue missions, but in an asymmetric digital war on terror, information sorting and retrieval can be turned into virtual or physical search and destroy operations. This is the playground for the Total Information Awareness officers and their panoptic Eye of Providence: subverting search engine capabilities for their all-seeing Eye’s massive surveillance of personal information flowing across the digital networks, breeding information paranoia and data panic to haunt the crossroads of search technologies. And the liberty to engage in social, cultural and intellectual activities free from oversight, in privacy and autonomy vanishes from sight. Data mining and retrieval applications are developed in vastly expensive software suites that are then beyond the reach of civil society organizations, independent researchers or critical initiatives. These powerful applications are not supported as tools for the public, but are used as weapons against this public by those who can afford it. Further development of semantic technology will enhance the uncanny ability to identify, understand and manipulate individuals without their knowledge or awareness. In the name of diversity, accessible and transparent applications are required, and the cryptology of open secrets must remain open source. A dynamic system of heterogeneous multiplicity including peer-to-peer exchange interfaces and open source search strategies, scalable personal information crawlers and anonymous engines are needed, as well as decentralized cluster architectures without central servers. Tools of cultural intelligence production should be in the hands of the many and not the privilege of the few. A truly free market of competing ideas requires access to the tools of computer aided analyzing and inferring, and a democratic diversity of making sense means the broad accessibility of automated information processes.
Augmented Cyborg Cognition
Many experts see advancements in information processing moving towards a stronger human-computer symbiosis in a range of fields that include bio-cybernetics and cognitive sciences. Human system integration is the buzzword for new human/machine interfaces in speed and depth enhanced information retrieval and decision making. Augmented Cognition wants to beat human cognitive limitations through adaptive computation. An adaptive user interface involves sensors for determining user state and emotions, and inference engines and classifiers to evaluate incoming sensor information. Computational systems continually adapt to their users and through sensing, learning, and inferences understand trends, patterns, and situations relevant to context and goals. Away from systems of linear or static text and the electronic typewriter towards advanced statistical text analytics, information mining and enhanced pattern recognition. Obviously the jet fighter pilot is currently the prototypical cyborg. But both pilots and the information workers on the ground have to filter relevant information from vast amounts of data in no time and act on the results. It appears only natural that DARPA is a leading player in a technology that shapes the future of warfare and information dominance. Since any sufficiently advanced technology is indistinguishable from magic, without a practical understanding of how it operates, this emerging technology remains a black art. Arcane sciences of high-powered and excessively funded labs set out to influence information landscaping.
Embedded Information Politics
Futuristic applications and computational complexity aside, technologies of the mind are political philosophy masked as neutral code. Innocent utilities that blend into the routine of everyday work and leisure, shading, blinding or subtly bending our perception in various ways, weave cognitive threads into the fabric of reality. Deliberately designed to yield results in a limited frame of reference, or naïve mechanisms of ideologies in specific domains, these cognitive tools are always political. Classification is not merely a retrieval tool, but also an embedded element of the ongoing construction of a work context and its associated dynamic processes and mechanisms. The logic of everyday language and political rhetoric typically evolve from a hierarchy of semantic objects, where its assumptions are presented as god-given and “natural”. However, in the daily reality of info overflow it is imperative to acknowledge both arbitrariness and willful design, and that hierarchies are not miraculously produced by nature itself. What is at stake is nothing less than the informational constitution of societies and their institutions. Throughout the heterogeneous fields of search research and the formation of applied sciences at the foundations of a democratic public, cultural intelligence is the thread to look for.