Three questions, one ledger
Who gets paid. Who owns it. Who's watching.
Every large company already reports what its CEO makes and what its typical worker makes. Almost no one reads the filings. This page draws them to one scale, then asks the two questions that come next.
Floor to top, one scale, from the company's own filing.
How much of the company the people who run it actually hold.
A design for AI that audits companies instead of serving them.
Write to your representatives in two minutes
CEO pay as a multiple of the median worker
Ratios are each company's own SEC or annual-report disclosure for its latest fiscal year, except the UK figure, which is the High Pay Centre's index-wide median. Companies use different methods, so treat cross-company comparison as approximate. Green marks ratios under 100 to 1: same laws, a fraction of the gap. New companies are added regularly; the data file is public.
It wasn't always like this
In 1965 the ratio was about 20 to 1. Your grandfather's CEO made a very good living, not a fortune.
The gap on this page is not a law of nature. It is about forty years old. Economists who reconstructed executive pay from the 1930s onward found big-company CEOs earning roughly 20 times a typical worker through the 1950s, 60s, and 70s: a professional's salary scaled up. Then, after 1980, it took off.
Historical figures are the Economic Policy Institute's series for the 350 largest US firms, measured against a typical full-time production worker in the same industry. The Starbucks bar is that company's SEC disclosure against its own median employee, part-timers included, so it isn't the same yardstick. The fair comparison is Starbucks' 1,794 against today's average of roughly 290: an extreme outlier even by modern standards, and modern standards are already ten to fifteen times what they were in 1965.
Three things changed after 1980
Pay became stock
Options and equity awards replaced salary as the main form of executive pay. From then on, CEO compensation tracked the stock market instead of the company's payroll. When the market went up, the top went up. Nobody else's pay was tied to it.
The top tax rate collapsed
The highest marginal income tax rate was 91 percent in the 1950s and 70 percent until 1981. It fell below 40 percent by the end of that decade. At 91 percent, an extra million in pay was mostly a gift to the Treasury and not worth the fight. At 37 percent, it was worth everything.
Worker pay stopped moving
Productivity kept climbing. Median wages, adjusted for inflation, roughly flatlined. The bottom of the ratio stopped growing at the same moment the top took off. The gap is the two lines pulling apart for forty years.
That's the question this page is asking you to carry into the room with whoever represents you: if it was 20 to 1 when the country built the interstate highways and put a man on the moon, why is 290 normal now, and why is 1,794 legal?
The second question
Pay is the part you can see. Ownership is where the money lives.
Cap every executive on this page at twenty times their median worker and you move roughly $150–200 billion a year worldwide, about a tenth of one percent of the global economy. Real, and worth doing, but not the whole story. A salary is labor income. Most extreme wealth is ownership: shares, dividends, capital gains. A pay cap does not touch it.
So the ledger is adding a second column: what share of each company is held by its rank-and-file employees. For most large companies the answer is close to zero. That is a number nobody publishes and everybody would understand.
Five ways ownership has already been reset somewhere
| Mechanism | Where it already works | What it does |
|---|---|---|
| Broad employee equity | Publix, WinCo, John Lewis (UK), thousands of ESOPs | Every worker gets shares every year as part of pay. Nothing is taken; new ownership is created and spread. |
| Citizens' wealth fund | Alaska Permanent Fund, Norway's sovereign fund | The public holds a slice of the economy and everyone gets the dividend. |
| Tax ownership like work | Proposed widely, partial in several countries | Equal rates on capital and labor income; a real inheritance tax above a very high threshold. |
| Antitrust | US 1911–1984, EU today | Concentrated ownership rides on concentrated markets. Break up the market, spread the ownership. |
| Workers on the board | Germany, above 2,000 employees | Doesn't change who owns shares. Changes who owns decisions. |
The honest objections: capital taxes may reduce investment; wealth funds can be politicized; single-company employee stock is risky if the company fails, which is why Enron's workers were wiped out; founders who take large risks arguably earn large rewards. Serious versions of any reset are built around those, with diversified employee funds instead of single-company stock and thresholds high enough to touch only the top fraction of a percent.
AI companies audit each other. Citizens audit the AI. And exposing corruption pays better than hiding it.
The third question
Who watches
The gap on the ledger persists for one reason: nobody is required to close it and nobody is paid to expose it. Change both. Four layers, each proven somewhere, never assembled. This is also how you keep AI itself honest while it is still small enough to hold.
Three AI monitors watch each other and every large company
Three independent AI systems, run by organizations in different countries with different funding and different training. Each audits the other two, plus the payroll, executive pay, ownership, supplier contracts, and political spending of every company above a size threshold, in every country it sells in. Their auditing code is public. When they disagree, the disagreement is published too.
