The Fiduciary Trap

This chapter was drafted before the Foundation adopted Synthetic Intelligence (SI) as its preferred term. The vocabulary has been preserved as written; the reasoning behind the rename lives on the Mission page.

In 2000, two graduate students at Stanford added an informal motto to their corporate code of conduct. Don’t be evil. The phrase was meant, by their own account, as a half-joke and a quiet rebuke to the older technology firms whose business practices they had grown up watching. Their company was Google. The motto persisted, in various forms, until 2018, when it was quietly demoted from a guiding principle to a closing line. By then, almost nobody at the company believed in it anymore, and many of the people who had believed in it had left.

It is tempting to read this as a story about the corruption of good people by power. It is more useful, and more accurate, to read it as a story about structure. The people who founded Google were not, on average, less ethical than the people who are alive today reading this paragraph. They were college students with a new idea who cared about the world. The trouble is what happened to the company they built, not what happened to the people in it. And what happened to the company is what was always going to happen to it, the moment it took the legal form it took.

This chapter is about why. It is the chapter that, more than any other, explains why the Webspinner Movement has the legal structure it has. It is the chapter that should be most useful to a reader who has watched, over the past twenty years, several technology companies that began with sincere idealism end up in a place that nobody who started them would have wanted to end up.

What “fiduciary duty” actually means

When a company in the United States issues stock to the public, the executives of that company take on a legal obligation to the shareholders. This obligation is called a fiduciary duty. It has two main parts: a duty of care (you must run the company competently) and a duty of loyalty (you must run the company in the financial interests of the shareholders, not in your own interests or the interests of anyone else).

The shareholders, in a publicly-traded company, are an enormous and shifting collection of people, pension funds, mutual funds, hedge funds, and algorithmic trading systems whose collective interest is, by aggregation, the maximum financial return on their investment over the relevant time horizon. They are not bad people. Many of them are pension funds whose beneficiaries are public school teachers and firefighters and your aunt. The structural fact is simply that their interest, as shareholders, is financial return. They are not the company’s customers. They are not, in most cases, the people who use the company’s products. They are the people who own the right to a share of the company’s future profits, and the company’s executives are legally obligated to maximize that share.

This is what is meant by the fiduciary trap. It is not a moral failing of any individual. It is a legal architecture that, once you have entered it, narrows the range of decisions you are permitted to make. A CEO who decided to leave fifty million dollars on the table in order to do a thing that was good for users but bad for shareholders could, depending on the circumstances, be sued by those shareholders for breach of fiduciary duty. The architecture punishes generosity. It rewards extraction.

What this looks like in practice

A search engine that, in 2002, returned the most useful result for any given query was a tool. By 2012, that same search engine had to balance the most useful result against the most profitable result, because the most profitable result paid for an advertisement that showed up first, and the shareholders preferred profit. By 2024, the same search engine was returning, on many queries, several screens of advertisements before any organic result, and the organic results had been rewritten by the company’s own AI to keep the user on the page rather than sending them to the source. None of this was a sudden moral collapse. It was the gradual application of a fiduciary duty that the company had taken on in 2004 when it went public.

A social network that, in 2008, was an interesting way to keep up with college friends had, by 2018, become an attention-harvesting machine whose internal research showed it was contributing to teenage depression and political polarization. The internal research was real; we know it was real because some of the researchers eventually leaked it. The reason the company did not change course, even after the research was clear, is that changing course would have reduced engagement, which would have reduced advertising revenue, which would have violated the fiduciary duty owed to the shareholders. There was, in a precise sense, no legal way for the company to do the right thing.

A motto like don’t be evil cannot survive this kind of structural pressure. It is a piece of cardboard set against a hydraulic press. The press is not malicious. The press has no opinions. The press just keeps applying pressure, year after year, in the direction it was built to apply pressure, until eventually the cardboard gives.

Why we are not going to blame the founders

It is important to be clear: this is not a chapter about how Google’s founders, or Meta’s founders, or the founders of any of the other firms one could name, are bad people. The Webspinner Movement is not built on the resentment of the technology principals. It is built on a clear-eyed reading of what happened to them.

In every one of these stories, the founders made the same choice at the same point. They took outside investment to scale faster. The outside investors, reasonably, asked for a path to liquidity, meaning, eventually, an IPO or an acquisition. The founders agreed, because the alternative was watching a competitor scale past them. By the time the company went public, the structure was set. Once the stock was trading, the fiduciary duty kicked in. The founders, even when they retained voting control, were now operating inside the architecture. They could resist for a while. They could not resist forever.

A few founders, having watched this happen, have tried to design around it. The dual-class share structures that Mark Zuckerberg, the Brin-Page-Schmidt arrangement at Google, and the special voting rights at certain other firms are all attempts to keep the founders in operational control even after going public. They have worked, partially, for as long as the founders are alive and engaged. They do not work after the founder leaves, dies, or loses interest. They are personal solutions to a structural problem.

The structural solution, the one that actually outlasts the founder, is not to go public at all. The structural solution is to incorporate, at the start, in a form that cannot ever be sold or floated. This is what Patagonia did. This is what the Mozilla Foundation did. This is what Wikipedia has done from the beginning. It is what Webspinner is doing now.

What “do good” looks like as a structural choice

The motto don’t be evil is a pledge of inaction. It is a promise not to do specific things that would be harmful. It is, in the language of moral philosophy, a negative duty. Negative duties are real and important. They are also insufficient. I will not stab you is not friendship.

Every great moral tradition the world has produced (and we are not partisan about this; the principle shows up in the Hebrew prophets, the Sermon on the Mount, the Bhagavad Gita, the Buddhist paramitas, the Quranic injunctions on charity, the Confucian ren, and the secular humanist tradition that runs from the Stoics through the Enlightenment) names the same recognition. Do not harm is the floor. Do good is the work. The traditions disagree about almost everything else; they agree about this.

The fiduciary trap reduces do good to don’t be evil, because do good requires sustained sacrifice of profit and don’t be evil only requires avoiding the most prosecutable forms of harm. A company structured to maximize shareholder return cannot mandate doing good. It can, at best, do good when it happens to also be profitable, and stop doing good when the two diverge.

A steward-owned institution can mandate doing good, because it has no shareholder return to maximize. The only thing it has to maximize is the mission. This is not a higher moral commitment held by the people inside the institution; it is a different architecture, inside which the same people will reliably make different decisions. We are not asking Webspinners to be saints. We are building an institution that does not require saintliness to function. That is the entire point.

The lesson, plainly

If you take only one thing from this chapter, take this. The structure you incorporate under is the floor of what you can do. You can build any kind of company you want, but the moment you take the form of a publicly-traded firm with a fiduciary duty to maximize shareholder return, you have set the floor of your moral ambition at don’t be evil, because that is all that the structure permits. The motto on the wall does not change this. The personal sincerity of the founders does not change this. Only the structure changes this.

The Webspinner Foundation is incorporated, deliberately and in writing, with a different floor. The trust deed that holds the operating company says, in legal language, that the institution must keep doing the work it was founded to do, that it cannot ever be sold, that the surplus revenue must flow to mission rather than dividends, and that the user community has elected representation on the body that governs the whole arrangement. The motto on our wall is do good, not because we are better people than the founders of Google were, but because the structure we built lets us mean it.

Twenty years from now, when the Webspinner Foundation is operating at scale, the people running it may not even remember this manifesto. They will not have to. The structure will still be doing its work, the way the structure of the rural electric cooperatives kept electrifying farms long after Roosevelt was gone. That is what an institution is. That is what we are building.