The round hole has cybernetic lineage
The frame the Directory has been working with treats the substrate as consistent and the demand structure as variable. The ADHD substrate functions as it functions across contexts. The demand structure varies — calibrated tightly in some environments, loosely in others. When the demand structure is calibrated for a different substrate’s compliance capacity, what shows up in the ADHD substrate gets caught as deficit. The substrate isn’t broken. The demand structure isn’t meeting it.
The round hole wasn’t always cybernetic. Post-feudal standardisation tightened it once, mercantile bookkeeping conventions tightened it again, post-industrial Taylorisation tightened it further. Each phase narrowed the calibration around the concept of a particular cognitive architecture idealised out of a statistical average and selected against the rest (the natural, inherent variation). The selection criterion shifted with each phase. What stayed constant was that the demand structure had a calibration target and that the target wasn’t the only possible substrate.
Post-1945 the demand structure took a sharp turn. The Macy Conferences from 1946 to 1953 established cybernetics — feedback loops, control systems, the recasting of mind as information processing across modular channels. The cognitive ideal modelled by cybernetics — sustained attention, executive control, rapid context-switching across information channels, suppression of affect to keep signal clean — was then exported into management theory, computer interface design, organisational scheduling, and education. The round hole became intrinsically cybernetic. The substrate the round hole now rejected was the substrate that didn’t process attention as cybernetic compliance with externally-imposed feedback loops.
Digital work is, thus, the cybernetic round hole operating at full resolution. Laptop plus Slack plus Jira plus calendar plus email plus video calls plus browser tabs plus phone notifications — all routing attention through cybernetic feedback loops, all measuring compliance with the routing as typical productivity. What we call the “neurodivergent” substrate doesn’t fit this: not because the substrate is broken but because the substrate isn’t cybernetic by default. It samples broadly, attends to salience over hierarchy, integrates affect with cognition rather than separating them, and doesn’t sustain compliance with externally-imposed feedback loops for arbitrary durations.
This is the lineage the Deshmukh paper, analysed in this article, steps into without naming. The paper’s framing of digital work as architecturally mismatched with ADHD substrate is accurate. But what the paper doesn’t say is that the mismatch is the cybernetic settlement’s selection criterion at work. The round hole isn’t an accident or an oversight. It’s our inheritance of what’s unfolded.
What Deshmukh's survey reveals about the demand structure
The Deshmukh paper is a conceptual framework paper, not large-N empirical research. The author states explicitly that the proposed system is “currently conceptual and in a very early stage of prototyping, with no deployment data.” What the paper offers as empirical material is a co-design survey of 25 ADHD-affected digital professionals — software engineers, IT workers, mid-senior knowledge workers, ages 21 to 54, formally and self-diagnosed, working across remote, hybrid, and in-office contexts. Read with that caveat in place, the patterns the survey surfaces are recognisable and worth naming.
Tab overload — 37% reported 21 or more browser tabs open at any time, 22% reported 12 to 20, 20% don’t track their tab usage at all. Over 80% bookmark content intending to return and never do. Attention fragmentation — over 70% admitted to being constantly or frequently pulled from high-priority tasks. One quoted participant: “I get 8 hours of work done in 1 or 2 hours or I get nothing done for days.” Productivity tool fatigue — Trello, Todoist, Notion, calendar reminders described as rigid, impersonal, and guilt-inducing.
Read these numbers as readings off the cybernetic demand structure operating on ADHD substrate. The tabs are sampled channels — the substrate working as it works. The bookmarking-without-return is the substrate’s broad attention sampling without the cybernetic loop closure that standard work demands. The pulled-away pattern is the substrate sampling salience as it changes, not sustaining arbitrary feedback compliance. The tool-fatigue response to productivity software is the substrate refusing apparatus calibrated for a different cognitive architecture. None of this is deficit. All of it is the substrate functioning while embedded in a demand structure that wasn’t built for it.
