“Yes, I clicked on this item, and it did not displease me.”

I do not like to use Facebook. There it is, upfront. I will occasionally pull up Facebook and here is how such interactions proceed: Glance at newsfeed; think that “in order to be useful I need to strip my Facebook of all but, like, three people;” and then I close the tab. Michael Cobb’s piece in the recent PMLA, “A Little Like Reading,” makes an argument that, despite my reservations of Facebook generally, I would agree with.  I would argue, however, that his examples need some serious context (and serious context), that they receive! The thing that first strikes me about Cobb’s piece, which argues for the potential virtue of Facebook-style reading of numerous tiny clips of information, is how informed our man Cobb, really is.  Bringing to bear on his discussion of Facebook well-informed discussions of Susan Sontag, Theodore Adorno, and Walter Benjamin on Baudelaire, at the very least (after all, the man’s staff photograph is with Snoop Dogg; he is clearly well-networked).

While the stakes of Cobb’s argument are not such that one can easily disagree with his conclusions, I do wonder what are the implications of his process: that to do something interesting with the “light, like” reading Cobb must necessarily contextualize these likings in a network of theoreticians.  Obviously, Cobb does not disavow this practice, and I in no way mean to “call him out” for this seeming conflict.  Cobb admits and advocates for this kind of contextualization, “Rarely can we not interpret a thing, encounter any work of art without a whole bunch of knowledge, a whole bunch of desire, and all of those mental categories that make art into something useful” (205).

Playing in parallel here is what he is considering in the debate between the “good” and “bad” kinds of negativity.  The bad kind, presumably, is the “endless nothings,” and the “positive” kind—the difference is still somewhat unclear and perhaps someone can illuminate this in the comments if my stab seems too in the dark—which seems the same except it is productively set against “the mind.”  Here he quotes Adorno quoting Hegel: “The life of the mind only attains its truth when discovering itself in absolute desolation” (206).  The virtue of nothingness, then, is that it allows a freedom of the mind, of mental associative activity, which is hampered by a too rigorously structured something.  Here, again is Cobb on the productivity of the “nought” of Facebook:  “You stare into something that can feel like a hopeless oblivion, and then it can stroke divine thought, or a peculiar combination, which will not yield sovereignty in meaning, or precise ordering of time, or even a good essay” (206). As I mentioned, the stakes are, in a way, quite low. He continues that the myriad feeds of information can “intrigue you as much as…the best and precise narrations in that book you wrote about religious hate speech eight years ago” (206). Finding the mind, intrigue generally: it is hard to disagree that these things could emerge from looking at Facebook.  However, Cobb seems necessarily to have couched this gentle, ephemeral insight in a heavily contextualized essay.  What his argument ends up looking like is that one can be sparked by interesting patterns in media like Facebook, but in order to articulate said sparks, one still needs deep immersion in other “deep” discourses.  The mind can only be discovered in the “nought” if one has already synthesized a great deal of structured something. Again, it is not a contradiction of any kind; it is simply the one-layer-removed from Cobb’s argument that I find complicates the essay in useful ways.

Cobb gives no specific examples of how this works with an example actually from Facebook, perhaps, no snark intended, because the examples are simply too ephemeral. The question is, then, if this is simply “the mind finding itself” as Adorno and Hegel and Cobb want to suggest, is what Facebook reveals only the structure of the mind already examining it?  Perhaps, then, this explains why an article about reading Facebook is really an article about interpreting Melville, Carson and Dickinson through certain well-regarded theorists and vice-versa.

Advertisements

Manovich’s Human-Scale Data

While Lev Manovich will reject a number of definitions in his reformation of “media visualization,” the project reaches a newly organic stage in the final examples of his “Visualizing Vertov” piece.  The relationship between eye and interface, and the shifts in this relationship, undergird each of “Visualizing Vertov,” “Media Visualization” and “How to Compare One Million Images?”  The focus, however, will remain visualizing data in a way to make it, once again, human scale.  Something most human, I argue, occurs in the final examples of “Visualizing Vertov.”

