Intents

In 1961, the economist William Vickrey proposed an auction with a single unconventional rule: the highest bidder wins, but pays the second-highest price.

The effect was counterintuitive. In a standard auction, rational bidders shade their bids below their true valuation, because paying exactly what something is worth to you means you gain nothing from winning. But in Vickrey's design, the optimal strategy flips: you bid exactly what you believe the item is worth. Not less, not more. Since you'll only pay the second-highest price, there is no advantage to bidding below your true valuation and no advantage to bidding above it. The mechanism makes honesty optimal.

Vickrey had demonstrated something profound: the same participants, with the same preferences, reveal completely different information about what they want depending solely on the design of the system they operate within. The people don't change. The intent doesn't change. What changes is whether the mechanism is compatible with expressing it.

Leonid Hurwicz generalized this insight into a framework he called mechanism design. Traditional economics takes the market as given and asks how agents behave within it. Hurwicz reversed the question: if you want a specific outcome — efficiency, truthful revelation, fairness — what mechanism would produce it?

The central concept is incentive compatibility: a mechanism is well-designed when participants do best by acting on their true preferences. A well-designed auction makes honest bidding the dominant strategy. A badly-designed one encourages gaming: shading bids, sniping, collusion. The participants are the same. The preferences are the same. What changes is the mechanism, and the mechanism determines whether true intent surfaces or gets buried.

Hurwicz, along with Roger Myerson and Eric Maskin, received the Nobel Prize in Economics in 2007 for formalizing what, once stated, sounds obvious: the design of the system determines whether the people inside it can express what they actually want.

The Distance Between Intent and Outcome

Most systems in the modern world are not incentive-compatible with human intent.

Consider what happens when your intent is simple: save money for the future. The financial system does not ask what you want and optimize for it. Instead, it presents a maze of products — mutual funds with embedded fees, insurance wrappers that benefit the seller, tax-advantaged accounts with complex rules, advisory relationships where the advisor's compensation depends on which product they recommend. You must navigate this maze, hoping that the path you take through the intermediaries' structure approximates your original intent.

Or consider: find the answer to a question. A search engine does not simply answer it. It returns a page where your query competes with paid placements, algorithmic suggestions, and content optimized for ranking rather than accuracy. The mechanism is designed to serve the advertiser's intent, not the searcher's. You get results, but you must filter, evaluate, and verify — performing work that the system could do but won't, because doing so would conflict with its revenue model.

Or: send money to someone in another country. Your intent is a single transfer. What you get is a chain of correspondent banks, each taking a fee, each adding days of delay, each imposing compliance requirements designed for their own regulatory survival. The SWIFT network, as the Path Dependence essay described, routes your intent through 1970s infrastructure not because it serves you but because the infrastructure serves itself.

In each case, the gap between what you want and what you get is not a failure of information or technology. It is a feature of the mechanism. The intermediaries are not failing to serve your intent — they are successfully serving their own. And the mechanism makes this optimal for them.

This is the structural problem that the previous eight essays have been describing from different angles.

The Anacyclosis essay showed that political systems decay when they stop serving collective intent and begin serving the preservation of existing power. The Non-Delegation essay showed that delegating critical functions to intermediaries creates vulnerability precisely because the intermediary's intent diverges from yours. The Ads essay showed that attention-extracting platforms are mechanisms designed to subvert user intent for commercial capture. The Degrees of Freedom essay showed that regulation, intended to align systems with public intent, often constrains the dimensions along which intent can be expressed. The Path Dependence essay showed that systems lock in early choices, making it progressively harder to redirect toward current intent. The Chemoton essay showed that living systems must actively maintain alignment between their internal state and their environment — a continuous expression of biological intent. The Criticality essay showed that when the gap between a system's actual behavior and its participants' intent grows large enough, the correction is not gradual but catastrophic.

The common pattern is the accumulation of distance between intent and outcome. Systems begin as intent-serving mechanisms and evolve into intent-distorting ones, not through conspiracy but through the natural accumulation of intermediary objectives, regulatory constraints, and institutional inertia.

Declarative Systems

In computer science, a foundational distinction separates two ways of interacting with systems.

In imperative programming, you specify exactly how to produce a result: step one, step two, step three, in order, with every operation explicit. The programmer controls the path.

In declarative programming, you specify what result you want and the system determines how to produce it. SQL, the language for querying databases, is declarative: you describe the data you want, and the query optimizer decides how to retrieve it. You never specify which index to scan, which order to join tables, or how to allocate memory. You state your intent, and the system figures out the execution.

