Metadata as a Problem for Thinking

I should have said this before, but the word itself is a monstrosity. Whosoever would attach a Greek prefix to a Latin root should be driven out of the city, egads. But let's overlook this superficial fact, at least for the time being. We saw before that metadata is an engineering problem. But metadata is a problem for other reasons too; metadata is a problem for thinking.

The notion that metadata might be a problem for society emerged onto the world stage with the Snowden revelations, although people have been worried about such issues already for some time. The conversation then was about the collection of so-called metadata -- telephone call records, who called whom, and so on -- and the lawful or unlawful ends to which such data might be used by state and commercial actors.

Here I'm not so much interested in whether metadata is a problem for society, but rather how metadata relates to thought and whether metadata might be a problem for thinking.

Data has long been a problem for thought. Data comes from the world, after all, and it's unclear where the mind sits vis-à-vis the world. In a phenomenological sense, data are all the things given up to experience. Data is furnished through specific structures of cognition. Such structures are social as well as cultural. But they're also physiological, limited and conditioned by the human senses. There's no such thing as raw data, of course, but at the same time data remains fairly primitive in the phenomenological sense. Data isn't raw in the sense of unstructured, but it remains raw to the extent that it becomes the ingredients for thinking. Much of the history of philosophy is an attempt to account for this fundamental incompatibility, the incompatibility between thinking, on the one hand, and, on the other, a world thought. To date, no one has solved this problem, probably because it's not a problem to be solved, so much as an encounter to be experienced. "Metadata" thus implicates the mind-body relation itself, the fundamental dualism of human consciousness. If data has a meta, is it us?


While the philologists will certainly object, it seems only natural and right to connect the Greek "meta" with English terms like "media" and "middle." The dentals "d" and "t" frequently slip and evolve into one another. And as Liddell and Scott attest, "meta" in the genitive and dative means "between" or "among." In the accusative it means "into the middle of" and thus generates a kind of sequence or succession, as is the essence of the accusative more generally. For this last reason, "meta" in English has come to mean "after" or "beyond," or even something like the "external" or "outside," as in the expression "going meta." Although that risks going too far; the essence of "meta" is not "beyond" so much as "between" or "among." So if we define media as middle, as I want to do, then a cluster of terms begin to coalesce around meta/media/middle.

(Philosophers love to recount the old anecdote about how Aristotle's "Metaphysics" got its name. The volume was apparently filed on the shelf *after* Aristotle's Physics, and since the Greek "meta" means "after," the volume was christened "After the Physics," or, for short, "Metaphysics." Whether apocryphal or not, the anecdote has long served to boost the biases of secular rationalism, as if to suggest that metaphysics in western philosophy can be explained away as a mere mishap of librarianship. At the same time, the anecdote also misses the essential point: "meta" might mean "after" but, more importantly, it means "amidst.")

The "meta" in metadata is also a problem for thought in a less obvious way. As I just suggested, the "matter" terms are, in a sense, the "media" terms. The meta is, in a sense, the middle. And the middle has long been a problem for philosophy. The middle is often the thing that philosophy can't think, or won't think. Indeed feminists have long made this very point, that "in between" states are perennially marginalized in philosophy. Some of the foundational principles from logic are most to blame here. I'm thinking of the law of non-contradiction and the principle of the excluded middle, both of which try to do it do away with middles as such. If you wonder why philosophy has always had such a hard time with ambiguity, multiplicity, or superposition it's because the state of being-in-between has been excluded from the foundations of logic since the beginning, or at least since Aristotle. (Attempts have been made to remedy this through things like paraconsistent logic but the results are rather unappetizing.) Because of this, many have turned away from philosophy, toward fields better equipped to think the middle as such, whether in feminist theory or media studies or in some other field altogether.

Shifting away from "amidst," eventually "meta" came to mean an external frame or set of conditions that constitute things. And to think the "meta conditions" came to be synonymous with thinking as a whole. Under secular modernity, the question "what is x?" was answered by outlining the conditions of possibility that pertained to the generation of x. A thing was defined via the conditions of the thing; thing = conditions. Perhaps no better example of this paradigm may be found than in the work of Immanuel Kant, who laid out the general conditions for thinking anything whatsoever (what Kant termed the transcendental conditions). With Kant, thinking fully mediatizes. To think is to think the meta, to think the media.

Now the relation between "digital" and "metadata" comes into view. I've said before that the digital entails some sort of cut that both separates and defines. (English terms like slice and section, but also the German Schnitt, the French tranche all contain this dual sense, both hewing off but also forming a piece.) So the digital means the cut, the mark, or better yet the frame or border that defines a space. And if "metadata" means framing -- information in formation -- then metadata are the purest expression of the digital. Such data is the digital in a very literal sense, because digital means the introduction of a frame. To assign metadata is to "be" digital. In the end, that's all the digital is.

Or consider Kant again and his famous expression "7 + 5 = 12" from the first Critique. We can now say that the left side of the expression, "7 + 5," has more metadata than the right side, "12." It has more discrete structure; it displays an increased digitality. It also literally contains more information, since two integers plus an operator is larger than a single integer. Thus "12" is quite literally the synthetic term for Kant, since it integrates three analytical terms into a single one. (Whether all mathematics is synthetic, as Kant argued, remains to be seen.)

So meta means amidst (being in the middle), but it also means frame, as in the conditions of possibility framing any knowledge whatsoever. Is this a contradiction? And does it matter? I wrote before on the problem of meaning in AI. Meaning seems to be contained not so much in one side or the other, in the frame or in the middle, but in the friction and vacillation between the two. Humans do it well. Digital machines not so much. Again, the obvious culprit is that digital tech has, for 2500 years at least, separated the middle from the meta, favoring the latter over the former. And for this reason modern digital computing has more or less given up on meaning. The formalist turn in mathematics during the early 20th C already set the stage for modern computing by asserting that mathematical truths were uncoupled from anything "real." All that matters for formalists is whether symbols are manipulated correctly on a page. So by mid-century, few were scandalized when Claude Shannon put forth his famous definition of information in which meaning played absolutely no part.

Metadata isn't a problem for digital computation. As I've said, "digital" and "metadata" mean practically the same thing. Yet metadata remain a problem for thinking, since to think is to recognize the essential tension between media as middle and media as frame. To think is to recognize the middle, full stop.

(And if, by contrast, digital tech were to include the middle, it might perform better, but then again it would cease being digital.)