In “One Million Images” Manovich focuses on the eye and interface but with a more sever critique of the eye.  Not only is the eye not scalable to different data collections, but “[t]he second problem with using our eyes is that we are not very good at registering subtle differences between images.”  Wisely, I think, he shifts to a complaint of the interface more singularly in “Media Visualization:” “Although it may appear that the reasons for this are the limitations of human vision and human information processing, I think that it is actually the fault of current interface designs…This access method does not allow us to understand the ‘shape’ of overall collection and notice interesting patters” [sic].  Here, we can again notice interesting patterns!  Phew. The trick is to change the scale of the project to one at which we can use our apperceive apparatus to notice “shape.”

However, the idea of media visualization does not, I would argue, come to fruition until the last couple of examples in his “Visualizing Vertov” piece: “Anatomy of a Shot” and “Visualizing Movement.” Notably, figures 8.2 and 9.4.  These images represent shots “averaged” into a single image.

MANOVICH_8.2

FIG 8.2

MANOVICH_9.4

FIG 9.4

The process works differently here than in the graphs and “montages” of images. Shots averaged together most completely fulfill the definition of “media visualization” in which  “pictures are translated into pictures” but also, pictures, as well as they can, do not hide the original images of the media under consideration.  Manovich makes a special note of this move away from this traditional graphs: “Typical information visualization involves first translating the world into numbers and then visualizing relations between these numbers.  In contrast, media visualization involves translating a set of images into a new images which can reveal patterns in the set. In short, pictures are translated into pictures.” The result, at first and for a good while, is “images in a collection superimposed on a graph” (5).  Such graphs offer little more information that the graph itself would, except perhaps to understand the axes more intuitively.

What seems, however, the final result of this project, most nuanced interpretation, and most engaging image built from images, are these composite images of shots. They do not, admittedly, have the potential to convey raw-ish information like the superimposed-upon graphs, but differently do more with the visual data.  Namely, they seem to have the potential, of any of these visualizations, to tell us something we did not know, and to beg further questions, rather than, as Manovich often freely admits, verify claims. The image apprehends something itself of the shot beyond the algorithm used to generate it.

These approaches have problems, too.  For example, with projects 8 and 9 in “Visualizing Vertov,” Manovich tacitly acknowledges the inability of such projects to scale (the problem in the other two pieces).  In fact, the projects seem only to reveal movement at a single scale.  If an object within a shot moves too quickly, the movement will be erased from the composite image.  However, if an object moves too slowly, the object would appear stationary in the composite shot.  In this way, an object’s ability to be perceived has a direct relationship with the “scale” of the data set, the scale here being the length of the shot. I note these particular examples, however, because these exact complications generated by the composite images offer the greatest potential for a future project.

The bag of nails that you already own

A professor of mine tells me that the trick to academic writing is translating associative connections into logical, causal connections–a fact of the academy that I find showmanshipy yet absolutely true.  With topic modeling, the relationship between scholar, association and logic strains yet further under the computing weight of algorithmically generated association.

Ben Schmidt, in his “When you have a MALLET, everything looks like a nail” post, notes one of the methodological “saves” of working with two-diminutional graphical data plotted on a familiar plane rather than language (bags of words): one can intuitively recognize error.  He identifies the whaling map in which the LDA algorithm grouped together the eastern seaboard shipping and some pacific whaling into a single “topic.” Schmidt writes,

 This is a case where I’m really being saved by the restrictive feature space of data. If I were interpreting these MALLET results as text, I might notice it, for example, but start to tell a just-so story about how transatlantic shipping and Pacific whaling really are connected. (Which they are; but so is everything else.) The absurdity of doing that with geographic data like this is pretty clear; but interpretive leaps are extraordinarily easy to make with texts.

