“Empirical assessment of emotions in literary texts has sometimes relied on human annotation” but now an algorithmic “system determines which of the words exist in our emotion lexicon and calculates ratios such as the number of words associated with an emotion to the total number of emotion words in the text” for novels, Shakespeare, and Brothers Grimm fairy tales (pp. 2, 3). There are no references to Franco Moretti, literary criticism and critical theory, or precedents in fairy-tale analysis such as the ATU Index. Google’s n-gram corpus data and visualization software, which allows you to “Search lots of books” (as opposed to a specific number? are we simply drawing lots? from the “mind of God,” as a Google co-founder once famously called Google), is the Script/ure used (examples of Google data visualizations here).

NRC author Saif Mohammed sees a bright future for the technology in affect surveillance: “Ultimately, he thinks emotion-based software may help us better keep tabs on depression or online harassment and bullying, track emotions on Twitter to forecast riots and better deploy resources, or measure sentiment towards government policies and controversial issues.”

The post Government Info-Tech Research Officer “Track[s] Emotions in Novels and Fairy Tales” appeared first on &.

In view of our reading of Latour’s rearticulation of the assumptions informing the concepts of “facts” and “values” into new terms that recognize the processual cyclicity of institutionalizing always-debatable uncertainties into useful certainties, and in view of Alice Munro’s having just been awarded the Nobel Prize in Literature, an article entitled “Denied Nobel Prize Yet Again, Margaret Atwood Plots Post-Apocalyptic Revenge” seems timely. The article appears on the website Newslo, which boasts (with apropos unabashedness with respect to facticity) the title of “The First Ever Hybrid New/Satire Platform!” The title captures accurately, however, the Newslo site’s hallmark article-viewing functionality: a toggle for “Show Facts” and “Hide Facts” foots every article. Activating the former button highlights the portion of the article that has been institutionalized to the realm of what we would previously have called facticity; re-negating these highlights (back to the article’s default appearance) returns smoothly and non-libelously the factual portions into formal homogeneity with the sensationalist satirical tabloid remainder. “Just Enough News,” runs the site’s motto, which is just as well an ironic motto for news in general nowadays: there’s too much news, and yet never enough “(f)actual” news in that news (and never enough fiction in our “fiction“). How can we better define that “just enough” (with pun on “just”)? We need what Latour proposes: a better “power to take into account” (Facebook account or otherwise): we need to evolve our filter-feeders.

Indeed, Newslo’s filtering system typifies the related problem of facticity in regard to digital public spheres that we discussed both yesterday in relation to a Quebec provincial voting commercial, which we argued exemplified the complex relations governing the way in which the mob rule of social media gets billed as democracy via that social media itself, as well as last month in relation to our discussion of print newspapers versus databased newspapers and the personalized newsfeed. “Filter bubbles,” Ted Talker Eli Pariser calls it, this Internet censorship phenomenon in which our bias towards our own politics of interest can actually work against our own interests – especially when, as Pariser relates, Facebook one day decides to run an algorithm (just for “you,” i.e. for you) that converts all your visible news into whatever your Good News happens to be. While the talk is scarcely worth skimming beyond what I am graciously filtering for you, Pariser does end up proposing that databases like Google adopt what amounts to an extension of the logic of Newslo’s system: namely, including among our “sort by” options checkboxes for things like “Relevant”, “Important”, “Uncomfortable”, “Challenging”, and “Other Points of View”. The metaphysics of presence points to an obvious problem with this system, or rather with the utopian version of it (a database will never represent what is truly Other to it – we, and even Google, simply cannot know how “Other Points of View” is itself being filtered; and is a comfortable way to find the “uncomfortable” possible?), but nevertheless the gesture seems valuable, and maybe even represents a crude attempt at doing for the database what John Law proposes scholars do for research in general: making explicit to our readers the always-messy contingency of that research. Perhaps databases should have a “messy” filter (kind of like the trope of the button which you’re told never to press because no one knows what it does), which when you press it simply puts everything through a glitch-randomizer and fucks everything up.

It is such mess, though, that Latour likewise encourages as the one of two requirements of his proposed initial phase for institutionalization (what in the old language we would call fact-making), the phase of taking into account all of the entities that we have at our disposal to debate for institutionalization candidacy: namely, the requirement of perplexity, a “provocation (in the etymological sense of ‘production of voices’)” in which “the number of candidate entities must not be arbitrarily reduced in the interests of facility or convenience” (110). Its corollary is the requirement of relevance – considering the relevance of the various stakeholding voices to perplex what is being, by debate, perplexed. This latter somewhat relates to Pariser’s proposed (re)search filter category, but more to that mark is the requirement of hierarchization in Latour’s second of the two institutionalization phases, arranging in rank order. For, Pariser’s category of “Relevance” (and rather than simply pertaining to the already-familiar category of “sort by relevance” vis-a-vis search terms) comes in critical response to Mark Zuckerburg’s cited remark that “[a] squirrel dying in front of your house may be more relevant to your interests right now than people dying in Africa.” Pariser’s point is that, if newsfeeds or other search results are ever to even approach something more like the impossible ideal of information democracy to which popular media discourse so often articulates them, then the squirrel pictures that come up in our newsfeeds should perhaps not be placed at the same level of relevance as humanitarian crises in what Latour calls society’s “hierarchy of values” (107) (which also, notably, functions by a “[r]equirement of publicity” [111, emphasis added]).

The comparison to be made, then, is between the (in the old terms) “fact-evaluation” process that Latour proposes for constitutional democratic government, and the way in which information is similarly filtered to and filterable by the public of such democracies within their digital news and (re)search platforms. Latour’s proposed process begins with perplexity and relevance – in other words, the collection of search results (the terms – relevance – and their results – initially, perplexity) that then, in the second part of the process, the move toward actual instutionalization (“closure”), must get hierarchized accordingly in terms of their hierarchical ethical valuation. In sum, it would seem that, in extension to Newslo’s and Pariser’s unique filtration systems, and indeed as a corollary component to the type of public network resultant of precisely the kind of (public sector) debate apparatus that Latour proposes, our new digital filtration systems should involve not just parameters like “Show Facts” and “Relevant” but precisely such a process as Latour’s, with its unique re-articulation of “facts” and “values” into less contradictory categories. Then a system such as Newslo’s could only constitute absurdity insofar as “facts” would no longer be a category for filtration. But more importantly, because the relative contingencies of what we used to call “facts” (and get mixed up with “values”), and the (traces of the) process by which they became such, might be explicitly built into all our articles and newsfeeds and other search databases. That is, the equivalent of “Show facts”, for example, would show (by a kind of markup, perhaps) what had been successfully put through Latour’s process of institutionalization.

Otherwise, Margaret Atwood might still be in the running, according to her own personalized fantasy newsfeed. For the division between “facts” and “values” (where the latter equals the explicitly skewed satire in Newslo’s articles) is never so clear as Newslo makes it. Hence the need and use for Latour’s proposed new articulations.

Works Cited

Latour, Bruno. “A New Separation of Powers.” Politics of Nature: How to bring the Sciences into Democracy. Cambridge: Harvard University Press, 2004. 91-127. Print

Law, John. “Making a Mess with Method.” Lancaster: The Centre for Science Studies, Lancaster University, 2003.
http://www.comp.lancs.ac.uk/sociology/papers/Law-Making-a-Mess-with-Method.pdf

The post Show Facts, Hide Facts: Applying Latour to Database Structures appeared first on &.