Data, data, data

So, as of late I've been working with numerous projects that in various ways deal with data. Data as object, data as subject, data as ontology, data epistemologies, and data as difference. Data, as a phenomenon, might seem self-evident, but is actually a rather mysterious one. I spend quite a lot of time trying to work with the ontology of data (I don't get invited to a lot of parties off the basis of this, I have to say…), and I'm increasingly thinking that data is the trauma of contemporary society – an unknown that we've created a joint silence/unquestioning around so as not to have to face the strangeness of it all. So, today, an essay I wrote for the ADD-project. This is the first draft, and I will be doing a rewrite for the version that will be published, but you get the early version here, long before publication. Ain't you lucky. 

Anyway, let's keep talking about data, and as we do, stay classy! Alf

On the Topography of Data Intensities

As much as our current age celebrates data, the vernacular and the vocabulary around it is sorely lacking. In fact, for something so arguably important, it is even maddeningly difficult to define. We speak of data as something absolute, an unproblematic category of conceptual matter that manages to simultaneously to be ever-present, to the point of over-saturation, yet also wholly without any qualities beyond its own. A locale or field that has been intensely mapped and mined for data will, to all except the functions of a data scientist (by any other name), appear much like one that is wholly untouched by datafication; no matter how exploitative data practices are, they leave their object untouched. Further, data often presents itself to us as something that exists only by way of representation, or, more precisely, only certain perceptions can be considered 'data' even when all perceptions can be made into such. The more one considers the category of data, the more it starts to resemble one of Leibniz monads: Basic, non-material, and without parts, who do not physically interact yet at the same time reflects the whole of the universe from its own internal principles.

Whereas data is typically spoken of as having relations with things such as information and knowledge, this is only partially true. An attempt to present a definition of data might result in something like this:

Data is a recorded difference that has been fixed in a way that can enable some level of permanence and distributability, and which thus can be utilized as input for some further operation.

It isn't the most elegant of definitions, but it goes some way towards capturing the complex materiality of data. It is more than something out there in the world, yet not something that necessarily weighs down the world. Furthermore, it also highlights the basic fact that the world is overflowing with recorded differences, yet not all have been fixed in a way that makes them processable. Data is thus both a trace, and more than a trace. It is, by necessity, a trace that is put into use, and through this extends into the world. Consider the rings on a tree. These are, as most remember from childhood, a record of the manner a tree has grown, and can be counted. Each ring is a recorded difference, fixed within the tree in a permanent manner. Yet, the tree can be cut down, allowing for the counting of these differences, and through this operationalization assess the age of the tree. Yet, when we see a dense forest, we do not consider this to be a thicket of data intensity.

It would seem that data intensities do not emerge simply out of the capacity to record differences, but the further operationalization of this. Early man could use a stone to make a trace in another stone, but left at this, the data intensity remains very low indeed. It took the establishment of numerous things, including the systematization of traces into a medium through which data could be distributed, for any real datafication to occur. This started out materially – cuneiform writing on clay, runic carving of stone, the transfer of ink onto parchment – and progressed towards more mechanical, industrial modes. The printing press, the telegraph cable, the gramophone, radio signals, the typewriter, the Xerox; as Friedrich Kittler has argued, the media is not just the message, it is the infrastructure of culture and society itself. Data intensities emerge out of this, the thicket not of a forest, but cables and pipes, processors and copiers.

Yet, this too seems much like an engineer's dream, too obsessed with the technological make-up of it all. One can, as my wife tells me, have an endless array of technological tools at one's disposal, yet capture very little. The mere existence of recorded differences, the mediums to store and share these, and technologies to use the end-result for novel ends does not explain why any of this would occur. Why have we built a world of greater and greater data densities, and what externalities does all this bring?

Intensities, in this context, are best understood through the lens of desire. Typically, when we think of desire, our minds drift towards notions of romance and erotics. However, the form of desire I refer to here is different: it is the desire for accumulation, profit, power, or knowledge, or some combination thereof. These diverse desires permeate societies, including democratic ones, and the reason we experience our current data intensities is rooted in a profound desire to amass data for various operations. This desire is not born solely from technological advancements, but from the possibilities these technologies enable. While technologies themselves are not entirely inert – Bruno Latour reminds us that technical actants exert their own kind of agency – the primary force driving data intensities is human desire and the organizational structures built around it.

In this regard, major tech companies like Google and Facebook function as engines of desire. They are fervently committed to gathering specific types of knowledge and data that can be leveraged in particular ways to generate profit or value. Yet desire always has a target; thus, when we talk about desire, we also acknowledge non-desire. We consider what is marginalized or ignored. Just as being enamored with a person can render someone myopic, making them feel and act as if no one else exists besides their beloved, the pursuit of data intensities creates a shadow world; a world that represents what is overlooked or pushed aside in the desirous process.

