Hernando de Soto

Foreword

By Edo Segal

The asset I never counted was the one I was standing on.

In Chapter 14 of *The Orange Pill*, I wrote about a developer in Lagos. I used her three times across the book — as proof that the floor was rising, that AI tools were democratizing who gets to build. I believed every word. I still do. The tools are real. The capability is genuine. I watched my own engineers in Trivandrum achieve things in days that would have taken months.

But de Soto broke something open for me. Not about the tools. About what surrounds them.

Hernando de Soto spent forty years asking a question that sounds simple and isn't: Why do the poor stay poor when they already possess assets? His researchers counted the informally held real estate of the developing world and arrived at a number north of nine trillion dollars — more than the combined stock market value of the twenty wealthiest nations. The houses were real. The businesses were running. The people were industrious and creative. None of it generated capital. Not because the assets lacked value, but because no formal system existed to recognize that value, represent it, and allow it to participate in the broader economy.

Dead capital. Functional, valuable, and economically inert.

I read that and felt the floor tilt.

Because I had been celebrating the collapse of the imagination-to-artifact ratio — the fact that anyone with an idea and Claude Code can now build a working prototype in hours — without asking the harder question. Can she deploy it on infrastructure she can afford? Can she reach customers through channels that see her? Can she protect what she built? Can the revenue flow through payment rails that actually reach her bank account — assuming she has one?

The tool crossed the gap. The institutions did not. And the distance between a working prototype and a livelihood is not a technology problem. It is exactly the kind of institutional problem de Soto has spent his career mapping.

This lens matters right now because the AI discourse is stuck inside a fishbowl. We measure capability. We celebrate access. We talk about democratization as though handing someone a powerful tool is the same as handing them a functioning economy. De Soto's framework exposes the invisible infrastructure that the developed world breathes without noticing — the property registries, the payment rails, the contract enforcement, the credit systems — and asks what happens when billions of newly empowered builders exist outside all of it.

The answer is dead intelligence. Real, brilliant, amplified by extraordinary tools, and locked out of the system that converts building into wealth.

That reframing changed how I think about the dams this moment requires.

-- Edo Segal ^ Opus 4.6

About Hernando de Soto

1941-present

Hernando de Soto (1941–present) is a Peruvian economist and the founder and president of the Institute for Liberty and Democracy (ILD) in Lima. Born in Arequipa, Peru, he studied in Switzerland and worked in international finance and trade before returning to Peru in the 1980s, where his research into the informal economy transformed the global debate on poverty and development. His landmark books *The Other Path: The Invisible Revolution in the Third World* (1986) and *The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else* (2000) argued that the persistence of poverty in developing nations stems not from a lack of assets or entrepreneurial energy but from the absence of formal representational systems — property titles, business registrations, enforceable contracts — that allow assets to function as capital. His concept of "dead capital," referring to the trillions of dollars in extralegal assets held by the global poor that cannot be leveraged, traded, or used as collateral, reshaped development economics and influenced property rights reform in dozens of countries. Named one of the five leading Latin American innovators of the century by *Time* magazine and one of the hundred most influential people in the world, de Soto continues to advocate for institutional frameworks — increasingly incorporating blockchain and digital infrastructure — that extend economic inclusion to the populations capitalism has historically excluded.

Chapter 1: The Mystery of Dead Intelligence

In the early 1980s, a team of researchers at the Institute for Liberty and Democracy in Lima, Peru, conducted an experiment that would reshape the global debate about poverty. They attempted to register a small garment workshop as a legal business, following every rule, filling out every form, standing in every queue. The process took 289 days. It required visiting multiple government offices, making dozens of trips, and paying fees that, for a person living at the median Peruvian income, represented thirty-one times the monthly minimum wage. At the end of those 289 days, the researchers had a registered business. They also had an answer to a question that had confounded development economists for decades: why do the poor remain poor despite being, by any observable measure, industrious, creative, and resourceful?

The answer was not cultural. It was not genetic. It was not a failure of character or ambition. The answer was architectural. The institutional infrastructure that allows assets to generate capital — property titles, business registration, enforceable contracts, access to credit — was designed, often unwittingly, to exclude the majority of the population. The poor possessed assets. Hernando de Soto's research team estimated that, across the developing world, the total value of real estate held but not legally titled by the poor exceeded $9.3 trillion — a figure larger than the total value of all companies listed on the stock exchanges of the twenty largest developed nations combined. These assets were real. They sheltered families, housed businesses, fed communities. But they could not serve as collateral for a loan. They could not be sold on an open market. They could not be leveraged, contracted upon, or accumulated across generations. They were dead capital: valuable, functional, and economically inert.

De Soto's framework, articulated most fully in The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else, rests on a single insight of remarkable precision: capital is not a thing. It is a process. It is the process by which an asset is represented in a formal system that captures its economic potential and makes that potential available for productive use. A house is a physical structure. A titled house is simultaneously a physical structure and a legal entity — an entry in a registry, a node in a network of enforceable claims, a thing that can lead what de Soto calls "a parallel life" as capital. The title does not change the house. It changes what the house can do. It allows the house to participate in transactions, relationships, and accumulation processes that the physical structure alone cannot access.

The absence of this representational system is not a minor inefficiency. It is the primary mechanism by which the majority of the world's population is excluded from the wealth-generating machinery of capitalism. The exclusion is not deliberate in most cases. It is structural. The systems were built by and for a narrow segment of the population, and they remain inaccessible to everyone else — not because the everyone-else lacks assets, but because the system lacks the capacity to recognize what they possess.

The central mystery of the AI economy in the twenty-first century is structurally identical.

Consider what happened in the winter of 2025, as The Orange Pill documents with considerable vividness. A threshold was crossed. The machines learned to speak in human language. The imagination-to-artifact ratio — the distance between a human idea and its realization as a working thing — collapsed to the width of a conversation. A person with an idea and the ability to describe it in natural language could now produce a working prototype in hours. Not a sketch. Not a plan. A functioning artifact: code that ran, interfaces that responded, logic that held up under testing.

This collapse is genuinely historic. De Soto's framework does not dispute it. When de Soto documented the assets of the global poor, he did not question whether those assets were real. He questioned whether the system existed that could convert them into capital. The same precision applies here. The intelligence is real. The tools are powerful. The prototypes work. The question is what happens next.

