In this brief essay I wish to introduce the draft as a keyword for thinking about the designs of ethnography today, and specifically about the task of producing difficult descriptions. I am interested in how ethnography proceeds in and through drafts: how it is temporarily and tentatively inscribed in notebooks, sketches, documents, and illustrations, and how drafts are often drawn and written with others, circulating as invitations for commentary or modification at different times and in different spaces, outlining and accompanying the ongoingness of social process.

However, by drawing attention to the role that drafts play as technologies of description, I also hope to expand on our understanding of the nature of description for anthropological theory more broadly. The salience of drafts as epistemic objects today, I suggest, bespeaks important reconfigurations in venues and methods, exchange and experimentation, as well as the political economy of anthropological inquiry. The tentativeness and instability that drafts inscribe and inaugurate—their role as both semiotic and material operators—reorients the problem of description to operate not just in an epistemo-ontological register, but in an infrastructural one too. Let me explain.

Grappling with the challenges of difficult descriptions is, of course, a far from recent anthropological enterprise. In her book What Would Animals Say if We Asked the Right Questions?, Vinciane Despret (2016) builds on the work of Eduardo Viveiros de Castro and Marilyn Strathern to put forward figures of description that she calls versions. Versions do not aim for direct and correlative meaning but aspire to disturb the semantic and conceptual registers of meaning-making. To engage in the production of descriptions as versions is to engage and dwell in the production of equivocations (after Viveiros de Castro). Equivocation, for Despret (2016, 172), is an operation of meaning that adumbrates openings and cracks; it is an “operator of bifurcations” and a “creator of partial connections.”

Versions, equivocations, and partial connections invoke worlds-in-tension and worlds-within-details where the positional geometry of bodies, words, and affects yields anything but obvious or correlative translations—worlds that do not overlap, that do not add up. Instead, these are anthropological descriptors that value the promises of a certain imponderability and not-knowing, the value of unanticipated effects.

Drafts share with versions this trust in the uncertain and the promissory. They, too, are invitations, openings. However, the openings that drafts perform are not just descriptive. They are also inscriptive, for drafts are their own concurrent materiality. Drafts are at once ethnographic effects and ethnographic infrastructures. Moreover, they are infrastructures specifically designed to enable descriptions to remain redescribable.

What does it mean to say that a description must remain redescribable? Aren’t all descriptions redescribable in one form or another? The short answer is no. The labor of redescription is not just a literary or metaphorical skill but demands mobilizing specific legal, technical and organisational assemblages, which underwrite the presence or absence of what I have elsewhere called capacities for redescription (Corsín Jiménez 2015). Let me illustrate by way of a brief vignette from the history of software versioning.

Developed in 1972 by Marc J. Rochkind at Bell Laboratories in Murray Hills, New Jersey, the first Version Control System (VCS) enabled software developers to access historical versions of source code and project files (see Rochkind 1975). Over the years, VCSes afforded developers the concurrent and distributed storage, retrieval, logging, and identification of revisions. Working at different terminals (and, in time, in different parts of the world), developers could keep an eye on the itinerary of any project, tracing back its history and, if needed, restoring it to past states or, alternatively, forking new paths from any particular juncture. The history of VCSes is closely associated with the development of free/libre and open-source software (see Kelty 2008). It takes a particular assemblage of intellectual property rights, technical protocols, and collaborative dynamics for a version to remain versionable, for a description to remain redescribable. However, VCSes have played a key part in the development of the stack megastructures of platform capitalism too, as is the case with the archival registries of cloud-computing services such as Dropbox or Google Drive. There is no romanticism in VCSes, only descriptive possibilities.

The bifurcations operated by VCSes are therefore not unlike those described by Despret; yet, unlike the latter, they also make visible their infrastructural sources and requirements, their legal forms and obligations, and their material conditions of possibility. As versions they world effects; as drafts they inscribe the solicitousness of worlds.

Drafts, then, capture this double work of descriptions as both inscriptions and effects. They layer themselves as invitations to reckon with the solicitudes and exigencies, the languages and the venues, the registers and the techniques of difficult descriptions.

This inquiry into the nature of anthropological description seems a timely one today: we have been invited to consider arts of noticing, entanglements, ethnographic refusals, epistemic partnerships, ontological designs. Anthropology appears to be entering yet another time of awakening and discomfort, groping uneasily for complicities and complexities in worlds that are on the brink of collapse, if not already ruinous and decimated. How to align ourselves with others’ solicitations? What designs for the solicitousness of ethnography?

In this brief essay I have offered the draft as one possible design for our ethnographic moment. Drafts help us see how descriptions and ethnographies work as infrastructures. Drafts are unassuming infrastructures that accompany our journeys through translational spaces. Drafts draft a certain tremulousness, a certain hopefulness, a certain solicitation, at every turn making visible the sources and resources that remain available for redescription. However, there is no romance in the draft. Operative in planetary stacks and platforms, some drafts hold in duplicitous and ambiguous tension our capacities for redescription. Thus, while some drafts may be discarded and others may be brought to a close, it seems that it is always possible—necessary even—to point to the drafts through which worlds world and draft along.