Who Owns Intelligence?

Welcome. Notes from the Field is a working notebook on human agency, company-building, and the changing value of intelligence. I hope to bring commentary on problems that are happening, focused on the journey over the polished end state.

This first edition explores ownership of intelligence and compensation.

When I left a role I had been in for ten years, I spent a few weeks rifling through the proverbial file cabinets, sorting out what context was still real and what had gone stale to prepare my transition.

After doing that, I built a number of AI skills and an old school, printed "Field Guide." Collectively, a blend of satire and very real context that together made up the digital footprint of a decade. The digital and physical ephemera of 10 years of learning. Distilled content, experiences, conversations, tests and failures. I joked that the artifacts should have come with a residual schedule. The joke worked because everyone understood the absurdity.

The challenge going forward is that the underlying logic is not absurd.

The lines we draw around what stays with the employee and what transfers to the company are inherited convention. Focused on leaving behind a computer, the physical assets, not a reasoned answer to what intelligence is worth or who it belongs to.

The judgment I built over ten years is mine. The frameworks I developed for diagnosing expansion problems, the ways of reading a user base that I refined across thousands of decisions, the institutional pattern recognition that took a decade to accumulate: I carried all of that out with me. The business got the documentation. The artifacts. But the underlying intelligence went home with me even when I tried my hardest to transfer it.

At the same time, the business will keep using what I built. The systems I designed. The teams I structured. The strategic decisions that are still playing out. For years after I left, the returns on my decade of contribution will keep arriving, and there was no mechanism connecting those returns to me.

That is not a complaint. It is an observation about a structural gap. Companies have instruments for capital and labor. They do not have one for intelligence.

There should be.

Companies Are Bundles of Intelligence

Every business is a bundle of intelligence. Not assets, headcount or code. Intelligence: the accumulated knowledge of how to serve a specific customer, solve a specific problem, and do it better than the alternative. When that intelligence exceeds the cost of deploying it, the business generates profit. When it doesn't, the business fails.

That's the theory of the firm. Capital allocates resources. Labor executes. But intelligence is what makes the whole thing worth more than the sum of its inputs. That model is inherited from centuries of work valued for its inputs, not its outputs. Humans built widgets and turned cogs.

We don't have a model for how this works in an era where computers act more like intelligent beings, where work is heavily measured and context digitized. I want to think about this from first principles.

Before you can think about how proceeds should be distributed, you have to be clear about who the intelligence belongs to.

Intelligence Is Rented, Not Sold

This is clear to me, and critical for our long-term viability as a healthy civic society. It is what differentiates the western liberal capitalist tradition from other approaches.

Intelligence belongs to the person who holds it. Not the company they work for, the institution that trained them, or the platform they use to express it. Intelligence is sovereign to the human. It lives in a person, travels with a person, and when that person leaves an organization, it leaves with them.

Yes, an employer keeps the documentation, the code, the outputs. Yes, a platform may keep the traces, the data, the actions. But the judgment, the pattern recognition, the accumulated understanding: that goes home with the employee every night and walks out the door when they resign.

The appropriate transaction, then, is rental.

Money for time becomes money for intelligence.

You rent your intelligence to an enterprise for the duration of your engagement. The enterprise combines your intelligence with others to build collective value that no single contributor could build alone. The proceeds flow back to contributors based on what each brought. When you disengage, the rental ends. The artifacts you left behind continue generating value, decaying over time as context goes stale and skills drift. You retain a residual claim proportional to what remains.

This is not how most people experience employment. Most people, in practice, sell their time and effort for a wage. They transfer it to an organization that owns the outputs and captures the returns. The contributor gets a wage that doesn't reflect the actual intelligence contributed.

Equity and Wages Are Too Blunt

The current system offers two options.

The first is cash, generally as W2 employment or increasingly 1099 (contractor) payment. You are compensated for your active participation. When you stop participating, compensation stops. The day you leave, the transfer ends. The system has no concept of residual value. The institutional knowledge you built, the processes you designed, the relationships you developed: the company keeps them. You keep nothing.

The second is equity. You own a percentage of the business in perpetuity. The ownership doesn't decay or adjust for how much you're currently contributing. The person who wrote the original product spec in 2015 owns the same percentage in 2025, regardless of whether they've touched the company since. Equity captures your contribution at a single point in time and locks it permanently.

Collectively, these represent compensation as a binary state. You either own or you don't. You're either employed or you're not. There is no middle, no decay, and no acknowledgment that intelligence contribution is continuous, partial, and persistent.

Both systems are wrong about how intelligence actually works.

The Half-Life of Contribution

Intelligence doesn't arrive all at once and then disappear. It compounds, and it depreciates.

A founder's early insight about the customer problem might define the company for a decade. An engineer's system architecture might remain the core structure of the product long after the engineer has left. A salesperson's relationship capital might keep generating revenue through referrals and reputation years after they've moved on.

But none of that value is permanent.

The founder's insight gets superseded as the market evolves. The architecture gets refactored. The relationships fade without maintenance.