A prototype is not a business. An artifact is not capital. The gap between producing something and converting that production into sustainable economic value — value that compounds, that attracts further investment, that supports a livelihood, that builds over time into an economic foundation — is the gap that de Soto's entire career has been devoted to mapping. And it is precisely this gap that the triumphalist narrative of AI democratization, for all its genuine moral energy, consistently and dangerously overlooks.

The Orange Pill acknowledges this gap. To Edo Segal's considerable credit, the book does not pretend that tool access equals economic equality. He writes directly that the democratization is "real but partial," that inequalities of access, connectivity, and capital remain, that English-language bias shapes the tools, that institutional support is unevenly distributed. But the acknowledgment operates as a qualification — a responsible caveat appended to an argument whose center of gravity remains the exhilarating expansion of who gets to build. De Soto's framework inverts the emphasis. The caveat becomes the thesis. The partiality is not a footnote to the story of democratization. It is the story.

Billions of people possess intelligence, creativity, and the drive to build. Many of them now possess tools of extraordinary power. What they do not possess is the formal representational system that would allow their intelligence to generate capital. Their intelligence is dead capital in the same structural sense that the houses of Lima's informal settlements are dead capital: real, valuable, but locked outside the system that would allow it to do productive work.

The analogy requires precision to avoid the charge of metaphorical overreach. De Soto's dead capital is a specific economic phenomenon. A house without a title cannot serve as collateral, cannot be sold on an open market, cannot be used as a legal address for a registered business, cannot participate in the web of formal transactions that generates wealth in the developed world. Each of these incapacities is a specific institutional failure, not an atmospheric grievance. The deadness is architectural. It can be mapped, measured, and — de Soto has argued for forty years — remedied through specific institutional reforms.

Dead intelligence exhibits the same structural specificity. A developer in Lagos builds a working application with Claude Code. The application functions. It solves a real problem. It has potential users. But it remains dead intelligence unless a series of institutional conversions occur, each dependent on infrastructure that may or may not exist in her context. The application must be deployed — which requires cloud infrastructure she can afford, with latency acceptable to her users, with uptime guarantees that a free tier cannot provide. It must be discoverable — which requires marketplace access, search optimization, or distribution networks that connect her to potential users. It must be monetizable — which requires payment infrastructure that works across borders, in her currency, with fees that do not consume her margin. It must be protectable — which requires intellectual property frameworks that recognize her ownership and provide recourse if her work is copied. It must be sustainable — which requires either revenue sufficient to fund continued development or access to financing through credit, investment, or grants.

Each of these requirements is a component of the representational system that turns intelligence into capital. None of them is provided by the AI tool. All of them are necessary for the tool's output to generate lasting economic value. And their distribution across the global population is as unequal as the distribution of formal property rights that de Soto has spent his career documenting.

The parallel extends to the experiential reality of exclusion. De Soto has described what it feels like to stand outside the formal economy — to possess real assets, to work with extraordinary energy and ingenuity, and yet to remain economically invisible. The informal entrepreneur in Lima operates a business that serves real customers, employs real workers, and generates real revenue. But because the business is not registered, its revenue cannot be deposited in a bank. Because the entrepreneur's house is not titled, it cannot serve as collateral for a loan that would expand the business. Because the contracts with suppliers are not enforceable in court, every transaction carries a risk premium that eats into margins. The entrepreneur is productive. The system is blind to her productivity.

The extralegal AI builder experiences a structurally identical blindness. She builds a product that works, that users want, that solves a real problem. But the formal economy — the system of venture capital, enterprise procurement, app store distribution, and professional credentialing — does not see her. She has no corporate entity. She has no track record legible to investors. She has no compliance certifications, no audit trails, no security guarantees of the kind that enterprise customers require. Her product exists in a space that is technically sophisticated and institutionally invisible.

De Soto's most provocative claim about the developing world was that the poor are not the problem. The system is the problem. The poor possess assets, energy, and ingenuity in abundance. What they lack is a system designed to recognize and leverage what they possess. The institutional infrastructure of capitalism — the property registries, the contract law, the business registration systems, the credit mechanisms — was built over centuries in the West, through trial and error, through political struggle, through the gradual formalization of customs and practices that had previously operated informally. The West has forgotten how its own infrastructure was built, and therefore assumes that what the developing world lacks is the assets rather than the system.

The AI economy is reproducing this error at extraordinary speed. The assumption embedded in the democratization narrative — that providing tools is sufficient, that access to Claude Code is the relevant variable, that the imagination-to-artifact collapse is the story — is the same assumption that has haunted development economics for decades: that providing resources is sufficient without building the systems that allow resources to generate capital. The houses of Lima's informal settlements are not less valuable because they lack titles. They are less productive. The intelligence of the world's extralegal builders is not less real because it lacks institutional infrastructure. It is less generative. The asset is present. The capital process is absent. The intelligence is dead.

Building the representational infrastructure that turns dead intelligence into living capital is the central institutional challenge of the AI age. It is not a technology problem. The technology works. It is an institutional problem — a problem of systems, structures, registries, protections, and pathways that allow the output of human intelligence, amplified by AI tools, to participate in the formal economy. Without that infrastructure, the democratization of AI capability will produce what the democratization of assets without property rights has always produced: a world in which the majority possesses real value and the minority possesses the system that converts value into wealth.

De Soto's career has demonstrated that this outcome is not inevitable. The infrastructure can be built. Property rights can be extended. Formal systems can be designed to include rather than exclude. But the construction requires seeing the problem clearly, and seeing the problem clearly requires recognizing that the bottleneck is not the asset and not the tool but the system — the representational infrastructure that allows things to lead parallel lives as capital.

The mystery of dead intelligence is not why people lack ideas. It is why ideas remain dead — functional, valuable, and locked outside the system that would allow them to generate the economic life their creators deserve.