Citizens chosen by lottery check the AI
Panels of ordinary people selected at random from every participating country, not elected and not appointed. Short terms, paid well, rotated constantly, identities protected while serving. They can demand plain-language explanations, order re-analysis, and escalate findings. The AI cannot overrule them.
Whoever exposes corruption gets paid
Anyone, anywhere, who reveals concealment or wrongdoing receives 20 to 40 percent of the penalty, funded from the penalty itself. The largest payouts go to exposing collusion among the monitors or inside a citizen panel, because those are the failures that would break the system. Legal protection and income replacement come with it.
The US SEC whistleblower program has paid out billions since 2011 for exactly this reason: it made reporting fraud more profitable than participating in it.
Humans judge. The AI only finds and publishes.
No monitor imposes a penalty. Findings that survive citizen review go to courts and public enforcement bodies in the relevant country. The AI is the auditor. It is never the judge.
Why this reaches racism
The bottom of every ladder on this page is disproportionately Black, brown, immigrant, and women: cashiers, baristas, warehouse workers, and the people in the factories overseas. Raise the floor and those are the people raised. Publish pay by job and location and unequal pay for equal work has nowhere to hide. Pick the watchers by lottery from everyone and the system cannot quietly become one group watching itself. It doesn't end racism. It takes away the dark, and racism needs the dark.
The rules it would enforce
The monitors enforce whatever the public sets. This is the starting proposal. It applies to every company above the size threshold, in every country, including their suppliers.
| Tier | Rule |
|---|---|
| Floor | A local living wage in every country of operation, suppliers included. Not the legal minimum: enough for housing, food, healthcare, and savings where the person actually lives. |
| Frontline | Living wage plus meaningful equity. Full-time available to everyone who wants it. |
| Skilled | Market pay. Bands published by level and location. |
| Executives | Total compensation capped at 15 to 20 times the company's global median employee. |
| Innovation | Royalties on specific products, paid to the people who built them, capped per person per breakthrough. Reward invention where it happens, not two levels up. |
| Ownership | Broad employee equity every year, held through a diversified fund so no worker's savings ride on one company. |
| Profit share | A fixed percentage of net income distributed to all employees. |
| Transparency | Pay ratio and bands published every year, in plain language, in every language the company operates in. |
Applied to Apple, the CEO goes from $74 million to about $3 million, every store worker gains $10 to $15k, every factory worker in the supply chain reaches a living wage, and the company stays the most profitable on earth. The cost fits inside one year of stock buybacks. The same arithmetic holds for nearly every company on the ledger.
Is anyone building this?
What exists
- Companies use AI to audit themselves. The vendor works for the company.
- Governments are regulating AI. They are not using it to audit corporations.
- Whistleblower bounties exist and work at the SEC and a few other agencies.
- Citizens' assemblies have run successfully in Ireland, France, and dozens of cities.
- Pay ratios are public by law in the US and UK, buried in filings almost no one reads.
What is missing
- A publicly owned AI that audits companies rather than serving them.
- Any arrangement where AI labs monitor each other.
- A connection between the bounty and the detector.
- Anyone with the incentive to build it. The labs sell compliance tools to the companies they would be auditing.
Every component is proven. The assembly is unbuilt. That is either a reason it can't work, or a reason it hasn't been tried. This is the conversation that needs to be happening in public, now, while the choice is still open.
For journalists, researchers, and anyone who wants to poke holes
Every number links to a filing. The data is a single public file. The design is a public document. Take any of it, check it, use it, improve it. If you find an error, tell me and it will be fixed within the hour, publicly.
Take action The data (JSON) The full design ryan@wilde.energy
Known weak points, so you don't have to find them
- All three major AI labs today share a country, investors, and a hiring pool. That is one faction, not three.
- Correlated blind spots: AIs trained on similar data miss the same things. The citizen panels exist to catch this, and might not.
- The system guarantees compliance, not fairness. If people agree on bad rules, it will enforce bad rules.
- Capital can flow to countries that opt out. The only lever is market access: sell here, get audited here.
- Public ownership only works if the public watches. Apathy beats every other failure mode.
- The author is one person with a laptop, not an institution. That's either the weakness or the point.
Why Rebecca
This is named for Rebecca Sue Smith, my best friend, who taught me to look at how people are treated and refuse to call it normal. She isn't here to see it. She'd have wanted it built anyway. — Ryan Wilde
A ledger is where you write down what is owed. That seemed like the right thing to put her name on.