Arctic Archives: The slough politics of GitHub’s Code Vault

READING CIRCLE.

In 2020, GitHub printed 21 terabytes of data scraped from GitHub users’ repositories onto a series of film reels to store in a decommissioned mine shaft in the arctic. The “Code Vault” calls itself an archiving project that aims to “protect the world’s software from apocalypse,” and preserve a “modern way of life.” Yet, as Katherine Clare Mackinnon’s work shows, the underlying logics of the project stem less from archival theory than from architectural thinking. At the center is GitHub’s “Pace Layer Strategy” for deep-time preservation, inspired by Stewart Brand’s How Buildings Learn (1994), which in turn draws on earlier concepts of “shearing layers” from Christopher Alexander and Frank Duffy. Revisiting these texts, Mackinnon traces a genealogy of conceptual translations – from theories of change and durability in buildings, through object-oriented software design, to contemporary infrastructures of digital preservation.

This session takes Mackinnon’s article as its point of departure and brings it into conversation with excerpts from Brand and Alexander. Together, we will explore how architectural models of materiality, temporality, and change are reconfigured in digital contexts, and what is at stake in this shift. What happens to concepts such as maintenance, durability, and care when they are translated into software infrastructures? And how might something like GitHub’s “pace layer strategy” not only describe preservation strategies, but actively produce hierarchies of obsolescence - anticipating and naturalizing processes of loss over time?

The session is conceived as an interdisciplinary discussion across architecture, digital culture, and media studies, and all participants are warmly invited to engage with the texts in advance.

Katie MacKinnon is a Postdoctoral Fellow on Data loss (DALOSS): the politics of disappearance, destruction and dispossession in digital societies at the University of Copenhagen (PI: Nanna Bonde Thylstrup).

She researches historical, political, and ethical implications of long-term data storage across datasets, models, archives and in the production of internet histories and futures.

Please sign up at intersect@ku.dk and we will send you the texts.