Cette semaine, j’ai pris le temps pour parcourir un livre que j’avais reçu à la rentrée, “Raw Data” is an Oxymoron, collection de textes édité par Lisa Gitelman en 2013, aux MIT Press. Une lecture critique avait été faite l’an passé sur les blogs impactofsocialsciences de la LSE et infofutures de la bibliothèque de NYU. J’avais alors eu envie de parcourir ce (petit) livre, et j’ai été ravi qu’on me l’envoie cet été. Je noterais aussi que j’ai été surpris de voir autant d’auteures femmes sur ce sujet, où d’ordinaire, on lit beaucoup le point de vue d’hommes. Et c’est passionnant. Je ne vais pas faire une lecture critique du livre, mais plutôt rappeler que mon blog est un carnet de recherche ouvert, où je garde mes notes de lectures, entre autres ici certains passages du livre que j’ai surligné lors de ma lecture. Comme le rappellent Lisa Gitelman et Virginia Jackson dans l’introduction, le titre du livre vient de Memory Practices in the Sciences de Geoffrey Bowker (qui conclue le livre, on en reparlera)
“Raw data” is both an oxymoron and a bad idea;
Dans l’introduction, on nous rappelle effectivement que les données sont construites, avec une discussion autour de l’objectivité et de la subjectivité,
Data are familiarly “collected,” “entered,” “compiled,” “stored,” “processed,” “mined,” and “interpreted.” Less obvious are the ways in which the final term in this sequence—interpretation—haunts its predecessors. At a certain level the collection and management of data may be said to presuppose interpretation. “Data [do] not just exist,” Lev Manovich explains, they have to be “generated.” Data need to be imagined as data to exist and function as such, and the imagination of data entails an interpretive base.
In English, “data” was first used in the seventeenth century. Yet it is not wrong to associate the emergence of the concept and that of modernity. The rise of the concept in the seventeenth and eighteenth centuries is tightly linked to the development of modern concepts of knowledge and argumentation.
So, what was data prior to the twentieth century? And how did it acquire its pre-analytical, pre-factual status? In this, etymology is a good starting point. The word “data” comes to English from Latin. It is the plural of the Latin word datum, which itself is the neuter past participle of the verb dare, to give. A “datum” in English, then, is something given in an argument, something taken for granted. This is in contrast to “fact,” which derives from the neuter past participle of the Latin verb facere, to do, whence we have the English word “fact,” for that which was done, occurred, or exists. The etymology of “data” also contrasts with that of “evidence,” from the Latin verb vid ̄re, to see. There are important distinctions here: facts are ontological, evidence is epistemological, data is rhetorical. A datum may also be a fact, just as a fact may be evidence. But, from its first vernacular formulation, the existence of a datum has been independent of any consideration of corresponding ontological truth. When a fact is proven false, it ceases to be a fact. False data is data nonetheless. In English, “data” is a fairly recent word, though not as recent as one might guess. The earliest use of the term discovered by the Oxford English Dictionary occurs in a 1646 theological tract that refers to “a heap of data.” It is notable that this first Oxford English Dictionary citation is to the plural, “data,” rather than the singular, “datum.” While “datum,” too, appeared in seventeenth-century English, its usage then, as now, was limited—so limited, that in contrast to the well-accepted usage of the plural form, some critics have doubted whether the Latin datum was ever naturalized to English at all.
On gardera toutefois en mémoire la petite phrase
However controversial they may have been, in seventeenth-century English, neither “data” nor “datum” was particularly common. In these early years, the term “data” was still employed, especially in the realm of mathematics, where it retained the technical sense that it has in Euclid, as quantities given in mathematical problems, as opposed to the quaesita, or quantities sought, and in the realm of theology, where it referred to scriptural truths—whether principles or facts—that were given by God and therefore not susceptible to questioning.
Plus précisément, il note que
In early modern Latin, as in classical Latin, data is everywhere. But data in Latin rarely translates to “data” in English.
