There Is No Software
Source: https://monoskop.org/images/4/43/Kittler_Friedrich_1995_There_Is_No_Software.pdf ↗
In five pages, Kittler mounts a provocation that has shaped two decades of debate: "software" as a distinct category does not exist.
What we call software, he argues, is a marketing abstraction layered over voltage differences in silicon — a convenient fiction that obscures the material reality of hardware operations.
The essay traces how each layer of abstraction, from high-level languages to operating systems, progressively distances users from the machine while claiming to bring them closer to meaning.
Kittler draws on his broader media-archaeological method, insisting that scholars attend to the physical substrate rather than the symbolic surface.
The argument is deliberately extreme and not entirely fair, but its usefulness lies precisely in the discomfort it produces: it forces anyone who works with software to justify the category they take for granted.
Available as a free PDF.
Central argument
Kittler argues that 'software' does not exist as a genuine ontological category but is instead a marketing abstraction layered over the physical reality of voltage differences in silicon circuits. Each layer of abstraction — from operating systems to high-level programming languages — does not bring users closer to meaning but further from the material substrate that actually performs computation. The essay uses media-archaeological method to insist that what we call software is a convenient fiction that obscures rather than reveals the nature of digital machines.
Critique
Kittler's argument collapses a useful and arguably necessary distinction: that abstractions, even if they do not correspond to discrete physical entities, can have real causal and social force — code has legal standing, it fails in reproducible ways, it encodes decisions. By treating all abstraction as mystification, he risks making the same error in reverse, privileging hardware as the 'real' layer while ignoring that voltage differences are themselves described through theoretical models. The essay's deliberate extremism, acknowledged even by the curator, produces productive friction but forecloses analysis of how abstraction layers actually shape what is possible or impossible to build.
Why it matters for product
For a CPO, Kittler's provocation is a direct challenge to any roadmap conversation that treats features as self-evident objects: if 'software' is an abstraction, so is 'a search bar' or 'a notification system' — categories that shape team structures, OKRs, and ownership boundaries without anyone examining whether they map to anything coherent at the delivery level. The essay also sharpens thinking about technical debt: what organisations often call debt is precisely the accumulated cost of abstraction layers that obscure the substrate, making it harder to reason about system behavior when those layers break down. A CPO who has internalised this argument will push engineering partners harder on what the abstraction is actually hiding before committing to architectural bets.