---

Chapter 2: The Bell Jar: Why Capability Without Infrastructure Remains Trapped

De Soto introduced the metaphor of the bell jar in The Mystery of Capital to describe a phenomenon that statistical measures of poverty consistently fail to capture. The bell jar is not a wall. A wall is visible. A wall announces its presence. You can see the wall, resent the wall, organize to tear the wall down. The bell jar is glass. You can see through it. You can observe, with perfect clarity, the formal economy operating on the other side — generating wealth, compounding advantage, converting assets into capital with a fluency that looks, from outside the glass, almost effortless. You can even imitate the motions. You can build a house that looks like the houses on the other side. You can start a business that functions like the businesses on the other side. But you cannot reach through. Your assets, however real, however functional, however indistinguishable in their physical qualities from the assets on the formal side of the glass, remain on your side. And on your side, they are dead.

The bell jar's cruelty is its transparency. If the barrier were opaque — if the excluded could not see the formal economy at all — the frustration would be different. It would be the frustration of ignorance, of not knowing what was possible. But the bell jar is transparent. The excluded can see exactly what is possible. They can see other people's houses serving as collateral, other people's businesses attracting investment, other people's contracts enforced by courts that treat their claims as legitimate. They can see the process by which assets become capital. They simply cannot access it.

De Soto documented this phenomenon across four continents. In Haiti, his team found that the assets held informally by the poor — untitled land, unregistered businesses, informal savings — totaled approximately $5.2 billion, which at the time was more than four times the total assets of all the formal businesses operating in the country combined. In Egypt, the figure was $240 billion — fifty-five times the total foreign direct investment recorded in the country's entire history. In every case, the pattern was identical: vast assets, fully visible to the people who held them, fully functional in their immediate use, and fully dead as capital.

AI has constructed a new bell jar with remarkable speed, and its transparency is, if anything, more tormenting than the original.

Consider the experience of the builder on the outside. She has access to the same AI tools — Claude Code, ChatGPT, open-source models that run on consumer hardware. The tools cost between zero and a hundred dollars per month. She can produce artifacts that are technically indistinguishable from those produced by builders in San Francisco, London, or Tel Aviv. The code compiles. The interfaces render. The logic holds. In terms of the artifact itself, the bell jar appears to have dissolved entirely. The imagination-to-artifact ratio has collapsed for her just as it has collapsed for the engineer at Google.

But the artifact is only the beginning. The conversion of artifact to capital requires traversing a series of institutional layers, each of which constitutes a pane of the bell jar. And these panes are arranged not as a single barrier but as concentric shells, each one filtering out a larger proportion of the world's builders.

The first pane is infrastructure access. Cloud deployment — the mechanism by which a local application becomes a globally accessible service — requires servers, bandwidth, and uptime guarantees that cost money. The pricing structures of major cloud providers are designed for the economics of the developed world. Amazon Web Services charges the same rate for a server in Lagos as for a server in Virginia, but the developer in Lagos earns a fraction of the developer in Virginia's income. A monthly cloud bill of two hundred dollars that represents a rounding error in a San Francisco startup's budget represents a significant financial commitment for an independent developer in Nigeria, Kenya, or Bangladesh. The free tiers that cloud providers offer are designed as on-ramps to paid services, not as sustainable infrastructure for building a business. They impose limits on traffic, storage, and compute that are incompatible with serving a real user base at scale.

The second pane is financial infrastructure. Accepting payment from international customers requires a payment gateway. Stripe, the dominant platform for online payments, operates in approximately forty-seven countries. The world has approximately 195 countries. The majority of the world's population lives in countries where the most widely used payment infrastructure is either unavailable, restricted, or burdened with fees and regulatory requirements that make small-scale commerce prohibitively expensive. A developer in Lagos who builds a software-as-a-service product and wants to charge ten dollars per month per user faces a conversion challenge that is not technical but institutional: how does the money get from the user's account to hers, across borders, currencies, and regulatory regimes, in a way that is reliable enough to sustain a business?

The third pane is intellectual property protection. The developer who builds something valuable needs to be able to protect it from copying. Intellectual property law varies dramatically across jurisdictions. Enforcement is expensive even in countries with robust legal systems and effectively impossible in countries without them. A developer in the informal AI economy who discovers that a larger company has replicated her product has limited recourse — not because the law does not nominally protect her, but because the cost of legal action exceeds the value of what she is protecting, and the enforcement mechanisms that would make protection practical do not extend to her context.

The fourth pane is market access. Discovery — the process by which potential users find a product — is dominated by platforms controlled by a small number of companies. App stores, search engines, social media algorithms, and enterprise procurement processes all function as gatekeepers, and their gatekeeping criteria are shaped by the assumptions and incentives of the developed world. A product built for a local market in West Africa may never appear in the discovery channels that global users rely on. A product built for a global market by a developer in West Africa may be disadvantaged by the reputational and credentialing filters that procurement processes impose.

The fifth pane is social capital — the networks of mentors, investors, collaborators, and early adopters that the technology industry calls "ecosystem." The Orange Pill acknowledges this directly: the developer in Lagos does not have "the same salary, not the same network, not the same institutional support, not the same safety net if the project fails." This acknowledgment is honest and important. But in the architecture of the book's argument, it functions as a limitation, a qualifier. In de Soto's framework, it is the primary mechanism of exclusion.

Access to social capital is not an add-on to the formal system. It is the formal system's nervous system. Venture capital does not flow through application forms alone; it flows through introductions, warm referrals, the trust networks that connect founders to funders. Enterprise customers do not find software through open searches alone; they find it through industry conferences, peer recommendations, analyst reports — channels that presuppose a level of institutional embeddedness that the extralegal builder, by definition, does not possess.

Each pane of the bell jar is individually surmountable. Developers from outside the formal ecosystem do break through — some through exceptional talent, some through fortunate connections, some through sheer persistence. But the bell jar's power is not in any individual pane. It is in the cumulative effect. Each pane filters out a percentage of the builders who encounter it. The product of those percentages is the effective exclusion rate, and it is high — high enough that the vast majority of builders outside the formal ecosystem will produce artifacts that function perfectly and generate no capital.

This is the structural reality that the democratization narrative, focused on the tool, fails to capture. The tool has been democratized. The system has not. And the system is where capital is born.