Intelligence has a half-life, and that half-life is different for different kinds of intelligence, businesses, and periods of time.

The current system pretends this isn't true. Equity holders own forever. W2 employees own nothing after their last day. The reality, which both systems paper over, is that value from intelligence contribution is a curve: it peaks during active engagement, then decays at a rate determined by how transferable, documented, and durable the underlying knowledge is.

The next compensation model has to answer a simple question: what happens when your work keeps working after you leave?

What Intelligence Share Would Change

I'd like to propose a new model for compensation: Intelligence Share. Intelligence Share is a distribution model that follows that curve.

The core principle is simple: proceeds are distributed in proportion to the intelligence contributed to generating them, with active contribution weighted more heavily than historical contribution, and historical contribution decaying over time toward zero.

This changes several things at once.

It changes the incentive structure around documentation. Under equity and W2, undocumented knowledge is job security. The more of what you know lives only in your head, the harder you are to replace and the more leverage you have. The rational move is to hoard.

Under Intelligence Share, the opposite is true. Documented knowledge, systems you leave behind, skills you create, processes you write down: these are the artifacts that continue to generate residual value after your active engagement ends. If your contribution decays when you leave but your documentation persists, then the documentation is the thing worth building. The incentive flips from hoarding to sharing.

It changes the logic of partial participation as well. Today, the tax code and standard employment law recognize two modes: employed or not employed. The structures available for fractional engagement are awkward at best: consulting contracts, advisory arrangements, part-time employment. All slow-moving and still binary. Compensation for time or actions, not intelligence.

Intelligence Share makes partial participation the default logic. You contribute what you contribute. You receive proportional to that. If you disengage, your prior contributions decay on their natural curve. If you re-engage, you're contributing again. The model doesn't require you to fit into a binary category.

It changes who benefits from what you build. Right now, the primary beneficiaries of an employee's intelligence are the equity holders who own the company and the employer who captures the value of that intelligence as ongoing labor. The employee has a wage. Sometimes a small equity grant that vests over four years, regardless of when the actual value was created during that time.

Under Intelligence Share, the distribution follows the intelligence. If you built the thing that's generating the revenue, you receive a portion of that revenue, decaying appropriately over time. If you rent the core intelligence, meaning you use commoditized tools, licensed systems, or external expertise you don't own, there's less proprietary value to share in the first place.

The Hard Part: Attribution

That last point is the model's built-in diagnostic. It forces an honest answer to how much of the intelligence driving the business is actually owned versus rented.

A business built on top of off-the-shelf software, external consultants, and publicly available frameworks is renting most of its intelligence. The proceeds from that business are mostly a function of deployment and execution, not of proprietary knowledge. There's less to share because less is genuinely owned.

A business that builds proprietary systems, develops institutional knowledge that doesn't exist elsewhere, and creates durable IP is owning its intelligence. The proceeds reflect that. And the participants who built that intelligence have a legitimate claim to a portion of them that persists beyond their active engagement.

The hard part is attribution. Intelligence is rarely created alone. It is mixed with capital, timing, management, luck, and other people's work. Any Intelligence Share model would need to avoid turning every company into a compensation courtroom. That does not make the problem imaginary. It means the instrument has to be designed carefully.

This is not an argument for whether one model is better than another. Some businesses should rent their intelligence. It's often cheaper and faster. It's an argument that the distribution of proceeds should reflect the actual intelligence equation: what we built, who built it, and how much of it still matters today.

Why This Matters Now

We are in an era when intelligence is increasingly the only durable competitive advantage. Capital has been commoditized. Labor can be augmented or replaced by AI. Distribution is no longer controlled by incumbents with physical infrastructure. The thing that compounds, the thing that creates durable returns above the cost of capital, is organizational intelligence: the accumulated, specific, hard-to-replicate knowledge of how to do something better than anyone else.

Compensation instruments haven't kept up. Equity captures your contribution at the moment you sign and holds it permanently. W2 compensates active participation and stops the moment you don't. Both were designed for a world where value lived in capital and physical assets, not in knowledge that compounds over years and decays over decades. They get the attribution wrong, the timing wrong, and the incentives wrong.

Not because the people who designed them were careless; because they were built for a different problem.

The frontier labs are accelerating the gap. Through training data and user interactions, the labs are acquiring the accumulated intelligence of an era. When you use a product that improves by learning from your usage, you are not renting the model. You are selling it something. The problem is not that every data transaction is theft. The problem is that most people do not understand what they are transferring, how it compounds, or who captures the value. Your patterns of thought, your judgment, your ways of framing problems: these become inputs to a system owned by someone else, generating value you will never see.

Most businesses, if they asked that question clearly, would not like the answer.

I don't have the legal structure for Intelligence Share (yet). But I'm increasingly convinced the instrument matters. AI is making human intelligence easier to capture, copy, and redeploy. The companies that want durable trust with their contributors will need a better answer than wages, equity, and terms of service.

The joke I made on the way out is starting to feel less like a joke.