Why hierarchy is a poor operating system for knowledge work

The first corporate org chart was drawn in 1855 by Daniel McCallum, general superintendent of the New York and Erie Railroad, with the engineer George Holt Henshaw. It looks like a tree, with the board of directors at the bottom as the roots and the stations up in the branches. McCallum was solving a real problem. You cannot run a railroad several hundred miles long by walking the floor, and the telegraph had just started pouring more information onto executives than any of them could process alone.

Fifty-odd years later, Frederick Taylor published The Principles of Scientific Management (1911) and gave that structure its doctrine. Break work into elementary tasks, hand each task to a worker, time it, repeat it faster. Everything was measured in widgets produced per unit of time.

It worked for productivity. The work itself was boring, exhausting, and its repetitive nature was soul-crushing. Many of the people arriving at the factory had been artisans and small shopkeepers, running their own affairs. They went from full self-direction to being treated as extensions of machinery, tuned continuously for more output.

Workers complained. The answer was to invent a class of workers whose job was to keep the shop floor in check, managers, with their apparatus of measurement, planning and control.

The economist Stephen Marglin put that reading on paper in 1974 in What Do Bosses Do?, where he argues the major innovation of the industrial revolution was organisational before it was technical. The linear hierarchy of the crafts, master, journeyman, apprentice, where every rung led to the next one, gave way to a pyramid, boss, foreman, worker, where the worker stays a worker. The historian David Landes answered him in 1986 in the Journal of Economic History, arguing the factory won on technical efficiency grounds. The debate is still open. Whether control was the goal or the side effect, it is part of the design.

Today's org chart

A hundred and seventy years after McCallum, a company org chart still looks a lot like his. We still have order givers on one side and order takers on the other, separated by one or several layers of managers whose job is to check that the orders got through.

The layer has grown. Working from Bureau of Labor Statistics figures, Gary Hamel and Michele Zanini counted 23.8 million managers, supervisors and administrators in the 2014 American workforce, one for every 4.7 employees, 17.6% of headcount and close to 30% of total compensation. In the Harvard Business Review of September 2016 they put the cost of excess bureaucracy at more than $3 trillion a year, roughly 17% of US GDP. The estimate is arguable. The trend it measures is much less so.

Work, on the other hand, changed shape. Knowledge work has no cadence and no widget to count, and the person who spots the problem first is almost always the one furthest from the top. We kept the structure, swapped the content, and now wonder why the thing grinds.

1. A single point of failure

A hierarchical system rests on a handful of decision makers, sometimes one, expected to be the source of truth at all times, in a world where every decision crosses more variables than it used to.

Raaj Sah and Joseph Stiglitz modelled this back in 1986 in the American Economic Review. Two architectures compete: the hierarchy, where a project has to clear several filters in sequence, and the polyarchy, where several units decide in parallel. The hierarchy rejects a larger share of good projects. The polyarchy accepts a larger share of bad ones. So the real choice is which kind of error you would rather make. A nuclear plant is right to filter. A company still looking for its market is better off placing more bets.

2. Every layer filters what travels upward

In a hierarchical system, the big decisions lean less on ground-level data than on filtered information: hearsay, reports, decks polished three times over. That information crossed several floors of internal politics before it arrived.

The pyramid is good at pushing information down. An order leaves the top, each relay repeats it, it lands. It is bad at pulling information up, because the useful information is diffuse, scattered, and the signals are weak.

It is also bad at it because nobody enjoys carrying bad news. In 2003, Frances Milliken, Elizabeth Morrison and Patricia Hewlin interviewed forty full-time employees across a range of industries for the Journal of Management Studies. 85% had been in a situation where they judged an issue important and said nothing to their supervisor. The sample is small, and twenty years of research on organisational silence has pointed the same way since.

In a fast-moving market, what you want is as many sensors on the ground as possible, catching weak signals early. The pyramid has plenty of them. It wires them the wrong way round.

3. Disconnected decisions breed disconnected teams

Information travels up and down the same filtered channels and warps in a thousand small ways. What comes out are decisions so far from the ground that teams receive them dumbfounded.

A fracture opens between leadership and the teams, who struggle to own those decisions enough to carry them out well, because they cannot make sense of them. Then disillusion sets in. Motivated people turn into automatons, wait to be told what to do, and watch the clock.

Gallup has been tracking that slope for fifteen years. In the State of the Global Workplace report published in 2026, the share of engaged employees worldwide falls to 20% over 2025, the lowest since 2020, at an estimated cost of $10 trillion in lost productivity, or 9% of global GDP. Manager engagement itself went from 27% to 22% in a single year. The layer meant to pass the energy along has none left to pass.

Bain supplies the other half of the picture. With the Economist Intelligence Unit, the firm surveyed 300 companies with revenues above $500 million: the average company loses a quarter of its productive capacity to organisational drag, redundant coordination and meetings that decide nothing. The top quartile loses half as much.

Disengagement settles in slowly, because of the way work is structured.

4. You are missing your team's actual potential

Before the factory, salaried work was often a stage on the way to independence. In England, the Statute of Artificers of 1563 required seven years of apprenticeship before anyone could practise a trade. Seven years, then you became a journeyman, then a master with your own shop. In practice many stayed journeymen for life. The path existed all the same, and everyone knew it.