The question becomes, what is the threshold for a reasonable connection.  Indeed, Schmidt’s interpretation seems particularly not literary.  It seems to me that the “just-so” story about the connection between these two seemingly unrelated patterns would be not only what an academic of literature would accidentally expand on, but would be precisely the bit of information that he or she would be most likely to expand on, turn into a conference presentation, and tote about the conference circuit as a lively report on some unexpected associations (hence, I suppose, Schmidt’s warning).

Ryan Heuser and Long Le-Khac’s “Learning to read data” offers a counter-balance to the impulse to avoid spurious associations (or associations above the spuriosity threshold, which, as Schmidt implies, we must place somewhere).  The problem at the other end of the spectrum is throwing away data that does not already confirm what we believe, that is, eliminating data that does not support the conceptual associations and categories that we have already built.

  A troubling corollary to this is a tendency to throw away data that does not fit our established concepts. When Cohen discards a striking correlation between “belief,” “atheism,” and “Aristotle” as an accident of the data, he does just this. Whether or not the correlation is accidental should be decided by statistical analysis rather than the feeling that it doesn’t make sense. If we required all data to make sense—that is, fit our established concepts—quantitative methods would never produce new knowledge. If the digital humanities are to be more than simply an efficient tool for confirming what we already know, then we need to check this tendency to seek validation.

It seems as though Schmidt may be on the verge of doing just this—or, at least, encouraging literary people that thrive on association to do this—throwing away data that does not fit into a pre-established topic.   What is the happy mean here?  Heuser and Le-Khac advocate for doing some follow-up statistical modeling to check out the validity of these inchoate associations (when it rains algorithms…).  This is, however, where a more traditional literary scholarship could also take over.  Perhaps after getting a whiff of some new associative logic, it is time to set off into one’s text(s) and attempt a demonstration on the grounds of compelling and satisfying persuasive writing. Or do we want to see our field move further forward than this?

Reiterative Criticism

What if we treat algorithms as primary theoretical texts?  This seems to be, beyond suggestions rehashed from other theorists, one of Setphen Ramsay’s primary contributions to the matrices of questions that constitute the Digital Humanities in Reading Machines.

 In order to write the program, the critic must consider the “how to” of a deformative operation, but once that text is written, the output will be one that places the same critic into a critical relationship not only with the text of the result but with the text of the program as well. (66)

Aside from his work with the Oulipo and the Fibonacci sequence, however, Ramsey provides few examples of how this interaction may function.  In looking at the code he does make an important distinction between the “procedural,” and the “functional.”  While the functional describes a relationship, the procedural, that is, code, can “perform the relationships it describes” (65).

My question, if one would want to use code, procedures, as interpretive “texts,” where does this kind of text stand in relation to other forms of critical-text-as-primary-text?  To respond, I look to the similarity of Derrida’s Of Grammatology to Ramsay’s description of code as a primary text.  Here, Derrida describes Rousseau’s hiding and showing himself through writing:

 Let us note that the economy is perhaps indicated in the following: the operation that substitutes writing for speech also replaces presence by value: to the I am or the I am present thus sacrificed, a what I am or a what I am worth is preferred. “If I were present, one would never know what I was worth.” I renounce my present life, my present and concrete existence in order to make myself known in the ideality of truth and value. A well-known schema. (142)

Ultimately, Derrida works toward his conclusion of irreconcilable meaning: “Difference produces what it forbids, makes possible the very thing that it makes impossible” (143). However, what interests me in the above passage is the way in which Derrida restates—neatly by the sentence—Rousseau’s logic a number of times and turns it over to make it seem a schema!*  Once he states it three times, after all, it seems quite old-hat.

In this way, Derrida codifies and generates a procedure and a formula from the intelligence he sees in Rousseau’s writing: procedural in that Derrida’s writing is performing the work that he describes (deconstructing) while, at once, generating for us a formula (a schema) that describes Rousseau’s logic. The formula, not as immediately apparent, one might say is the average amalgamation of the statements that Derrida puts forth, the abstract description of all three statements.

