Rethinking How We Organise Work

AI tools are everywhere, but the big productivity boom many leaders expected has not arrived. Most organisations report modest gains at best, despite widespread adoption of tools like chatbots, copilots, and workflow automation.

A growing body of evidence suggests the real bottleneck is not the tasks themselves, but the way we coordinate people to do them.

This article outlines a research program that asks a simple but radical question: if AI and blockchain can dramatically reduce the cost of coordination, do we still need organisations to look the way they do today?

Why Organisations Exist

A Practical Take on Transaction Cost Economics

In everyday language, most leaders assume organisations exist to "bring people together to get things done." Transaction Cost Economics (TCE) sharpens that intuition considerably.

What is Transaction Cost Economics?

Transaction Cost Economics is a way of explaining why some activities are done inside firms (with employees, managers, and hierarchies) while others are done through markets (contracts with suppliers, freelancers, platforms). The core idea, as explained in Wikipedia, is that doing business is not free: every interaction involves hidden transaction costs such as:

• Search costs: finding the right person or partner for a task.

• Bargaining costs: agreeing on scope, price, deadlines, and responsibilities.

• Monitoring costs: checking whether work is being done well and on time.

• Enforcement costs: dealing with disputes, rework, or non-performance.

Ronald Coase argued in his landmark 1937 paper, The Nature of the Firm, that firms exist because, historically, coordinating through a hierarchy (managers directing employees) was cheaper overall than coordinating everything via contracts and markets. Oliver Williamsonson later built on this by showing that three factors, how unique the assets involved are, how uncertain the environment is, and how often transactions recur, determine where the "boundary of the firm" should sit.

Critically, Williamson also identified two behavioural realities that drive costs up: bounded rationality (we cannot foresee all risks in advance) and opportunism (people sometimes act in their own interests at others' expense). These make arm's-length market contracts risky, and push activities inside firms where management can monitor and intervene.

In practice, TCE asks leaders: for each type of work, is it cheaper and safer to do this in-house under management control, or to outsource it, once all hidden coordination costs are included?

The Hidden Half of Knowledge Work: Coordination Overhead Modern knowledge work is full of invisible transaction costs. Pilot observations from software engineering teams suggest only around half of working time goes into direct value creation, coding, design, problem-solving, with the remainder absorbed by coordination and tool friction.

Key categories of overhead include:

Synchronisation costs: meetings, standups, planning sessions. Handoff costs: transferring context between people and teams. Waiting costs: blocked work due to approval chains, dependencies, or missing information. Context-switching costs: juggling tools, re-finding information, shifting between tasks.

Most studies of AI's labour market impact classify entire occupations as "exposed" to automation, treating tasks as independent of each other, and rarely measuring the coordination layer explicitly. For leaders, this means the biggest AI opportunity may not be automating core tasks, but redesigning the way work is coordinated, a distinction with enormous practical implications.

The AI and Blockchain Shift: New Coordination Technologies The original TCE story assumed that monitoring, verifying, and enforcing agreements are expensive human activities. Today, several technologies directly target those costs:

AI-powered verification: AI models can review code, documents, or support tickets for quality and compliance, reducing manual monitoring overhead. Reputation systems: persistent, portable records of performance reduce search costs, because you can trust someone's history even without a prior working relationship. Smart contracts: as Chris Berg explains in his primer on institutional cryptoeconomics, self-executing agreements on blockchains automatically release payments or trigger actions when predefined conditions are met, cutting bargaining and enforcement costs for certain transaction types. Cryptographic audit trails: tamper-evident logs create trustworthy records without continuous managerial oversight. Task graphs and automated dependency tracking: structured representations of work that make who is doing what, and what is blocked, visible without endless status meetings.

Together, these technologies raise a provocative possibility: for some kinds of knowledge work, market-like coordination (people or AI agents picking up tasks from a pool under smart contracts) might now be cheaper than traditional management hierarchies.

Trust as the Bottleneck: Game Theory and Coordination Traps Underneath transaction costs lies a simpler concept: trust. When people do not fully trust each other, they protect themselves through meetings, approvals, documentation, and oversight. Each action is individually rational but collectively wasteful.

What is a Nash Equilibrium?

Game theory studies situations where each participant's best move depends on what others do. A Nash equilibrium is a stable state where no one can improve their outcome by changing strategy alone, given what others are doing. The crucial, and often frustrating, catch is that a Nash equilibrium can be stable and still be bad for everyone involved.

In organisations, many coordination patterns resemble exactly these "bad" Nash equilibria:

Information hoarding: if others keep information to themselves, sharing openly feels risky. Defensive documentation: if others over-document to protect themselves, you must too or appear negligent. Long approval chains: if managers don't trust teams, additional verification layers become the norm. Meeting overload: if decisions happen in informal gatherings, you attend everything to avoid being left out.

Each person is acting sensibly in context, yet the system as a whole is inefficient. As coordination game theory shows, social dilemmas arising from self-interested behaviour can persist even when everyone would prefer a better outcome.

The research program asks: can cryptographic verification, algorithmic enforcement, and transparent reputation systems change the underlying "game" so that cooperation and lean coordination become the new stable equilibrium, even among strangers who have never worked together?

Institutional Cryptoeconomics: Institutions as Code To understand that question, the project draws on institutional cryptoeconomics, an emerging field that treats blockchains as a new kind of economic institution.

