Artificial Limits

A forthcoming book

Artificial Limits

What progress owes a human day

by Muzzammil Sajjad

Most human possibility disappears without ever becoming evidence.

Institutions judge people from records, and a record is a remainder — what survived the journey into evidence, not the person who made it. That was always a costly approximation. It is becoming an expensive one, because the tools that now expand what a single person can carry are held by the same institutions doing the judging.

Which raises a question that arrives long before anyone has answered it: when work takes less human time, who receives the difference?

Start with the Human Dividend Get notified when it's out

Artificial Limits

Muzzammil Sajjad Forthcoming

The central idea

The Human Dividend

Expanded capacity must return a defensible benefit to human lives through an identifiable transfer.

Follow one hour. A worker completes a bounded task sixty minutes sooner. That hour first belongs to the organization, through the schedule that already exists. Nothing has yet moved into anyone's control.

What happens next is not decided by the technology. It is decided by whatever stands between production and destination — a contract, a collective agreement, a public rule, an ownership decision. Something always carries the gain from one account to another.

There is always a bridge, even when the bridge is simply the existing distribution of power.

The quiet failure path needs no villain. No agreement changes, the target rises, and the same worker receives another task. Output grows while pay, hours and control stay where they were. That does not prove motive. It shows the inherited schedule acting as the transfer rule — and doing nothing is not neutrality, because the hour cannot remain unallocated.

So the dividend is not a slogan about sharing the benefits of progress. It is an accounting obligation. Identify the gain against a counterfactual. Name the institution that transferred it. Name the beneficiary, rather than invoking an abstract workforce. State the horizon, because a benefit this quarter can become a burden next year. Then show the distribution — including the people whose losses cannot be erased by someone else's gain.

If those questions cannot be answered, the dividend is an aspiration, not an outcome.

Where this sits

It refuses both automatic claims

This is not an argument that every gain belongs to labor, and it is not an argument that every gain belongs to whoever owns the tools. Both claims are made confidently and neither survives contact with how production actually works.

As an employer

I do not find it credible to say that all acceleration belongs automatically to the worker. The business carried risk before the gain appeared. It paid for tools, training, coordination, unsuccessful experiments, and the periods when demand did not cooperate. A rule that makes investment irrational can destroy the field from which any dividend would come.

As a worker

I do not find the opposite claim credible either. Selling labor does not transfer every future benefit of judgment, learning and tool use to the buyer without limit. If the contract can absorb every improvement while holding pay, hours, security and authority still, then the person becomes more capable and no more free.

What the Human Dividend asks instead is narrower and harder to dodge: what did each party supply, what changed, what must be protected, and which human consequence completes the gain? The answer may contain time, pay, ownership, transition, lower prices, better service or investment. What it cannot contain is only a more successful institution, while the people inside and around it carry the same or greater necessity.

Plurality is a strength. Vagueness is not.

How the book gets there

The argument, in four movements

Each stage repairs the one before it — and then shows why the repair is not enough on its own.

  1. 01

    A record is a remainder

    Achievement leaves objects behind. Unrealized capacity usually leaves nothing. So institutions measure the fraction of a person that made it through, then treat the fraction as the whole. A useful record accumulates authority beyond its purpose, and the selection loop begins manufacturing its own validation. That is an artificial limit — not imaginary, but built. Built things can be brutally real.

  2. 02

    Sight becomes power

    The same capacity that lets one mind cross more disciplinary borders becomes something else in an institution's hands. Gains are uneven, not universal, and the new bottleneck is selection rather than execution. Adoption is not a virtue. The standard is warranted use: a defined purpose, a real comparison, consequences measured separately, and the ability to stop.

  3. 03

    Four repairs, each insufficient

    Better evidence, then a genuine route, then authority, then durability — the book proposes each in turn and then shows what it fails to fix. Recognition can improve while access stays closed. A route can exist and still be paternalistic without power. Authority says nothing about what remains in the person once the tool is gone.

  4. 04

    Nothing distributes itself

    Even a perfect chain of individual repairs leaves the largest question untouched. Productive capacity and compensation have never been joined by a law of nature. Without an institution capable of carrying the transfer, "human benefit" becomes a label placed over the existing destination of power.

Not only a diagnosis

What the book proposes

A critique that cannot say what it would accept is not a critique. Much of the book is given to decision procedures precise enough to be used, argued with, or refused — including by its own test.

The artificial-limit test

The label has to be earned. Name the necessary work, the representation used, the relevant evidence excluded, the mechanism that closed the route, the evidence an alternative could create, and the risk that alternative imposes. An adverse decision is not proved artificial until that chain holds. The book applies the test to its own proposals.

The strong baseline

A richer or newer system must improve on a strong, relevant baseline — or deliver a specified gain in agency, explanation, correction or contestability that the baseline cannot. The burden stays with the institution. Sunk effort is an organizational fact; it is not moral authority.

Usable power on a real clock

Correction asks whether the data are accurate. Contestation asks whether accurate data were relevant. Remedy asks what happens to the consequence already produced. A process can perform the first well and fail the others entirely — and a thirty-day right against a five-day process is fictional. The institution chose the decision speed, so the resulting urgency cannot fall solely on the person affected.

Follow the boundary

Every claim of benefit or harm has an edge that someone chose. Name the unit, the period, the place, the counterfactual and the affected population before the arithmetic begins. Some outcomes cannot be averaged into one total — and nonnetting does not forbid comparison, it disciplines it.

Where these lead — a named standard for the working day, and the institutions that would have to carry it — is the argument of the book's final chapters.

The companion project

One arena, tested in public

The book's chapters on infrastructure ask what a project must prove before other people can be required to carry its burden. That question is being decided right now, in hearings and permit dockets, by people who will not read a book first. So it has been given somewhere to keep working.

About

Muzzammil Sajjad

The argument in this book began as a private observation about the distance between what a person can see and what they can carry into the world.

Around 1997, at about seventeen, he designed a retail system he called Andromeda: a store that would begin with the person walking in rather than with the building. Tell the system what you are trying to do, and let the physical world organize itself around the answer. He was a graphic artist, and Photoshop was the instrument he had, so he built the idea in layers — change the layer and the outfit changed.

Andromeda was never built. He makes no claim that it would have succeeded, that it was the first of its kind, or that anything later descended from it. Similar ideas emerge independently when people meet similar problems. The point is narrower: the idea did not fail a test. It never reached one. Photoshop could make the idea visible; it could not perform the next hundred acts required to make the system real.

He had seen the same distance closer to home. His father, Asghar Sajjad, worked in petroleum engineering in Pakistan. After the family migrated to the United States, immediate obligations pulled him into work outside that field — a household needed to survive in the present, and recognition of prior experience belonged to an uncertain future. The knowledge and judgment did not vanish when the surrounding institutions could no longer read them easily. But the work a person does to survive quickly becomes the record by which strangers assess what that person can do.

He later worked in wireless retail and opened a store of his own, then moved toward graphic design, web design and creative problem-solving, where the visible work finally began to resemble the thinking that had been there the whole time. He is careful to say that not every limit was external: some opportunities were lost through his own choices, divided attention, and disciplines he had not yet developed. Respecting human possibility does not require flattering it.

He writes now about who is permitted to benefit when the distance between conception and execution closes — and about what expanded capacity owes the people whose lives it was supposed to improve.

Stay posted

Know when it's out

Publication news, and the occasional note when the argument moves. Nothing else.

Enter an email address, like name@example.com

We won't sell or share your address, and you can ask to be removed at any time.