The declarative approach works because it separates what from how. The user's concern is the outcome. The system's concern is the path. When these are separated, the system can optimize the path in ways the user could never specify manually, using information about its own internal state that the user doesn't have and doesn't need.

This distinction maps precisely onto the problem of intermediation. In the current financial system, every transaction is imperative: you must specify the exact path through intermediaries, choose the specific instrument, route through the specific exchange, comply with the specific regulatory framework. You are forced to think in terms of the system's structure rather than your own intent.

An intent-centric system inverts this. You declare the outcome you want. The system competes to deliver it.

Intents in Practice

This inversion is already operational in programmable finance.

In traditional decentralized exchanges, a user specifies every parameter of a trade: the exact token pair, the exact smart contract, the maximum slippage, the deadline. This is imperative. The user must understand the system's internal structure to interact with it.

Intent-based protocols work differently. The user signs a message expressing what they want: "I have 1 ETH and want at least 3,000 USDC." This message is not a transaction. It is a declaration of intent. Specialized actors called solvers then compete to fulfill it, searching across multiple liquidity sources, exchanges, and execution paths to find the best outcome. The user does not specify how. They specify what. The solver that fulfills the intent most efficiently wins the right to execute it.

Protocols like UniswapX and the account abstraction standard ERC-4337 implement variations of this pattern. The details differ, but the architecture is consistent: separate the user's intent from the execution path, and let competition among solvers optimize the path.

This is Vickrey's insight applied to financial infrastructure. The mechanism is designed so that solvers profit by serving the user's intent as faithfully as possible. Gaming the system — front-running, sandwich attacks, information extraction — becomes structurally harder because the solver's compensation depends on outcome quality, not information advantage.

The pattern extends beyond simple trades. Lending intents: "I want to borrow $10,000 at the lowest rate available across any protocol." Portfolio intents: "Maintain this asset allocation and rebalance when drift exceeds 5%." Cross-chain intents: "Move these assets to the chain where my intent can be fulfilled most efficiently." In each case, the user expresses what they want, and the system's competitive dynamics optimize the how.

Agents as Intent Machines

The Criticality essay described how AI agents, operating autonomously, will naturally select the most efficient infrastructure for their transactions. This is because agents are, by their nature, intent machines.

A human navigating a financial system must understand interfaces, evaluate options, manage accounts, and track multiple intermediaries. Much of their effort goes into translation: converting their intent into the specific instructions each intermediary requires. An AI agent does not navigate interfaces. It expresses intent directly.

When an agent is instructed to "find the cheapest supplier for this component," it does not browse websites, compare shipping rates, and fill out purchase orders the way a human would. It expresses a structured intent — specification, quantity, delivery requirements, budget constraint — and routes it through whichever system can fulfill it most efficiently. If intent-based financial infrastructure exists, the agent will use it, because declaring what you want and letting solvers compete is objectively more efficient than manually navigating intermediary structures.

This creates a self-reinforcing dynamic. As more commerce is conducted by agents, the demand for intent-native infrastructure increases. As intent infrastructure improves, more agents route through it. The volume of intent-based transactions grows independent of individual human adoption decisions, because the efficiency advantage is clear and agents optimize by default.

This is why agentic commerce compresses the adoption timeline for intent-centric systems. Human adoption requires changing habits, learning new interfaces, and overcoming inertia. Agent adoption requires only that the system be more efficient. There is no switching cost for software, no cultural resistance, no institutional loyalty. There is only optimization.

Intent as Architecture

The implications extend beyond finance.

Governance as intent. Voting in representative democracy is a crude intent-expression mechanism: once every few years, you choose a representative, and that representative makes thousands of decisions on your behalf. The mapping from your intent to policy outcomes is noisy, delayed, and heavily intermediated by party structures, lobbyists, and bureaucratic translation.

Intent-based governance would let citizens express preferences on outcomes directly: "I value lower pollution more than I value cheaper energy." Mechanisms like quadratic voting, futarchy, and liquid democracy are attempts to build governance systems where the mechanism is incentive-compatible with expressing true preferences, rather than forcing citizens to compress their complex values into a single binary choice between candidates.

Information as intent. When you search for something, your intent is an answer. The current mechanism — ad-supported search — is incentive-compatible with serving advertisers, not searchers. An intent-based information system would let you express what you want to know and have competing systems provide the answer, compensated based on accuracy and relevance rather than click-through rates.

