Your Boss Is a Government

The schedule arrives on a Friday. It changes next Tuesday.

Someone has to move a childcare arrangement. Someone has to give up a class. Someone who works two jobs must negotiate a collision between two managers, each of whom regards the other job as a personal complication.

On the spreadsheet, these are adjustments to staffing. In a household, they rearrange a life.

An employer is not a state. It does not possess the same lawful powers, and different countries place different limits on what it may do. Yet a workplace governs a substantial part of the day. It allocates tasks, assesses behavior, controls access to resources and attaches consequences to obedience or refusal.

We should be willing to discuss that authority as authority. Calling a decision private does not make its effects small.

The question is not whether work can function without coordination. It cannot. The question is whether coordination requires ordinary people to live under avoidable arbitrariness, and whether the people carrying out a plan deserve an organized role in shaping it.

The wage does not buy the whole person

A labor contract establishes obligations. A worker agrees to perform work, and an employer agrees to pay under applicable conditions. That arrangement is useful precisely because each party can know something about what is owed.

Trouble begins when the employer’s legitimate claim on work expands into a general claim on availability, silence or personal deference.

A worker may be expected to accept a last-minute change because the business needs flexibility. The same worker’s need for flexibility may be treated as lack of commitment. Management’s uncertainty becomes an operational reality; the household’s uncertainty becomes an attitude problem.

The costs have not vanished. They have been moved.

A business may genuinely face volatile demand. It cannot know every order in advance. A hospital must respond to emergencies. A small restaurant can be vulnerable to a bad week. These are reasons to negotiate workable arrangements, including appropriate compensation and shared planning. They do not establish that the least secure person should absorb every fluctuation for free.

Good management has to coordinate constraints on both sides. If a model is viable only because workers provide unpaid standby capacity and continually repair their private lives around it, the model is being subsidized by those households.

A working-class politics should make that subsidy visible.

Competition you can actually reach

The familiar answer to a bad employer is a better employer. Sometimes it is the right answer. A new job can improve pay, treatment and prospects. A worker should be able to pursue it without unnecessary barriers.

But a labor market is not a list of company names. Jobs differ by location, hours, qualifications, physical demands and compatibility with care. Searching takes time. Information is incomplete. A person may have several nominal alternatives and only one they can realistically reach.

The U.S. Treasury’s 2022 review of labor-market competition describes employer power arising through search frictions, restricted mobility, information gaps and other mechanisms as well as concentration. A market with several employers need not give workers the bargaining position assumed in a perfectly competitive model.¹

This point matters because it changes what we investigate. Counting employers is useful but insufficient. We also need to know whether workers can discover pay, transfer their skills, leave without disproportionate penalties and obtain a job on terms compatible with the rest of life.

Restrictions on mobility are especially revealing. If an employer insists that a worker is freely choosing the relationship, it should have a strong explanation for preventing that worker from choosing another one. There can be legitimate interests in protecting confidential information and specific investments. A broad restraint on a person’s livelihood needs more than a convenient invocation of those interests.

As of this book’s research date, the Federal Trade Commission’s attempted nationwide noncompete rule had been set aside and removed from the federal regulations. Readers should not assume that a general federal ban protects them; state rules differ.²

The policy case for limiting coercive restrictions remains. The practical details require current law, careful drafting and enforceable remedies.

Leaving and speaking

Albert Hirschman’s distinction between exit and voice is helpful here. Exit means leaving or switching. Voice means trying to improve an institution from within. Loyalty can affect which response a person chooses and how long they stay to seek change.³

A healthy labor system needs both usable exits and effective voice. A person should be able to leave a bad job. Those who remain should not be required to accept avoidable harm until everyone individually finds somewhere else.

There is a collective complication. When the people most able to leave depart, the institution can lose some of its most effective critics. Those with fewer alternatives remain. Better outside options can strengthen a negotiation, but mass departure can also weaken the organization needed to change conditions for the next person.