Significant scholarship has emerged over recent decades concerning the challenges associated with collecting data on human subjects. A prevalent critique within this body of work posits that, while there may be incidental benefits for the individuals whose data is collected, these benefits are overshadowed by the fact that the primary value derived from such data aggregation accrues to a limited number of corporations (often Big Tech-corporations such as Google and Facebook). This concentration of data and resultant power in the hands of a few entities is perceived as a substantial threat to democratic societies. The argument follows that increased data collection facilitates manipulation, enhances consumer targeting, and ultimately reduces individuals to mere commodities within the data economy.

While this perspective is valid and underscores the problematic nature of big tech's insatiable appetite for data, an often overlooked dimension is the inverse scenario; individuals or groups who remain relatively invisible to these data collection mechanisms. In a society heavily reliant on data, being less data-intensive can sometimes be a deliberate choice, signifying a form of autonomy or freedom from pervasive surveillance. However, more frequently, individuals become data-poor not by choice but due to systemic exclusion.

Several societal groups – such as the elderly, socially disadvantaged populations, and residents of rural areas – in fact do not struggle from having given away too much data, or being too data-intensive, but rather the opposite. To be considered not vital enough to tag and trace, funnel and filter, mine and map is today to be marginalized. To be ignored on the level of data is a way to become less visible, even invisible, within the data economy. This novel form of data inequality has not been adequately theorized, and I have suggested the term "data deserts" as a partial way to start highlighting the same.

My initial notion was that a data desert would be something akin to a food desert. In US sociology, one has pointed to the fact that disadvantaged urban areas can lack access to nutritious food due to the scarcity of grocery stores, compelling residents to rely on fast food, establishing a so-called 'food desert'. Similarly, data deserts represent areas or populations inadequately captured by data collection systems. Data deserts can manifest in various forms. For instance, certain urban ghettos might become invisible even to social services due to insufficient data capture. More common might be the case of how certain demographic groups like the elderly are not deemed significant enough to warrant comprehensive data collection, resulting in their marginalization within the digital landscape.

In contrast to those of us who appear as intensities, prominent peaks in the data topology, these marginalized groups represent flat areas on this metaphorical data world. The lack of adequate terminology to describe these disparities highlights an urgent need for further theoretical development. Just as in Edwin Abbott's parable of Flatland where three-dimensional beings possess inherent advantages over two-dimensional ones, individuals fully integrated into the data ecosystem wield power that those in data deserts cannot even comprehend. This emerging form of data inequality, an inequality of intensities, is today a major barrier for inclusive digital governance.

Yet, even talking about this issue is quite challenging, not least as we are trained to consider the world in ways quantifiable and without critiquing data. Instead we tend to fall back to data, simply suggesting that more data should be collected, without understanding that the contemporary topology of data is the way it is due to deep power structures in society. We lack data about Greenland not because we couldn't collect data about Greenland, but because there exists numerous structural features in our grand democratic society that prefers to keep Greenlandic data on a low-intensity level. We have little desire for the datafication of the elderly, as this might bring to bear responses that we are not prepared to deal with.

In 2013, MIT Press publishes the book "Raw Data" Is An Oxymoron,edited by Lisa Gitelman, which elegantly (if sometimes in a roundabout way) showed the politics, negotiation, and material conditions that was required to establish points and programs of data, and how all such was by necessity "cooked". Unfortunately, neither the book nor the concurrent consolidation of creative data studies as a field of study has had a broader impact on the social sciences; data is often still treated as being as raw as fine sashimi, if with a considerably longer sell-by date. As issues of data inequality will indubitably become more prevalent as our algorithmic society progresses and AI systems thirst for ever-bigger datasets, the issue of the aforementioned intensities, and the choices that underlie them, will almost certainly become more acute.

What is needed in this situation, more than the continuous promises from Big Tech to "do better", is to develop a more robust vocabulary around data and issues with data. Critical data studies has done a lot of legwork on this already, but is at current quite a small field and often overlooked by e.g. researchers into AI and AI society. It is as if data, these tiny little monads of ours, simply aren't considered important enough to critically engage with, at least not when there are algorithms and algorithmic systems to tussle with. Or maybe it is because people think they already know what data is, and thus simply bypass its manifold complexities out of intellectual arrogance. On a personal level, I can attest that I once thought I knew what data was, but the more I study it, the less certain I become.

In his book The Fold: Leibniz and the Baroque, Gille Deleuze praises Leibniz as a philosopher of folds, pleats, and curvatures, where things are continuously composed and recomposed in incessant inflection. He, in other words, finds a topological logic to Leibniz monadology; and thus a call to look where others wouldn't. In other words, an aim to look towards the various enfoldments and gatherings of the world, all with an eye to what escapes the fold, what remains outside. In this, it might be that what our world, and our data requires is less of the engineering logic that has defined much of data science – attend only to what is present, strive for efficiencies, and let the bits fall where they may – and instead more of a more baroque thinking, attuned to the folds of this world and the empty enclosures they create.