Hustle culture appears in Deshmukh’s literature review as the cultural overlay of the cybernetic demand structure — a citation to PLOS ONE work on how “non-attachmental and hustle-driven work cultures” undermine psychological stability and exacerbate attention-related distress, especially among workers with low emotional buffer capacity. Hustle culture isn’t an aberration of digital work. It’s the cybernetic demand structure made culturally compulsory. Workers internalise the feedback-loop discipline and then perform that discipline as virtue. My corpus has previously traced this directly in the cybernetic attention and waning human attention piece.
77% of Deshmukh’s surveyed participants ranked privacy as very important or mandatory. Many expressed discomfort with telemetry-based systems that monitor behaviour or transmit data externally. The privacy concern isn’t paranoia. It’s the participants recognising correctly that any workplace surveillance apparatus, framed as cognitive support, would extend the cybernetic feedback loops further into their cognitive lives. The architectural intelligence of the participants outruns the institutions designing tools for them.
The recursion — AI as more cybernetics, proposed as remedy for cybernetics
Deshmukh’s proposed framework is itself cybernetic from end to end. Lightweight on-device machine learning models sense behavioural signals — tab switching, app focus, inactivity windows. An adaptive feedback engine maps sensed behaviours to nudges, reflective prompts, ambient presence cues. The architecture operates as “feedback loops” within a “systems thinking” framing. The paper draws explicitly on Capra & Luisi’s Systems View of Life and Meadows’ Thinking in Systems primer. Body doubling becomes “ambient presence” mediated by the assistant. The entire framework is cybernetic in its underlying logic.
The recursion is the architectural payoff. Digital work is the cybernetic apparatus operating on worker attention at high resolution, producing the architectural mismatch the paper identifies. The paper then proposes more cybernetic apparatus, deployed at higher resolution on the same worker attention, as the remedy. The ADHD substrate that doesn’t fit the cybernetic demand structure is to be given a cybernetic prosthesis to help it comply. The substrate is not being met. It is being more thoroughly enfolded into the apparatus that produced the mismatch in the first place.
This isn’t a criticism unique to Deshmukh. The AI-for-neurodivergent-support market is overwhelmingly structured this way — more sensing, more feedback, more behavioural inference, more adaptive nudging. The recent cognitive co-regulation piece examines this dynamic at length. The diagnostic apparatus made the substrate visible as deficit; the remediation apparatus extends the same logic by offering compliance support. Both apparatuses share the cybernetic premise.
What looks like recognition, and what the framework leaves out
The Deshmukh framework appears to make several recognitions that should make the piece progressive. The architectural mismatch between digital work and ADHD substrate is named. Attention is reframed away from individual-trait language. Hustle culture is implicated. The design is privacy-first. Body doubling is acknowledged as a community practice. Each of these recognitions, examined more closely, sits entirely inside the cybernetic frame the paper is operating in.
The architectural mismatch is named so it can be addressed cybernetically — not so the demand structure is questioned. Attention is reframed as “emergent property of interaction” and “dynamic state influenced by a feedback-rich ecosystem” — which is the cybernetic definition of attention, not a relocation away from deficit framing. Once attention is conceived as the feedback-loop control function, any deviation from desired feedback-loop dynamics becomes detectable as a deficit of that function — at higher resolution than the trait framing could achieve. Hustle culture is cited via PLOS ONE work on attention-related distress, still inside the cybernetic frame. The privacy-first design relocates where the behavioural data sits without questioning the underlying surveillance logic — a privacy-first cybernetic sensor is still a cybernetic sensor. Body doubling is absorbed as a feature of the apparatus, with community practice translated into ambient mediated presence the assistant can deliver. None of these moves break out of the cybernetic frame. Each refines it.
What the framework leaves out is the political economy of who built digital work this way and who benefits from worker compliance behaviour being recorded as productivity data. Digital work environments aren’t natural phenomena. They were designed. The people who designed them designed them around assumptions about how workers should attend, what counts as on-task behaviour, and what cadences and tools should structure the working day. The fact that those assumptions exclude a substantial fraction of natural human cognitive variation isn’t an oversight to be patched with adaptive AI. It’s a feature of the design. The structural question, then, is not how to help ADHD workers comply with the cybernetic demand structure. The structural question is who keeps redesigning the cybernetic demand structure to require this much compliance, and why.
