Programmed Visions: Software and Memory
Source: https://mitpress.mit.edu/9780262518512/programmed-visions/ ↗
Chun examines the paradox at the heart of software: it promises permanence through storage yet operates through constant execution, repetition, and decay.
She argues that the ideology of software — the belief that code is inherently reliable, logical, and transparent — obscures how it actually functions as a technology of memory that is always degrading, always requiring regeneration.
The book connects the history of programming to broader histories of eugenics, race, and governance, tracing how the concept of "programmability" migrated from genetics to computing.
Chun's reading of source code as a fetish object — visible yet not what actually runs on the machine — challenges the common assumption that open source equals transparency.
Dense and theoretically demanding, it rewards readers willing to sit with its discomfort about what it means to delegate memory and decision-making to machines.
Central argument
Chun argues that software operates under a founding ideological contradiction: it presents itself as stable, transparent, and logical, but actually functions as a technology of memory that perpetually degrades and must be constantly regenerated through execution. She extends this critique to open source, contending that source code functions as a fetish — visible and readable, yet not what actually runs on the machine — which undermines the equation of openness with transparency. The book further traces how the concept of programmability carries a genealogy rooted in eugenics and governance, meaning that the apparent neutrality of computational logic conceals deeply political histories of control.
Critique
Chun's theoretical apparatus — drawing heavily on psychoanalysis, critical race theory, and media archaeology — can strain under the weight of its own ambition, risking the conflation of historical analogy with causal argument. The claim that programmability conceptually migrated from genetics to computing is provocative, but the book is more suggestive than demonstrative on the mechanisms of that transfer, which may frustrate readers seeking historical rigor over critical resonance. There is also a tension in the argument itself: if source code is a fetish that obscures execution, Chun's own textual readings of code risk reproducing the very displacement she diagnoses.
Why it matters for product
The fetish critique of open source has direct implications for product leaders who treat API transparency, audit logs, or observable pipelines as proxies for genuine accountability — visibility of code or data is not the same as intelligibility of system behavior, and confusing the two produces false confidence in oversight. Chun's framing of software as perpetually decaying and requiring regeneration reframes technical debt not as an engineering failure but as a structural condition: product strategy built on assumptions of system stability will systematically underestimate the organizational cost of keeping existing behavior alive while delivering new capability. Her argument about delegating memory to machines should also give pause to CPOs scaling AI-assisted decision-making — the question is not whether the system is auditable in principle, but whether the organization retains the capacity to reconstruct what it has effectively outsourced.