The workload fairy tale: why most knowledge work is optional

Executive overview

Most knowledge workers operate under a false assumption: that the amount of work they currently do is exactly what their role requires. Four-day workweek studies across Iceland, the UK, and Germany show productivity stayed flat or improved when hours were cut — revealing that most of what fills a 40-hour week is not the work that actually creates value.

The key work in most jobs requires far fewer than 40 hours to complete. The rest is pseudo-productivity, make-work, and optional activity.

Core insight: The workload fairy tale is a belief that your current workload is the minimum needed for success — but the data shows it isn't.

The workload fairy tale

• Knowledge workers have unusual autonomy over their workload compared to service or industrial workers
• No manager can typically describe what any individual employee is working on at a given moment
• Most workers themselves don't maintain a full picture of their commitments at any time
• This autonomy leads to the workload fairy tale: the belief that the current amount of work is precisely what the role demands
• Time-block planning reveals the truth: most weeks contain only 10–15 hours of work that directly produces value
• The rest is email, unnecessary meetings, pseudo-productivity, and non-promotable activities
• Four-day workweek studies confirm this — removing 5–8 hours per week left productivity unchanged

What the four-day workweek data actually shows

• Iceland: 2,500 workers (~1% of the workforce), 2015–2019 — productivity maintained or improved
• UK: 60+ companies, ~3,000 employees, 2023 — productivity maintained or improved in nearly every case
• Germany: 45 firms, 2024 — workers felt better, remained as productive, reported lower stress and burnout
• The scandalous implication: if removing time doesn't reduce output, what is the extra time actually producing?
• The real fix is not a shorter week — it is recognising and reducing the make-work itself

Hyperactive hive mind and communication tools

• In knowledge work, the dominant collaboration style is the hyperactive hive mind: working things out on the fly through unscheduled back-and-forth messaging
• This style requires constant monitoring of communication channels — email, Slack, Teams
• Continuous context-switching severely diminishes cognitive capacity and causes burnout by mid-afternoon
• Slack is not the problem — it is a better tool for a bad workflow
• Slack reduces the friction of hive-mind collaboration, which means workers must be even more hyperactive to keep up
• The solution is changing the collaboration style, not the tool

Managing multiple writing projects

• For similar project types, sequential work outperforms interleaving — switching between two scripts daily wastes cognitive overhead on every switch
• Large projects can be broken into big chunks and done sequentially before switching to another
• Different project types (e.g. a book plus shorter articles) can interleave if the smaller items are in a research or thinking phase
• Actual writing should focus on one project at a time; background research on another can overlap
• The goal is maintaining a single cognitive orientation — a general mental state oriented toward solving one large problem

Two levels of context-switching cost

• Immediate focus switching: costs 10–20 minutes to fully shift attention; the cause of damage when checking email or Slack repeatedly
• General orientation switching: a larger overhead that takes days to fully reassign when moving between major projects
• For tasks that take many weeks, daily switching costs are negligible; the orientation cost is what matters
• Work on a major project for multiple weeks before switching; push to a clear finishing point when possible
• Research and preparation can overlap across projects; actual deep execution should not

Lifestyle-centric career planning

• Lifestyle-centric planning means fixing a vision of what daily life should look like, then working backwards to career decisions
• It is not synonymous with low effort or "lifestyle entrepreneurship" — moral ambition can be a vivid component of the vision
• In your 40s, career capital is built and ready for reconfiguration — this is the ideal time to redesign
• Avoid solutionism: resist anchoring on a single fix (e.g. "go to 50% time") before articulating what the ideal life actually looks like
• Explore multiple paths, including schedule changes, role changes, or a partner shifting hours
• Bespoke opportunities — the right job at reduced hours, the autonomous role — don't appear on LinkedIn; they surface when you start asking and talking to people

Weekly planning and Sunday anxiety

• Planning on Friday for the following week eliminates background anxiety across the entire weekend
• The brain generates low-level unease when the week ahead is unresolved — Friday planning removes this
• If Sunday anxiety is severe, block a hard calendar stop on Fridays and treat it as a non-negotiable meeting

AI agents and knowledge work automation

• An AI agent is software wrapped around a language model that can take real-world actions based on model outputs
• The case for agents rested on scaling laws: larger models would keep getting smarter, enabling general automation
• Scaling laws have faltered — models beyond GPT-4 have not delivered equivalent leaps; Meta's "Behemoth" model was delayed because it was not sufficiently better than its predecessor
• The industry has shifted to reinforcement fine-tuning: taking a foundational model and training it on synthetic data sets to excel at specific task types (reasoning, maths, code)
• This produces bespoke capabilities, not general intelligence — and most real knowledge work lacks usable synthetic data sets
• Automating something as basic as answering a full email inbox requires idiosyncratic knowledge about people, history, and context that cannot be captured in a training set
• The genuine near-term gains: natural language interfaces to existing software (huge productivity lift, underappreciated) and smart search — conversational access to web, documents, and data
• Smart search is already the killer app of generative AI; Google search alone earns $175 billion per year — this is the revenue battleground
• Do not extrapolate from "we improved from A to B" to "we will reach C" — full job automation requires a credible technological path, which does not currently exist