AI is already moving in this direction. When you ask a language model a question, you express intent directly, and the system attempts to fulfill it without routing you through a list of ad-supported links. The reason this feels different from traditional search is not that the technology is smarter — it is that the mechanism is more closely aligned with the user's intent.

Identity as intent. The Non-Delegation essay described portable, self-custodial reputation as a way to preserve exit rights. Intent-based identity goes further: instead of presenting a fixed credential package to every system — passport, credit score, employment history — you express what you need to prove: "I am over 21," "I can afford this loan," "I have five years of experience in this field." A zero-knowledge proof verifies the claim without revealing the underlying data. Your intent is verification. The system serves that intent without requiring you to expose everything else.

The Inversion

The common thread is an architectural inversion. In the current paradigm, systems are designed and people must navigate them. In an intent-centric paradigm, people express what they want and systems compete to serve them.

This inversion resolves a tension the Degrees of Freedom essay identified: centralized regulation cannot match the dimensionality of the systems it attempts to control. But intent-centric architecture doesn't require centralized regulation to achieve coordination. The coordination emerges from competition among solvers, each motivated to fulfill intents more efficiently than their competitors. The system achieves alignment not through constraint but through incentive compatibility.

This is why intent-centric systems can operate at higher dimensionality than rule-based systems. A rule constrains behavior along a known axis: "do not exceed this speed," "hold this capital ratio," "report this information." Rules are low-dimensional projections of high-dimensional intent. An intent-based system doesn't need to project. The intent itself is the specification, and the execution path can vary across as many dimensions as the system offers.

It also resolves the path dependence problem. In imperative systems, the execution path is fixed: you must use this exchange, this bank, this settlement network, because the transaction is specified in terms of their structure. In declarative systems, the path is flexible: you express what you want, and the system finds the best path at the moment of execution. If a better path exists tomorrow, tomorrow's transaction uses it automatically. You are never locked into a specific intermediary because you never specified one.

And it resolves the criticality problem. The Criticality essay argued that a parallel system could absorb a phase transition if it reached sufficient maturity before the crisis. Intent-based infrastructure is how that parallel system becomes usable at scale. The reason programmable money doesn't yet substitute for traditional finance for most people isn't primarily technological — it's that the current interfaces are still imperative. They require users to understand protocols, manage gas fees, navigate bridges, and specify exact execution paths. Intent-based layers abstract this away. The user doesn't need to know which chain they're on, which protocol they're using, or how the settlement works. They express what they want. The system delivers.

This is what makes the parallel infrastructure accessible to the people who will need it when the critical event arrives.

The Mechanism of the Second Enlightenment

The First Enlightenment answered a question about authority: who rules, and by what right? Its answer — consent of the governed, separation of powers, individual rights — was a mechanism design for politics. It was a system designed to make governance incentive-compatible with the public's intent, at least more so than divine-right monarchy.

But that mechanism was designed for a world of slow communication, geographically bounded coordination, and human-speed decision-making. It works through delegation: you delegate your intent to a representative, who delegates to a bureaucracy, which delegates to regulatory agencies. Each delegation adds distance between your intent and the system's behavior.

The Second Enlightenment does not abolish delegation. It makes it optional. It builds systems where you can express intent directly, where intermediaries must compete to serve it rather than capture it, and where the mechanism is designed so that honesty — expressing what you truly want — is the optimal strategy.

This is what the series has been building toward. Not a critique of existing systems for the sake of critique, but a description of the architectural principle that resolves their accumulated failures: systems should be designed around the intents of their participants, not around the structures of their intermediaries.

Vickrey showed that a single rule change in an auction can make honesty optimal. The challenge of the next era is to apply that insight at civilizational scale: to build financial systems where expressing your true economic intent is the dominant strategy, governance systems where expressing your true political preferences is meaningful, information systems where expressing your true question gets you a true answer, and identity systems where expressing what you need to prove doesn't require surrendering what you need to protect.

The technology to do this exists. Programmable money makes financial intents executable. Zero-knowledge proofs make identity intents verifiable. AI agents make intent expression scalable. Competitive solver networks make intent fulfillment efficient. The architecture is not theoretical. It is being built.

What remains is the transition — and the Criticality essay described those dynamics. The old system accumulates distortions. The new system accumulates capability. At some threshold, the cost curves cross. And the cooperative dynamics of intent-based systems — where each new participant makes the network more useful, which attracts the next participant — ensure that the transition, once it begins, is fast.

The grains are still falling. But for the first time, the question is not only when the pile collapses, but what is being built to catch what comes after. The answer is an architecture designed, from the ground up, around what people actually want.