De Soto's work on property rights revealed a pattern that is directly instructive. In every country his team studied, the informal economy was not a marginal phenomenon. It was the majority economy. In Peru, approximately sixty percent of economic activity took place outside the formal legal system. In Egypt, the figure was similar. In the Philippines, sixty-five percent. These were not failed economies. They were functioning economies operating in parallel to the formal system — with their own rules, their own enforcement mechanisms, their own forms of trust and coordination. But they were invisible to the formal economy, and their invisibility meant that the wealth they generated could not compound.

The AI economy is developing its own informal majority. Millions of builders worldwide now create software, generate content, and construct products with AI tools outside the formal structures of employment, intellectual property, and corporate organization. They build on personal laptops. They deploy on free tiers. They sell through informal channels — direct messages, community forums, word of mouth. Their work is real and often valuable. But it exists in a space that the formal economy cannot see, and what the formal economy cannot see, it cannot invest in, insure, regulate, or protect.

The bell jar is not malicious. Nobody designed it to exclude. It emerged, as institutional barriers usually do, from the accumulation of systems built for one population and never extended to another. Cloud pricing was set for Western economics. Payment infrastructure was built for Western financial systems. Intellectual property law was drafted by Western legislatures. Discovery algorithms were trained on Western usage patterns. Each system is rational within its own context. Together, they constitute a bell jar that separates the world's builders into those whose intelligence can generate capital and those whose intelligence, however amplified by AI tools, remains economically inert.

The bell jar is not broken by providing better tools. It is broken by extending the formal system — the representational infrastructure that allows assets to lead parallel lives as capital — to include those it currently excludes. De Soto demonstrated this for physical property. The task now is to demonstrate it for intelligence.

---

Chapter 3: The Representational Gap in the AI Economy

The concept that unlocks de Soto's entire framework is representation. Not representation in the political sense — the right to vote, the right to be heard — but representation in the informational sense: the capacity of a formal system to capture the economic attributes of an asset and make those attributes available for transactions that extend beyond the asset's immediate physical context.

A house, in its physical form, performs one function: it shelters. A house represented in a formal property system performs dozens. The title allows the house to serve as collateral for a loan — not because the bank wants the house, but because the title converts the house into a legible, verifiable, transferable claim within a system the bank trusts. The same title allows the house to be sold to a stranger — someone who has never seen it, who lives in a different city, who trusts not the seller but the system that guarantees the seller's claim. The title allows the house to serve as a legal address, establishing the owner's location within a grid of enforceable rights and obligations. The title allows the house to be inherited, its value transmitted across generations through mechanisms that require no personal relationship between the giver and the receiver, only a formal system that both parties recognize.

None of these functions inhere in the physical house. They inhere in the representation. The house without the title performs one function. The house with the title performs many. The difference between the two — between a shelter and a capital asset — is entirely a product of the representational system.

De Soto estimated that the formal property systems of the developed world took between one hundred and two hundred years to build. They were not designed in a single act of legislation. They emerged through centuries of incremental formalization — the gradual conversion of informal customs, local arrangements, and extralegal practices into codified, standardized, enforceable systems. In the United States, the process involved the Homestead Acts, the development of title insurance, the creation of county recording offices, the standardization of surveying practices, and dozens of other institutional innovations, each one adding a layer to the representational infrastructure that allowed land to lead a parallel life as capital.

The developed world has forgotten this history. The formal property system is so deeply embedded in the institutional fabric of Western economies that it has become invisible — as invisible as the air in a fishbowl. When Western economists recommend property rights reform to developing countries, they often prescribe the endpoint — "establish a property registry" — without recognizing the centuries of institutional evolution that preceded the endpoint in their own countries. The recommendation is like telling someone to build a cathedral without explaining the invention of the flying buttress.

The AI economy has a representational gap of comparable magnitude, and the gap is widening faster than any institutional response can close it.

The code a developer writes with AI assistance is an asset. Like a house, it has immediate functional value: it performs a task, solves a problem, serves a user. But its capacity to generate capital — to attract investment, to sustain a business, to compound over time — depends on a representational system that allows the code to participate in transactions beyond its immediate functional context. This system has multiple components, each analogous to a component of the formal property system that de Soto has spent his career analyzing.

The first component is documentation and standardization. A house becomes legible to the formal economy when it is surveyed, measured, and described in standardized terms that any participant in the system can interpret. Code becomes legible to the formal economy when it is documented, version-controlled, and organized according to standards that potential collaborators, investors, and customers can evaluate. Version control platforms — GitHub, GitLab — serve a function analogous to the county recorder's office: they create a public, verifiable record of what exists, who created it, and how it has changed over time. But access to these platforms, and the literacy required to use them effectively, presupposes an institutional context that not all builders share. A developer working in an informal context, building with AI tools on a personal device, may produce code of extraordinary quality that has no formal representation whatsoever — no repository, no documentation, no version history, no public record of its existence.

The second component is licensing and intellectual property. A house becomes tradeable when the owner's claim is recognized by a system that can adjudicate disputes and enforce transfers. Code becomes tradeable when it is licensed — when a formal instrument specifies who owns it, under what conditions it can be used, and what rights the owner retains. The licensing infrastructure of the developed-world software economy is mature: open-source licenses, commercial licenses, software-as-a-service agreements, enterprise procurement contracts. But this infrastructure was built by and for the institutional participants of the developed-world economy. An extralegal builder who wants to license her code faces a thicket of legal complexity that assumes she has access to legal counsel, understands jurisdictional differences, and can afford to enforce her rights if they are violated. For most of the world's builders, these assumptions do not hold.

The third component is deployment infrastructure. A house generates economic value when it is connected to municipal services — water, electricity, roads, postal addresses — that allow it to participate in the broader economy. Code generates economic value when it is deployed on infrastructure — servers, networks, content delivery systems — that makes it accessible to users. As noted in the previous chapter, this infrastructure is priced for developed-world economics and unavailable or prohibitively expensive for a substantial majority of the world's builders. The deployment gap is not a technology gap. The technology exists. It is a pricing gap, a distribution gap, an institutional gap.