The remedy is not to trap capable workers in bad jobs for the public good. It is to create voice that does not depend on heroic individuals remaining indefinitely. A representative structure, enforceable agreement or independent safety process can preserve an institution’s memory after particular workers move on.

Think of the difference between a respected employee asking a manager to be reasonable and a negotiated rule that applies after both of them have left. The first can solve a problem. The second changes the system’s tendency to reproduce it.

That is what organized labor can make possible.

A union is a structure for answering back

A union brings workers together to bargain over conditions that they may be poorly placed to change alone. Its importance reaches beyond the size of a wage increase. It can establish a procedure, preserve a record, provide representation and make an individual dispute a matter the employer must address collectively.

The Bureau of Labor Statistics estimated that 10.0 percent of U.S. wage and salary workers were union members in 2025, representing 14.7 million people. The private-sector membership rate was 5.9 percent, compared with 32.9 percent in the public sector. These annual estimates excluded October because of the federal shutdown and used eleven-month averages, limiting strict comparison with normal annual series.⁴

Those figures describe membership. They do not tell us whether every represented workplace is democratic or well run. They do show how unusual formal union membership remains in private employment.

Historical research by Henry Farber, Daniel Herbst, Ilyana Kuziemko and Suresh Naidu, using newly assembled survey microdata and several methods, supports a significant inequality-reducing role for unions in the twentieth-century United States. Its historical and quasi-experimental findings are evidence against the idea that collective bargaining never changes distribution; they are not a guaranteed wage premium for every worker today.⁵

The broader moral case does not require a promise that a union will always win. Workers who sustain an enterprise deserve a legitimate institution through which they can contest its decisions. An employer can organize capital, purchase advice and bargain as a durable entity. Workers should not be told that acting together is an unnatural interference with a relationship the other side already conducts collectively.

The union must answer too

A worker can be poorly served by union leadership. Representatives can become complacent, protect insiders, exclude minorities, mishandle grievances or value their relationship with management more than their obligations to members.

These failures should be named without embarrassment. Defending labor does not require defending every labor official.

The democratic answer is stronger member power: meaningful elections, transparent finances, accessible grievance procedures, clear bargaining mandates and a genuine opportunity to question a proposed agreement. Members need understandable information about tradeoffs before a vote, not simply a recommendation wrapped in urgency.

A union also has to decide whom it sees. Permanent staff may have different interests from temporary workers. Senior workers may have more protection than new entrants. Workers in one occupation may be tempted to preserve an advantage at the expense of people doing adjacent work.

Solidarity is therefore a practice of widening the circle, not a word that proves the circle is already wide enough.

Public-sector bargaining adds another set of responsibilities. Workers deserve dignity and effective representation; the public depends on services and finances that must remain sustainable. Users of those services, including other working people, need a voice in the broader design. That creates a difficult negotiation, not a reason to silence one side.

Any institution claiming to speak for ordinary people should be willing to show how ordinary people can correct it.

The algorithm did not invent the target

A worker may now receive instructions from software rather than a supervisor. A route is assigned. A score changes. An alert announces that performance has fallen below a threshold. The interface can make a judgment appear impersonal and therefore inevitable.

But somebody selected the objective, the data, the threshold and the consequence. Somebody decided how much error was acceptable and who would bear it. Even when a system learns complex patterns, the institution remains responsible for choosing to use it in a particular way.

An automated decision can be more consistent than an arbitrary manager. It can also apply a bad rule consistently at scale. Both possibilities deserve attention.

The worker needs more than a general promise of responsible technology. They need to know what is being measured, how errors can be challenged, whether unusual circumstances can be considered and who has authority to reverse a result. Sensitive judgments should not disappear into a vendor relationship in which the employer blames the software and the vendor says the employer chose the policy.

There should be a named institution that owns the decision.