Bref, on apprend plein de choses sur le mot data. Par contre, pour le terme raw, une des réflexions les plus intéressantes se trouve dans le dernier chapitre, de Geoffrey Bowker, qui rappel que raw peut être compris dans le sens cru (par opposition à cuit), tel que l’utilisait Lévi-Strauss,
The concept “raw data” can be aligned with Lévi-Strauss’s use of the term “raw.” When he wrote The Raw and the Cooked, he was describing a vast mythological set originating analytically in South America among the Bororo and spreading throughout the world.2 His argument was that a series of binaries characterized this set, many of which were a variant of what we would call the nature/society divide. The natural was the raw (honey) and the social was the cooked (ashes).
On notera d’ailleurs que cette idée de cuisson était présente dans la phrase introductive du livre (quand on la prend dans son intégralité)
“Raw data” is both an oxymoron and a bad idea; to the contrary, data should be cooked with care.
Database development has followed this vein.The early databases were hierarchical—you needed to go down a detailed line of authority each time you wanted to retrieve a datum. Then we had relational databases, where there was still central control but much more flexible access (the database system, like society at the time, was seen as a fixed structure). Today we have moved into a world of object-oriented and object-relational databases, in which each data object lives in a Tardean paradise—any structure can be evanescent providing we know the inputs or outputs of any object within it. So databases have recapitulated social and organizational developments. And many organizations changed in the 1990s and 2000s in an effort to become more “object-oriented”; forgetting that the first object-oriented language (Simula, a precursor to Smalltalk) attempted to model work practice. Along the way, we have conceived ourselves and the natural entities in terms of data and information. We have flattened both the social and the natural into a single world so that there are no human actors and natural entities but only agents (speaking computationally) or actants (speaking semiotically) that share precisely the same features. It makes no sense in the dataverse to speak of the raw and the natural or the cooked and the social: to get into it you already need to be defined as a particular kind of monad.
J’avoue avoir beaucoup aimé cette comparaison. Sinon, je retiens aussi le chapitre Data Bite Man: The Work of Sustaining a Long-Term Study de David Ribes et Steven Jackson qui rappelle que les données sont éphémères,
Data are ephemeral creatures that threaten to become corrupted, lost, or meaningless if not properly cared for.
Ce chapitre, sur la collecte de données, et les difficutées pratiques m’a fait pensé à un repas de conférence sur le climat (et la modélisation statistique de données climatiques), à Montréal, où j’ai eu le privilège de manger face à un collecteur de données, de troncs d’arbres en l’occurence, qui pouvait rouler des jours durant, vers le nord, avec sa tronçonneuse, pour aller collecter quelques points de nos échantillons. Bref, ce livre regorge de perles. Et pour conclure, je garderais de Rita Raley l’idée de “data double”, qu’elle trouve chez Kevin Haggerty et Richard Ericson,
Surveillance technologies do not monitor people qua individuals, but instead operate through processes of disassembling and reassembling. People are broken down into a series of discrete informational flows which are stabilized and captured according to pre-established classificatory criteria. They are then transported to centralized locations to be reassembled and combined in ways that serve institutional agendas. Cumulatively, such information constitutes our “data double,” our virtual/informational profiles that circulate in various computers and contexts of practical application
Et je retiendrais aussi un paragraphe dans son chapitre Dataveillance and Countervaillance
We are thus in the midst of what is exuberantly called a “Data Renaissance,” in which new marketing worlds await exploration and raw material—raw data—awaits extrapolation, circulation, and speculation. Data has been figured as a “gold mine” and as “the new oil of the Internet and the new currency of the digital world,” the engine driving our latest speculative bubble. (Around the time of the worldwide financial crash, venture capital began pouring into online tracking.) Data speculation means amassing data so as to produce patterns, as opposed to having an idea for which one needs to collect supporting data. Raw data is the material for informational patterns still to come, its value unknown or uncertain until it is converted into the currency of information. And a robust data exchange, with so-termed data handlers and data brokers, has emerged to perform precisely this work of speculation. An illustrative example is BlueKai, “a marketplace where buyers and sellers trade high-quality targeting data like stocks,” more specifically, an auction for the near-instant circulation of user intent data (keyword searches, price searching and product comparison, destination cities from travel sites, activity on loan calculators).