The fourth component is financial infrastructure. A house becomes capital when it can serve as collateral — when a financial institution recognizes its value and extends credit on the basis of that recognition. Code becomes capital when it can attract investment — when a financial system exists that recognizes the value of software assets and provides mechanisms for converting that value into the funding required to build a business. Venture capital, the dominant funding mechanism for technology startups, is concentrated to an extraordinary degree. In 2024, approximately fifty percent of global venture capital investment went to companies headquartered in the United States. The majority of the remainder went to China, the United Kingdom, India, and a handful of other countries. For builders in the vast majority of the world's nations, the venture capital system is as inaccessible as the formal property system is to the homeowner in a Lima shantytown.

The fifth component is marketplace infrastructure — the system that connects producers to consumers. A house participates in the real estate market when it is listed, valued, and presented to potential buyers through platforms that aggregate supply and demand. Code participates in the software market when it is listed in app stores, discovered through search engines, evaluated through reviews and ratings, and presented to potential users through channels they trust. These marketplace systems are controlled by a small number of companies — Apple, Google, Amazon, Microsoft — whose curation, pricing, and ranking decisions determine which products reach which users. For a builder outside the developed-world ecosystem, gaining meaningful visibility in these marketplaces is not merely difficult. It is structurally disadvantaged, because the algorithms that determine visibility are trained on usage patterns that reflect the preferences, behaviors, and contexts of the developed world's users.

Together, these five components constitute the representational system that turns code into capital. And the gap between builders who have access to this system and builders who do not is the representational gap of the AI economy.

The gap is not new. Every technology economy has had one. But AI has made the gap more consequential, because AI has democratized the one thing the gap does not govern: the capacity to produce. Before AI, the representational gap was partially masked by the production gap. Builders in the developing world faced barriers not only to capitalizing their work but to producing it in the first place. Limited access to education, tools, and infrastructure meant that the production capacity itself was unequally distributed. AI collapsed the production gap. It did not touch the capitalization gap. The result is a world in which the capacity to build has been dramatically equalized while the capacity to convert building into sustaining economic value remains as unequal as it has ever been.

This is the representational gap in its starkest form: the distance between what the world's builders can now produce and what the world's institutions can recognize, value, and support. Closing it requires not better tools — the tools are already extraordinary — but the construction of representational infrastructure that extends to the billions of builders who currently operate outside the formal system.

De Soto spent decades arguing that the construction of representational infrastructure is the single most impactful intervention available to economic development policy. Not aid. Not microloans. Not training programs. Infrastructure — the systems that allow assets to lead parallel lives as capital. The argument applies with equal force to the AI economy. The most impactful intervention is not building better AI tools. It is building the institutional infrastructure that allows the tools' output to participate in the formal economy — to be documented, licensed, deployed, financed, and discovered through systems that recognize the builder's ownership and protect her rights.

The representational gap is the defining structural challenge of the AI age. Everything else — the speed of the models, the sophistication of the interfaces, the beauty of the artifacts — is secondary to the question of whether the institutions exist that can convert those artifacts into capital.

---

Chapter 4: Extralegal Builders: The Informal Economy of AI Creation

In Lima, in the 1980s, de Soto's researchers counted the buildings. Not the buildings in the formal city — the titled, registered, municipally serviced buildings that appeared on official maps and in government records. They counted the other buildings. The ones that did not appear on any map. The ones built without permits, on land held without title, in neighborhoods that the formal city pretended did not exist. They counted 276,000 of them in Lima alone. They estimated that the construction represented approximately $8.3 billion in investment — investment made by families, block by block, room by room, over years and decades, with no institutional support and no formal recognition.

These were not shacks. Many were substantial, multi-story structures, built with the same materials and techniques as the formal buildings a few blocks away. The physical distinction between a titled house and an untitled house in Lima was often negligible. The institutional distinction was everything. The titled house was visible to the formal economy. The untitled house was a ghost — physically present, economically absent.

De Soto called the inhabitants of these invisible buildings "extralegal entrepreneurs" — a term chosen with care. Not illegal. Not criminal. Not marginal. Extralegal: operating outside the formal legal system not by choice but by exclusion, because the cost and complexity of entering the formal system exceeded what a person of ordinary means could bear. The 289-day registration process was not an anomaly. It was the system working as designed — designed for the formal sector, by the formal sector, with no consideration for the people standing outside its walls.

The extralegal economy was not a symptom of failure. It was an economy in its own right, with its own rules, its own enforcement mechanisms, and its own, remarkably effective, systems of trust. In the informal markets of Lima, contracts were enforced through reputation. Disputes were resolved through community mediation. Property boundaries were maintained by the social knowledge of the neighborhood — by the dogs that barked, as de Soto memorably put it, when a stranger crossed a line that no survey had drawn but everyone recognized. The extralegal economy functioned. It often functioned well. But it functioned within a ceiling imposed by its own informality. Transactions were limited to people who knew each other. Credit was limited to what personal trust could underwrite. Growth was limited to what could be achieved without the leverage that formal systems provide.

The AI economy has produced its own extralegal sector, and the speed of its emergence has no historical precedent.

Consider the landscape as of early 2026. Millions of individuals worldwide — the precise number is unknowable, which is itself diagnostic — are building software, generating content, creating products, and delivering services using AI tools outside the formal structures of employment, corporate organization, and institutional technology. They work on personal devices. They deploy on free tiers of cloud services, accepting the limitations — traffic caps, storage ceilings, the absence of uptime guarantees — as the cost of admission to a system that does not recognize them as legitimate participants. They sell through channels that would have been unrecognizable to the technology industry of five years ago: direct messages on social media, community forums, Telegram groups, informal referral networks.

Their products are not inferior. This is the point that must be emphasized, because the instinct of the formal economy is to equate informality with inadequacy. A developer in Nairobi who builds a mobile application with Claude Code may produce something technically identical to an application built by a developer at a funded San Francisco startup. The code quality may be equivalent. The user experience may be comparable. The underlying logic may be equally sound. The difference is not in the product. The difference is in the institutional context surrounding the product.

The Orange Pill celebrates several exemplars of this phenomenon. Alex Finn, who built a revenue-generating business over the course of 2025 using AI tools and determination, working 2,639 hours with zero days off, is held up as proof that the imagination-to-artifact ratio has collapsed and that individual builders can now achieve what previously required teams and institutional backing. The celebration is understandable. The achievement is real. But de Soto's framework forces a different set of questions about what Finn's example actually demonstrates.