Indeed, what may upset traditional scholars about the computer-aided analysis is that, beyond word counting and data mining, the procedures are hidden.  Knowing the procedure, allowing one to put the code in open conversation with the critical work it performs, removes the threat: “For surely there is nothing in the procedure that makes any claim to truth value beyond what is already stipulated in the critical act” (80). However unclear one may find Derrida’s writing, there is his procedure, laid out and labeled.

*Not sure from what Spivak translates “schema,” nevertheless:

1.a. “Any one of certain forms or rules of the ‘productive imagination’…”

1.b. “An automatic or unconscious coding or organization of incoming physiological or psychological stimuli” (OED)

Works Cited:

Derrida, Jacques. Of Grammatology. Baltimore: Johns Hopkins UP, 1997.

Ramsay, Stephen. Reading Machines. Chicago: University of Illinois Press, 2011.

“schema, n.”. OED Online. December 2012. Oxford University Press. 12 February 2013

 

Semantic Tagging

If there is a controversy among the use of TEI, it seems to be the appropriate degree of “semantic” markup. David Golumbia argues that a minimal markup is ideal both for the longevity of archived texts and the utility of those texts:

No doubt, some researchers may wish to create highly rich, structured-data-based applications driven off of this archival language data, but it would be a grave error to try to incorporate such markup into the archive itself.  That would actually obscure the data itself, and the data is what we are trying to preserve in as ‘raw’ a form as possible. (Golumbia 117)

The minimally-marked texts function as a sort of raw data, true, but it seems even the most powerful tools have a limited functionality with raw data in this form. Voyant Tools gives one a taste of just what a software tool can do with such a semantically-undeclared, raw text: search and count words and then display those search results in a variety of more and less useful fashions.

While this potential is interesting (see the Radiolab story about the professor who, from looking at the declining variety of words in her texts, found that Agatha Christie had undiagnosed Alzheimer’s), the question arises: isn’t even this most basic level of encoding an interpretation?  I find that Paul Eggert and, later, James Cummings’ response to this compelling: TEI provides a “phenomenology, not an ontology” of the text (Cummings).  As though one could provide the ontology of the text? Text essence? Text Dasein?  Probably not, I believe, “book” Dasein. The problem seems to be one of expectation: semantic encoding leads intuitively to the expectation of ontological capture.  While considering metadata, then, perhaps there is a line to be drawn in what metadata is for purposes of display, and what metadata passes beyond visualization of the text.

Cummings presents an example that falls between the silent metadata tagging of XML texts and traditional critical scholarly work when he recommends a digital text that can have varying degrees of critical apparatus.  This is metadata but not hidden metadata and not metadata that manifests without one’s go-ahead (in Cummings example at least, it sounds as though one can choose which metadata is available at any given time). Here is Cummings’ description of the text many-editions-in-one digital text:

Electronic publications can, if suitably encoded and suitably supported by software, present the same text in many forms: as clear text, as diplomatic transcript of one witness or another, as critical reconstruction of an authorial text, with or without critical apparatus of variants, and with or without annotations aimed at the textual scholar, the historian, the literary scholar, the linguist, the graduate student, or the undergraduate. (Cummings)

This text, I imagine, does change the ontological status of the text, and this change manifests as phenomenological alteration.  Imagine, if one will, having a text of Moby Dick which could include either zero footnotes, or footnotes for the layman, footnotes for the undergrad, and footnotes for the grad student (would you like to read Moby Dick on easy, medium or hard?).  One would have a sense at any given point of incompleteness while reading the text, as if bypassing layers of associative, interpretive work available at only the toggle of a menu option.

Works Cited:

Cummings, James. “The Text Encoding Initiative and the Study of Literature.”  Companion to Digital Literary Studies.  Ed. Susan Schriebman, Ray Siemens. Oxford: Blackwell 2008. Web. January 30, 2013.

Golumbia, David.  The Cultural Logic of Computation. Cambridge: Harvard UP, 2009. Print.