The framework also leaves out the disclosure trap that any workplace tool calibrated to detect ADHD-typical behavioural patterns produces. My accommodation con piece traces this dynamic: support frameworks become disclosure traps when the structural environment hasn’t been redesigned to be safe by default. The on-device design Deshmukh proposes addresses where the behavioural inference sits. It doesn’t address what the behavioural inference is — which is a profile of cognitive variability that, once produced, can be repurposed against the worker in any workplace that already treats neurodivergence as stigma and remote-era cognitive variability as performance concern. The privacy-first sensor is still a sensor that profiles. The profile is still legible to anyone with access.
Finally, the framework leaves out the question of why “support” arrives as more sensing rather than less demand. The paper’s surveyed voices include: “I get 8 hours of work done in 1 or 2 hours or I get nothing done for days.” The structural response to that voice isn’t an AI that learns to nudge them through the days when nothing gets done. The structural response is a working arrangement that doesn’t require eight uniform hours of cybernetic compliance to count work as work. The apparatus the paper proposes addresses the worker’s symptoms. But the demand structure it leaves untouched is what produces those symptoms.
The structural response the paper doesn't reach for
An actual architectural alternative to digital-work-as-currently-engineered doesn’t begin with the worker. It begins with the demand structure. Fewer tools, not more. Fewer parallel cybernetic channels demanding sustained executive network control. Fewer compulsory turn-taking cadences. Fewer notifications. Less synchronous compliance. Less measurement of the activity surface that doesn’t map onto the substrate’s actual cognitive work.
I’ve been making this structural case here on the Directory across the relevant domains. In workplaces — the accommodation con as alibi to abdicate redesigning the work itself. In NHS care — the right-to-choose con as alibi to abdicate clinical responsibility. In housing and welfare — the architectural arc from school exclusion to neurodivergent homelessness. In each domain the same pattern can be found: the demand structure stays calibrated for one substrate (a concept of “typical”), the rest get caught at terminal outcomes, the institutional response is downstream-of-crisis remediation rather than upstream-of-prevention redesign that benefits all people. Digital work, as it happens, belongs in this list, alongside all of that. It is the cybernetic apparatus operating directly on human attention, calibrated for one ideal-substrate, rejecting the rest, while proposing further cybernetic scaffolding as the alternative to redesigning itself.
The paper’s title names this directly without intending to. “Toward Neurodivergent-Aware Productivity” gives the game away. Productivity is the cybernetic settlement’s unit of account. Neurodivergent-aware productivity is the substrate being absorbed into the unit of account rather than the unit of account being redesigned. The structural alternative is neither productivity nor an enhanced version of it. It is work whose internal architecture doesn’t require cybernetic compliance to count, evaluated by what gets produced rather than by the surface behaviours of producing it.
Citations
Deshmukh, R. (2026) — Toward Neurodivergent-Aware Productivity: A Systems and AI-Based Human-in-the-Loop Framework for ADHD-Affected Professionals — arXiv preprint (not yet peer-reviewed)
Rashid, L., Möckel, C. & Bohn, S. (2024) — The blessing and curse of “no strings attached”: An automated literature analysis of psychological health and non-attachmental work in the digitalization era — PLOS ONE
Eagle, T., Baltaxe-Admony, L. B. & Ringland, K. E. (2023) — Proposing Body Doubling as a Continuum of Space/Time and Mutuality: An Investigation with Neurodivergent Participants — ASSETS ’23
Hayles, N. K. (1999) — How We Became Posthuman: Virtual Bodies in Cybernetics, Literature, and Informatics — University of Chicago Press
Capra, F. & Luisi, P. L. (2014) — The Systems View of Life: A Unifying Vision — Cambridge University Press
Meadows, D. H. (2008) — Thinking in Systems: A Primer — Chelsea Green Publishing