What is Institutional Cryptoeconomics?

Traditional institutions, firms, markets, governments, exist to help people cooperate when trust is limited. They provide rules, dispute resolution, and enforcement, but they are costly to build and maintain. As Berg, Davidson, and Potts argue, institutional cryptoeconomics treats blockchain-based systems as "institutions in code" that can:

Encode rules (via smart contracts). Provide shared, tamper-evident records (via distributed ledgers). Align incentives (via tokens, staking, rewards, and penalties).

The central claim is that these cryptographic institutions can reduce the cost of trust, making it cheaper to know who did what, when, and under what rules. If that is true, they might support new organisational forms that outperform traditional firms for certain types of work.

One prominent example is the Decentralized Autonomous Organization (DAO): a community or project coordinated primarily through smart contracts and on-chain governance rather than managers and corporate structures. In a DAO, tasks, rewards, and voting rights can be codified and executed algorithmically, with transparent histories of contribution visible to all participants.

This research does not assume DAOs are the future of all work. Rather, it treats them as one test case in a broader comparison of coordination mechanisms, a way of stress-testing whether the theoretical promises of institutional cryptoeconomics hold up in practice.

From Jobs to Tasks: A More Granular View of Work Most workforce discussions treat roles like "software engineer" or "customer support agent" as single units. Task-based labour economics instead models jobs as bundles of tasks, each with different characteristics.

What is Task-Based Labour Economics?

Pioneered by Autor, Levy, and Murnane (2003), task-based labour economics asks: what are the individual tasks that make up a job, and which of those can be automated, augmented, or must remain human-only? It shifts attention from job titles to the specific activities people perform day-to-day.

This research extends that lens by:

Mapping task dependencies: how tasks rely on each other and when handoffs are required. Separating execution time from coordination overhead. Assessing AI suitability at the level of specific tasks, not whole occupations.

For organisations, this yields more actionable questions: Which tasks are bottlenecked by coordination rather than skill? Which are genuinely ripe for AI support? Which demand human judgement, empathy, or relationship-building and should remain human-led?

Three Competing Coordination Models The empirical work will compare three broad coordination mechanisms applied to real projects, with software development as an ideal starting domain:

1. AI-Augmented Traditional Hierarchy Existing management structures remain, but AI handles some status reporting, task suggestions, and quality pre-screening. This is the "light touch" transformation most organisations will recognise, and serves as the control condition.

2. Centralised Algorithmic Coordinator An AI system functions like a dynamic dispatcher: assigning tasks, monitoring progress, and adjusting priorities or pricing dynamically, while human managers set goals and handle exceptions. This tests algorithmic coordination under centralised control.

3. Decentralized Autonomous Organisation (DAO) A task marketplace mediated by smart contracts, where contributors, human or AI agents, claim tasks, stake reputation or tokens, and are paid automatically once work is verified. Reputation and staking mechanisms aim to align incentives and manage risk without traditional line management. This tests market-based coordination with algorithmic enforcement.

Across all three, the research measures:

Transaction costs and coordination overhead. Output quality and reliability. Worker trust, autonomy, and perceived fairness. Distribution of value: who captures what share of the surplus.

How the Research Will Run (In Practice) The study unfolds in four main phases:

Phase 1, Baseline measurement in traditional organisations Observe knowledge workers across five professions (software engineering, customer support, financial advisory, UX design, and data analysis), logging time use, tools, and interactions to quantify how much time goes to value creation versus coordination and tool friction.

Phase 2, Task decomposition and AI benchmarking Take real work artefacts (code, tickets, reports, designs), work backwards to identify the minimal set of tasks required, strip out avoidable coordination overhead, and benchmark frontier AI systems on those tasks against human performance.

Phase 3, Mechanism design and pilot implementations Design concrete coordination mechanisms based on insights from phases 1 and 2, then run real projects through each mechanism with partner organisations and communities, including a participant-observer role inside a newly-launched DAO.

Phase 4, Comparative analysis and frameworks Use the data to build practical decision frameworks: how to measure transaction costs in your organisation, how to assess AI suitability for specific tasks, and how to choose between coordination mechanisms given your context.

For practitioners, the output is not just academic theory but usable toolkits, measurement templates, assessment rubrics, and design patterns for alternative ways of organising work, validated across multiple professions and contexts.

What Leaders and Practitioners Stand to Gain If coordination overhead truly consumes 40–50% of knowledge work time, even modest improvements could unlock productivity gains larger than many direct automation forecasts suggest. Some concrete benefits this work aims to enable:

Clearer visibility of where time actually goes in your organisation. Evidence-based identification of high-impact AI and automation opportunities. Practical patterns for reshaping processes, teams, and platforms to reduce coordination friction. Better decision frameworks for when to rely on hierarchy, when to centralise coordination algorithmically, and when to experiment with more decentralised models. Insights on worker experience, how different mechanisms affect trust, fairness, and autonomy, which is crucial for responsible adoption of AI and algorithmic management.

Ultimately, the question is not "Will AI replace jobs?" but "How will AI and new coordination technologies change the way we structure collaboration itself?" This research is designed to provide grounded, empirical answers that leaders can act on, moving beyond hype cycles and anecdote to rigorous, practical measurement.