Worker participation before a system is introduced can improve both fairness and function. People doing the task often know when a metric encourages shortcuts, ignores necessary care or mistakes a difficult assignment for poor performance. Their knowledge can prevent an expensive error. It can also threaten a management story that the metric was chosen to support.

A real consultation needs the possibility of changing the system, not merely learning how to comply with it.

A few lines for the end of a shift

The Clock

The clock has counted every hour.
It has not counted what you held:
a name remembered in a hurry,
a fault repaired before it spread,
the patient work of making right
what someone else called good enough.
You leave. The numbers close the day.
Your life has not been entered there.

A poem cannot settle a wage agreement. It can remind us what a measurement leaves outside its frame.

The person who notices a mistake, calms a frightened customer or teaches a new colleague is contributing something. If the institution measures only speed, it may punish the care on which its own quality depends. If it measures only immediate output, it may consume skills and trust without noticing their depletion.

Workers deserve more than praise for this contribution. They deserve conditions under which providing it does not require exhausting themselves.

Bargaining above the individual workplace

Enterprise-level bargaining has an obvious strength: it can respond to the circumstances of a particular employer. It also has a limitation. A responsible employer may face competitors that cut labor costs through worse standards, making it harder to sustain an agreement.

Sectoral or coordinated bargaining can establish a broader floor while leaving room for local arrangements. The OECD’s 2019 comparative synthesis associates coordinated systems with favorable employment and wage-distribution outcomes, while emphasizing institutional design and flexibility. These comparisons are not randomized proof that copying a model will reproduce its results elsewhere.⁶

The practical question is what level of negotiation can address the problem. Some issues belong close to the workplace. Others arise because a whole sector competes by transferring costs onto labor. A common wage floor, training system or portable benefit can reduce that pressure.

Representation matters at every level. A sectoral agreement negotiated by a narrow set of incumbents can exclude small firms, new entrants or vulnerable workers. Broad bargaining should be designed to include those affected and remain open to evidence about its consequences.

The goal is an economy in which decent treatment is less likely to become a competitive disadvantage.

What can be done without pretending it is easy

A worker thinking about acting with colleagues needs accurate information about protections, risks and the appropriate route. In the United States, the National Labor Relations Board describes protections for certain concerted activity by covered employees, including some activity without a union. Coverage has important exclusions, and the circumstances of an action matter.⁷

A general book cannot promise that any particular strike, refusal or disclosure will be protected. Nor should it confuse a legal right with immunity from practical retaliation. People should seek appropriate union, agency or legal guidance before consequential action.

That care is compatible with courage. It is part of respecting the people who will bear the risk.

At an institutional level, several directions are clear. Make lawful organizing protections usable promptly. Reduce unnecessary barriers to changing jobs. Require comprehensible pay and scheduling information. Give workers meaningful representation in decisions that alter the conditions of their lives. Fund enforcement so that a right is more than a sentence waiting for someone wealthy enough to litigate it.

We should also ask employers to recognize the difference between control and competence. A manager who must listen may make a better decision. A worker who can speak without fear may identify a failure early. A negotiated rule can reduce conflict by making expectations clear.

None of this eliminates disagreement. It gives disagreement a civilized form and a chance of producing something better than submission.

Bring democracy to work

Many people spend more waking hours under workplace rules than in any political institution they consciously recognize. It is strange to celebrate their citizenship while treating their daily obedience as beyond democratic concern.

They do not need to vote on every operational detail. They need enforceable boundaries, collective representation and a fair share of the authority over decisions that shape their lives. They need a way to contest a rule without staking their entire livelihood on whether a supervisor takes offense.

This is not a small addition to democracy. It is one of the places where democracy becomes real enough to practice.

A person who learns that their judgment can matter at work carries a different expectation into the rest of society. They have experienced a decision being explained, challenged and changed. They know that coordination can survive disagreement.

The working day should not be a daily lesson in the futility of having a voice.