The End of Raikin’s Atelier

In Arkady Raikin’s 1970s sketch “Who Made the Suit?”, the satire on Soviet bureaucracy reaches its clearest expression. A customer appears in a grotesquely ill-fitting jacket and confronts an atelier staffed by a hundred specialists, each responsible only for a tiny fragment: one for buttons, another for sleeves, a third for pockets. The line “No complaints about the buttons – they’re sewn on for life!” became emblematic of diffusion of responsibility. Raikin illustrated the classic paradox of Adam Smith’s division of labour: greater specialisation increases operational efficiency but destroys the integrity of the final product. Process prevails over outcome, and collective irresponsibility is disguised as formal correctness.
Half a century later, the metaphor remains remarkably precise in describing digital reality in 2026. Raikin’s atelier has simply migrated from Gosplan workshops to the data centres of Big Tech. The clearest illustration is Windows File Explorer in versions 10 and 11. Designers in the Fluent Design team created 8K icons with flawless animation (the button department), the core team updated context menus (the pocket department), and AI code generators “stitched” the microservices. Everything is formally perfect. Yet the result is an interface in which file copying lags, search insists on querying Bing instead of the local drive, and menus demand extra clicks. The system’s response? “Any complaint about this specific button? No. That the suit (the OS) is unwearable is your problem – we work in sprints.”
Here Raikin encounters contemporary management theory. Narrow specialisation has produced silo mentality: departments function as isolated towers with atrophied horizontal connections. In Microsoft, as in the Soviet atelier, there is no single “master” who sees the entire garment. Bureaucracy, however, is only the visible symptom. The deeper cause lies elsewhere.
The real conflict is a dynamic temporal mismatch: between the accelerated pace of technological change (the exponential tempo of NTPR – scientific-technical progress and development), continued deepening of the division of labour, and ever-greater component diversification. Suppliers of “thread” (Intel/AMD chips, Google frameworks, AWS cloud services) alter the technological map every quarter, sometimes every month. Yesterday’s cotton thread – stable legacy code, proven APIs, old drivers – is declared obsolete. It is replaced by a plastic micro-cable of the new generation (new processor instructions, TPM 2.0, neural optimisations, cloud dependencies).
To preserve the original buttons and original thread – that is, to maintain compatibility with 2005 software, accounting packages and corporate workflows – the atelier (the developer) must pay more and more. Not once, but every day. Preserving the existing technological map has become a permanent, exponentially growing tax on slowing progress. The supplier has already shifted main capacity to micro-cable production, where economies of scale, automation and new contracts apply. The old production line remains on outdated equipment, staffed by scarce specialists approaching retirement, with logistics that grow costlier each year. To persuade the supplier to continue manufacturing “yesterday’s thread”, Microsoft, Google and other giants pay two, three, then five times more. This is no longer classical vendor lock-in. It is a dynamic tax on braking NTPR: the faster the supplier introduces the new, the more expensive it becomes to halt or even slow the process in order to maintain compatibility.
Add the planned saturation of the market with narrow specialists. Universities, bootcamps and corporate programmes mass-produce thousands of “button-sewers” and “pocket-makers” – operators of isolated functions. This eliminates competition as a natural selection mechanism. With a surplus of cheap, interchangeable labour, the market no longer weeds out ill-fitting suits. They are reproduced automatically, by inertia, without real pressure for quality.
The deeper the division of labour and the wider the diversification, the more severe the mismatch. Threads evolve according to their own logic (becoming stronger, smarter and cheaper at scale); buttons evolve according to theirs (more elegant, smaller, with 8K textures). Integration between them demands ever-greater resources. The outcome is classic over-engineering: a system in which individual components are perfected while the whole is distorted and expensive. Windows File Explorer is a living example: it swells not because programmers are lazy, but because every day payment is required to keep the old threads alive in a world where the supplier now works exclusively in micro-cable.
This temporal-speed conflict explains why the conveyor-belt model of mass consumption has finally reached its limit. Earlier, shortage drove customers into the atelier; today, hyper-personalisation demands individual fit. Yet the conveyor is inflexible: millions of lines of legacy code, thousands of narrow specialists across time zones, supply chains tied to yesterday’s specifications. Agile, originally conceived as an antidote to bureaucracy (cross-functional teams, short iterations), has degenerated into reactive crisis management: daily recutting of patterns to accommodate new materials, with no underlying architectural skeleton. Sprints become patches on patches. Coordination entropy (Ronald Coase’s transaction costs) consumes the entire margin.
Marx would have recognised the culmination of labour alienation. The programmer is no longer a master but an operator of a function, fully alienated from the product. AI intensifies the process: it generates code with 98 % validity, a junior reviews it, and the lead reports against KPIs. Expertise erosion is irreversible – carriers of “sacred knowledge” about the old code are disappearing. The system operates at peak local efficiency yet produces systemic waste. Windows is not unique. The same pattern appears in vehicles with fifty ECU modules, in banking microservices, and in smartphones with twenty abstraction layers.
The era of the universal product “for the average consumer” is over. The conveyor belt – Adam Smith’s triumph – has collided with its own complexity limits and with the dynamic tax on braking NTPR. What replaces it?
The return of the master 2.0 – now at a fundamentally new technological level. This opens the door to personalised production. Every large consumer (corporation, affluent individual, professional team) acquires its own “personal computer tailor” – a trusted agent accountable for the entire garment. This is technological neo-feudalism: a stratified model in which the elite have a personal code concierge while the masses continue to wear the ill-fitting universal suits.
Open source forms the platform for this transition. The bare kernel (Linux, as in NixOS or Guix) remains the universal pattern, while meta-layers of implementation (custom overlays, declarative configurations) become the personalised overlay. Custom Linux builds tailored to specific hardware and use cases already demonstrate the model: gaming PCs optimised for particular graphics cards and overclocking, multimedia workstations for professional editing, scientific stations with reproducible builds and domain-specific libraries. A plain Ubuntu assembly no longer interests anyone – it belongs to the mass-production era of the past. A single master (or micro-team), equipped with AI-agent apprentices, controls the entire cycle from bare metal to final fitting. The master negotiates directly with micro-cable suppliers, creating an isolated, stable ecosystem matched precisely to the user’s tasks. Technology’s “long arm” finally enables one individual (or small group) to bear full responsibility for the suit’s fit.
This constitutes the dialectical synthesis: from the eighteenth-century craftsman, through the industrial ordeal of specialisation, the dynamic tax on braking NTPR and the planned oversupply of specialists, back to the master – now equipped with the power of open code and artificial intelligence. Socially, stratification emerges: the elite in perfectly tailored garments, the masses in universal ill-fitting jackets with Patch Tuesday. Economically, there is a net gain: coordination costs fall and responsibility returns to the creator.
Raikin’s atelier marks the conclusion of the industrial era. We are currently in the most acute phase of the conveyor’s transition: giants such as Microsoft spend trillions maintaining yesterday’s threads, AI generates code without holistic oversight, and the dynamic tax on braking progress increases daily. The shift toward personalisation is already underway.
The future lies neither in scale nor in universality. It lies in the bare kernel and the personalised overlay – in a garment cut not for everyone, but for you. In this lies the final verdict on Raikin’s atelier and the emergence of the new master.