Finn operated in a specific institutional context. He had access to reliable broadband. He had access to payment infrastructure that allowed him to monetize his product. He had access to app stores and marketplaces that provided distribution. He had the legal standing to register a business, open a bank account, and enter into contracts with customers. He had the social and cultural capital — English-language fluency, familiarity with Silicon Valley norms, visibility on platforms where the technology industry congregates — that allowed his work to be discovered, discussed, and celebrated. His 2,639 hours of labor were extraordinary. The institutional infrastructure that allowed those hours to generate capital was ordinary — ordinary in the sense that it is the default for builders in the developed world, so thoroughly embedded in the environment that it is invisible to those who inhabit it.

The question de Soto's framework demands is: what happens when the same 2,639 hours are invested by a builder who lacks that institutional context? The answer, documented across four continents and forty years of development economics research, is: the hours produce an asset, not capital. The product functions but cannot reach a market. The revenue, if any, cannot flow through formal channels. The business cannot be registered, insured, or invested in. The builder's achievement, however extraordinary by any measure of effort and ingenuity, remains economically invisible.

The extralegal AI economy is large, growing, and institutionally invisible in precisely this way. De Soto's researchers counted buildings. The AI economy's extralegal sector has no equivalent census. No institution is counting the developers who build with AI tools outside the formal economy. No registry documents their products. No database tracks their revenue. The invisibility is not an oversight; it is the definitional characteristic of extralegal activity. What the formal system does not recognize, it does not count. What it does not count, it cannot serve.

The extralegal AI economy has its own enforcement mechanisms, its own trust systems, its own social contracts — just as the extralegal economies of Lima, Cairo, and Manila did. Reputation in online communities functions as a form of social collateral. Code shared on GitHub, even without formal licensing, generates a kind of informal property claim that the community recognizes and, within limits, respects. Discord servers and Telegram groups serve as informal marketplaces where products are discovered, evaluated, and traded through mechanisms that rely on personal trust rather than institutional guarantee.

These mechanisms work. Within limits, they work well. But they operate under the same ceiling that de Soto identified in every extralegal economy he studied: transactions are limited to those who share the trust network, growth is limited to what personal reputation can underwrite, and the accumulated value of the builder's work cannot be leveraged, transferred, or compounded through formal mechanisms.

De Soto argued that the extralegal economy is not a problem to be eliminated but a resource to be formalized. The energy, ingenuity, and productivity of the informal sector are not failures of the system. They are evidence of what the system is failing to capture. The appropriate response is not to suppress informal activity — an approach that has been tried repeatedly and has failed everywhere — but to extend the formal system to include it, to build the institutional bridges that allow extralegal assets to enter the formal economy and begin leading the parallel life as capital that generates wealth.

The same principle applies to the extralegal AI economy. The millions of builders who create with AI tools outside the formal system are not a problem. They are the single largest untapped resource the global technology economy possesses. Their productivity is already real. Their products already function. What is missing is the institutional infrastructure that would allow their work to generate capital — the representational system that would make their products visible, their ownership recognized, their revenue formal, and their growth sustainable.

Building that infrastructure requires understanding what the extralegal builders actually need, which is different from what the formal economy assumes they need. De Soto's fieldwork repeatedly demonstrated that well-intentioned development programs fail when they design solutions based on the formal economy's assumptions rather than the informal economy's realities. Property titling programs that require documentation the poor do not possess. Business registration systems that impose compliance costs the informal sector cannot bear. Credit programs that demand collateral the untitled cannot provide. In every case, the institutional bridge was designed from the formal side — designed by people who could not see the bell jar they were looking through.

The AI economy's institutional bridges are being designed with the same blindness. Intellectual property frameworks that assume access to legal counsel. Financial infrastructure that assumes a bank account in a supported jurisdiction. Distribution platforms that assume English-language proficiency and familiarity with Western technology norms. Each of these assumptions is reasonable within the context of the formal economy. Each is exclusionary when applied to the extralegal builder who constitutes the majority of the world's AI-empowered workforce.

The extralegal AI economy is not a marginal phenomenon. It is the majority economy of AI creation, in the same way that the extralegal property economy was the majority economy of the developing world when de Soto began his research. The formal AI economy — the funded startups, the enterprise deployments, the credentialed developers working within institutional structures — is the visible tip. The extralegal economy, invisible to the formal system, uncounted by any census, operating on personal devices and free tiers and informal trust networks, is the body of the iceberg.

What de Soto's four decades of institutional analysis demonstrate is that the body of the iceberg is not inert. It is alive with economic energy that the formal system is failing to capture. The question is not whether to acknowledge the extralegal builders. It is whether to build the institutional infrastructure that allows their energy to generate the capital their societies desperately need. The answer will determine whether the AI revolution produces shared prosperity or a technologically sophisticated reproduction of the oldest exclusion in the history of capitalism: the exclusion of the productive many from the institutional systems that convert productivity into wealth.

Chapter 5: The Imagination-to-Artifact Ratio and the Property Rights It Requires

The collapse of the imagination-to-artifact ratio is the central technological claim of The Orange Pill, and it is correct. When Edo Segal describes a person with an idea and the ability to articulate it in natural language producing a working prototype in hours — code that compiles, interfaces that render, logic that holds under testing — he is describing something that has no precedent in the history of human tool use. The barrier between conception and creation, which has constrained human productivity since the first stone was chipped into a blade, has been lowered to a degree that warrants the language of revolution. De Soto's framework does not dispute this. It demands that we look at what the revolution has not changed.

There is a distinction embedded in the concept of the imagination-to-artifact ratio that the technological enthusiasm of the moment tends to elide, and it is a distinction with consequences measured in billions of lives. The imagination-to-artifact ratio measures the distance between an idea and a thing. The imagination-to-value ratio measures the distance between an idea and a sustainable economic outcome. These are not the same distance. The first is a technological distance, governed by the capability of tools. The second is an institutional distance, governed by the systems that convert things into capital. AI has collapsed the first. It has not touched the second. And the second is where the lives are.