The industrial pyramid removed that horizon. You come in a worker, you leave a worker. What we inherited is a mode of prolonged adolescence at work, with a ceiling for everyone, the ceiling of the father figure, the boss, looking over your shoulder and stepping in at the first misstep.

Look at your teams as they are today. Your employees run at a fraction of their capacity, wait to be told what to do, and chase carrots.

5. The incentives are the wrong ones

I should have paid more attention to the business incentives class in school. It was probably the single most important class of my business school years.

Incentives drive everything. Find the ones that align company health AND employee health and you have something powerful. Promotions and individual bonuses look like that alignment. Up close they do something else.

Internal competition is a zero-sum game

A hierarchical structure breeds competition by construction. One person at the top, fewer managers than employees, so climbing means outshining your peers.

You are creating winners, some will say. What you are mostly creating is an individual game whose object is to climb a ladder, and the only way to climb it is for others to stay put. For you to win, your peers have to lose.

That climb can reward a real contribution. It does so less and less the higher you go. Alan Benson, Danielle Li and Kelly Shue tracked 53,035 sales workers at 214 American firms between 2005 and 2011, 1,531 of whom were promoted into management, and published in the Quarterly Journal of Economics in 2019. Firms systematically promote their best salespeople. Doubling a rep's pre-promotion sales then comes with a 7.5% decline in the sales of each of their subordinates. Other indicators already sitting in the data predicted managerial quality better, and firms go ahead without them.

The higher you climb, the more political the game gets. Who you know, who will back your rise, which manager likes you enough to sponsor you. Whoever reaches the top may well be the best, but the best at what? At growing the company, or at playing that particular game?

Hierarchies pit people, teams and departments against each other in a race for promotions, resources and budget. The weapons are familiar: charts that only go up and to the right, reports that serve their author more than their company, and an arsenal of political tactics perfected by management gurus with books to sell.

What about everyone else? They either play or they get played. Either they work their manager for the next promotion, or the environment wears them down until the day they find themselves quiet quitting.

You can win that game as an individual, provided you are motivated enough to play it. A company whose entire incentive system sets people against each other loses in the end.

The serious objection

Hierarchy lasted a hundred and seventy years because it earns its keep. Harold Leavitt made the point in the Harvard Business Review in 2003: hierarchies thrive because they deliver real value, because they absorb one managerial movement after another, and because they answer a need for order and security that few alternatives cover as well.

Abrupt transitions confirm it. Zappos adopted Holacracy in 2014, then in March 2015 Tony Hsieh offered severance to anyone who preferred to leave. 18% of the workforce took it, around 260 people. Pulling the hierarchy out without putting anything in its place leaves a vacuum that informal power fills fast. That is the main lesson from the liberated companies that ran out of steam.

The organisations that hold have replaced the management layer with something else. Buurtzorg, founded in 2006 by Jos de Blok, runs more than 10,000 Dutch nurses in self-managing teams of ten to twelve, with around fifty people in the back office and two directors. The case study the Commonwealth Fund published in 2015 reports 108 hours of care per client per year against 168 for the sector average, with better patient ratings. Morning Star, the largest tomato processor in the world, has operated without a single management position since 1990 on two rules: commitments are kept, and interactions stay voluntary.

Those organisations carry more explicit structure than average. It is simply distributed.

What takes the place of the management layer

Three building blocks show up everywhere.

Written roles

Each role carries a purpose, a domain it decides on alone, and accountabilities others can expect from it. A role is smaller and more mobile than a job, one person holds several, and it changes when the need changes. That is the whole difference between a role and a job description.

An explicit decision rule

Knowing who decides, on what scope, and what happens when a topic exceeds a single role. Consent decision-making answers that without requiring unanimity and without sending everything back to the CEO.

A place where the structure evolves

A regular governance meeting is where a team creates a role, amends another, or processes a tension about how they work. Without that space, the org chart freezes and the gap with real work widens month after month.

That is exactly what Rolebase sets up. An org chart where every role carries its purpose, its domain and its accountabilities, decisions recorded inside the role that made them, and a governance mode that defines who can change what, from Free mode where anyone edits everything to Strict mode where every structural change goes through a proposal. See what Rolebase does, or create your organisation for free up to five active members.

What are the main disadvantages of a hierarchical structure?

Decision-making concentrates in a few hands, ground-level information warps on the way up, teams execute decisions they barely understand, and promotions reward past individual performance rather than the ability to make others succeed.

Why does information travel up a hierarchy so badly?

Every level sorts what it passes on according to its own interests and biases. Milliken, Morrison and Hewlin found in 2003 that 85% of the employees they interviewed had stayed silent on something they judged important, either fearing they would be seen badly or believing it would change nothing.

Should you remove all managers?

Removing managers without installing anything else leaves a vacuum that informal power occupies immediately, as several failed transitions have shown. What works is replacing position-based authority with written roles, a known decision rule, and a regular space where the structure evolves.

Does a structure without hierarchy actually work at large size?

Buurtzorg runs more than 10,000 nurses in self-managing teams of ten to twelve with two directors, and Morning Star, the largest tomato processor in the world, has done without management positions since 1990. Those organisations carry an explicit structure, distributed rather than pyramidal.