Consider the sequence of events that must occur for an idea to generate lasting economic value. The idea must be conceived — a cognitive act. The idea must be articulated — a communicative act. The idea must be implemented — a technical act. And then, in a series of steps that the technology industry habitually compresses into the word "scaling," the implementation must be deployed, discovered, monetized, protected, maintained, and sustained over time — institutional acts, every one of them, dependent on infrastructure that exists independently of the tool that produced the implementation.

AI has revolutionized the first three steps. The fourth — the institutional sequence — remains governed by the same uneven infrastructure that de Soto has spent his career mapping. And the institutional sequence is where the value is created.

De Soto's research on property rights provides the most precise available framework for understanding why. In The Mystery of Capital, he identified six effects that a formal property system produces — six ways in which representation converts a physical asset into a generator of capital. Each of these effects has a direct analogue in the AI economy, and each reveals a dimension of the imagination-to-value gap that the imagination-to-artifact ratio does not capture.

The first effect is what de Soto called "fixing the economic potential of assets." A formal property system captures, in a standardized representation, the economically relevant attributes of an asset — its location, its dimensions, its ownership history, its encumbrances. This representation allows the asset's potential to be evaluated by parties who have never seen the asset itself, who may live in a different city or a different country, who trust not the owner but the system. Without this representation, the asset's potential is locked inside the local context — visible only to those who can physically inspect it, evaluable only by those who personally know the owner.

Code exhibits the same dynamic. A software product's economic potential — its market, its scalability, its revenue model, its competitive position — can be evaluated by investors, partners, and customers only to the extent that the product is represented in forms they can assess. A GitHub repository with clear documentation, version history, and licensing information is a representation that allows the product's potential to be evaluated at a distance. A product that exists only on a developer's laptop, undocumented, unlicensed, with no public record of its existence or provenance, has the same functional capability but none of the representational infrastructure that would allow its potential to be fixed — to be captured in a form that the formal economy can evaluate.

The second effect is "integrating dispersed information into one system." A formal property system creates what de Soto called an "information framework" — a single, authoritative record that integrates information about an asset from multiple sources: the surveyor's measurements, the recorder's documents, the tax assessor's valuations, the lender's liens. This integration is what allows the asset to participate in complex transactions. A buyer can assess a property's value, a bank can assess its risk, and an insurer can assess its vulnerability — all from the same integrated record, without needing to independently verify each component.

The AI economy has no equivalent integrated information system for the products of extralegal builders. A developer in Dhaka who builds a product with Claude Code may have fragments of representation scattered across multiple platforms — a repository on GitHub, a listing on a community forum, a payment link on a personal website, reviews from early users posted on social media. But these fragments are not integrated. No single system captures the product's provenance, ownership, technical specifications, user base, revenue history, and legal status in a form that a potential investor, partner, or enterprise customer could evaluate with confidence. The information exists. It is dispersed. And dispersed information, as de Soto demonstrated, is functionally equivalent to no information at all, because the cost of assembling it from scattered sources exceeds the benefit for any individual evaluator.

The third effect is "making people accountable." A formal property system attaches owners to assets in ways that create consequences for misuse. If a titled property owner defaults on a loan, the lender has recourse. If a registered business fails to deliver on a contract, the customer has legal standing. These accountability mechanisms are not punitive luxuries. They are the preconditions for trust between strangers — the trust that allows transactions to scale beyond the circle of people who know each other personally.

In the extralegal AI economy, accountability is enforced through reputation and social pressure — mechanisms that work within communities but do not scale. A developer who delivers a defective product to a customer in her community faces social consequences. A developer who delivers a defective product to a customer on the other side of the world faces none, because no institutional mechanism connects the transaction to enforceable consequences. The absence of formal accountability is not merely a risk for buyers. It is a ceiling for sellers. The developer who cannot offer institutional guarantees — a registered business, a contractual commitment, a legal jurisdiction for disputes — is excluded from the most valuable market segments: enterprise customers, government contracts, partnerships with established companies, all of which require accountability mechanisms that the extralegal builder cannot provide.

The fourth effect is "making assets fungible." A formal property system standardizes the representation of assets in ways that allow them to be compared, combined, and exchanged. A titled house in Lima can be compared to a titled house in Bogotá — not because the houses are physically similar, but because the representational system that describes them uses the same categories, the same standards, the same units of measurement. This fungibility is what creates markets. Without it, every transaction is a unique negotiation between parties who must independently assess the value of incommensurable assets.

Software products in the formal economy achieve a degree of fungibility through standardized interfaces — APIs, data formats, integration protocols — that allow products to interact with each other and with the broader technology ecosystem. Products in the extralegal economy often lack this fungibility. They are built to solve specific problems in specific contexts, without the standardization that would allow them to participate in the broader ecosystem. This is not because extralegal builders lack technical capability. It is because standardization requires awareness of standards, access to documentation that describes them, and the institutional incentive to adopt them — all of which are products of participation in the formal system.

The fifth effect is "networking people." A formal property system creates connections between people who would otherwise have no basis for interaction. The buyer who finds a property through a listing, the bank that extends a mortgage to a stranger, the insurer who covers a building it has never inspected — all of these connections are mediated by the representational system. The system creates the network. Without it, the network is limited to people who already know each other, and growth is limited to what personal networks can reach.

The sixth effect is "protecting transactions." A formal property system provides the enforcement mechanisms — courts, arbitration, insurance — that give parties confidence that the terms of a transaction will be honored. Without these protections, transactions carry a risk premium that suppresses economic activity. The informal entrepreneur who sells goods without a contract charges more or sells less, because the absence of enforcement raises the cost of trust. The extralegal AI builder who sells software without institutional protections faces the same calculus.

These six effects — fixing potential, integrating information, creating accountability, enabling fungibility, building networks, and protecting transactions — are the mechanisms by which a property system converts dead capital into living capital. They are not incidental to capitalism. They are the machinery of capitalism. And every one of them depends not on the asset itself but on the representational system that surrounds it.

The Orange Pill describes the moment when a student in Dhaka can access "the same coding leverage" as an engineer at Google. The claim is precise about the tool and silent about the system. The student in Dhaka can produce the same artifact. She cannot fix its economic potential, because no integrated information system captures her product's attributes in a form the formal economy can evaluate. She cannot benefit from accountability mechanisms, because no institutional framework connects her to enforceable commitments. She cannot make her product fungible, because she may lack access to the standards and protocols that allow products to interoperate. She cannot network through the formal system, because the formal system does not see her. She cannot protect her transactions, because the enforcement mechanisms that protect transactions in the formal economy do not extend to her context.

The imagination-to-artifact ratio has collapsed for her. The imagination-to-value ratio has not. And the distance between the two — the institutional distance that separates an artifact from capital — is precisely the distance that de Soto has spent forty years measuring, mapping, and proposing to close.

Closing it requires building property rights for the AI age — not metaphorical property rights, but specific, implementable institutional mechanisms that perform the six functions de Soto identified for the products of intelligence. Digital registries that fix the economic potential of AI-built products. Information frameworks that integrate dispersed data about provenance, ownership, and quality. Accountability systems that connect builders to enforceable commitments. Standards that make products fungible across contexts. Networks that connect extralegal builders to the formal economy. Enforcement mechanisms that protect transactions between parties who do not know each other and may never meet.

These are not futuristic proposals. They are the institutional building blocks that the developed world's technology economy already possesses, in fragmentary and imperfect forms, for its own participants. The task is not invention. It is extension — the deliberate, systematic extension of representational infrastructure to the billions of builders whom the current system excludes.

The imagination-to-artifact ratio tells us what the tools can do. The imagination-to-value ratio tells us what the institutions allow. The gap between them is the measure of dead intelligence — intelligence that functions but does not generate capital. Closing the gap is not a technology problem. It is the institutional challenge of the century.

---

Chapter 6: The Formal System That Turns Code into Capital

In the early nineteenth century, the United States faced a property crisis that bears an uncanny structural resemblance to the AI economy's current institutional deficit. As settlers pushed westward, they occupied land, built homes, cultivated fields, and established communities — all without formal legal title. They were, in de Soto's terminology, extralegal entrepreneurs: productive, invested, and institutionally invisible. The land they occupied was real. The improvements they made were valuable. But the federal government's land-management system — designed for orderly, sequential disposition of territory — could not keep pace with the reality of occupation.

The result was a continent-sized economy of dead capital. Millions of acres held and improved by settlers who could not prove ownership, could not sell their improvements on an open market, could not use their land as collateral, and lived under the constant threat that a better-connected claimant would arrive with a formal title and dispossess them. The settlers were not idle. They were not unproductive. They were locked out of the system that would have allowed their productivity to compound.

The resolution came not through a single legislative act but through decades of institutional innovation. The Pre-emption Act of 1841 recognized the rights of settlers who had already occupied land. The Homestead Act of 1862 created a formal pathway from occupation to ownership. County recording offices were established to maintain public records of title. Title insurance emerged as a private-sector mechanism for managing the risk of disputed claims. Surveying standards were developed to create the common measurement language that allowed parcels to be compared, combined, and traded. Mortgage lending, built on the foundation of verifiable title, created the credit system that allowed settlers to invest in improvements they could not otherwise afford.

Each of these innovations was a component of the formal system that turned land into capital. None of them changed the land itself. The fields were the same fields before and after the Homestead Act. What changed was the institutional wrapper — the representational infrastructure that allowed the land to lead the parallel life that generates wealth.

The formal system that turns code into capital is younger, less complete, and far more unevenly distributed than the formal system that turns land into capital. But its components are identifiable, its functions are analyzable, and its gaps — particularly the gaps that exclude the world's extralegal AI builders — are mappable.

The first component of the formal system is version control and provenance. Git repositories, particularly those hosted on platforms like GitHub, serve a function analogous to the county recorder's office: they create a public, timestamped, verifiable record of what was created, by whom, and when. This record establishes provenance — the chain of authorship that supports ownership claims. A developer with a well-maintained GitHub repository can demonstrate, to potential investors, partners, or courts, that she created the code at a specific time and has maintained it through a documented series of modifications. This provenance is not a luxury. It is the foundation on which every subsequent institutional function rests. Without it, the developer's claim to ownership is a personal assertion, not an institutional fact.

But version control platforms, though technically accessible to anyone with an internet connection, presuppose a set of practices, norms, and literacies that are products of participation in the formal technology ecosystem. Meaningful use of Git requires understanding branching, merging, commit conventions, and the social practices of collaborative development. Documentation standards — README files, API descriptions, contribution guidelines — are legible to participants in the formal ecosystem and opaque to those outside it. The platform is open. The literacy required to use it as a representational system is not equally distributed.

The second component is licensing infrastructure. Software licenses are the title deeds of the code economy. They specify who owns the code, under what conditions it can be used by others, and what rights the owner retains. The licensing ecosystem of the formal technology economy is mature and varied: permissive licenses like MIT and Apache that allow broad reuse, copyleft licenses like GPL that require derivative works to be similarly licensed, proprietary licenses that restrict use to paying customers, and the complex landscape of enterprise licensing agreements that govern the commercial software market.

For a developer within the formal ecosystem, choosing and applying a license is a standard part of the development process, supported by institutional knowledge, legal precedent, and tooling that automates much of the work. For a developer outside the formal ecosystem, the licensing landscape is a thicket of legal complexity that assumes access to resources she may not possess: legal counsel who understands software licensing, knowledge of jurisdictional differences in intellectual property law, and the financial capacity to enforce her rights if they are violated. The license exists as a legal instrument. The institutional context that makes it effective — the courts, the lawyers, the precedents, the enforcement mechanisms — is unevenly distributed.

The third component is deployment and distribution infrastructure. The mechanisms by which software reaches users — cloud hosting, app stores, package managers, content delivery networks — constitute the distribution system of the code economy. This system is controlled by a small number of companies whose platforms serve as the gatekeepers between producers and consumers. Apple's App Store and Google's Play Store together mediate access to approximately 99 percent of mobile users worldwide. Amazon Web Services, Microsoft Azure, and Google Cloud together host the majority of the world's cloud-deployed software. These platforms are not neutral conduits. They impose requirements — technical, financial, and compliance-related — that function as filters, admitting builders who meet the requirements and excluding those who do not.

The requirements are rational within the context of the platforms' business models. App store review processes protect users from malicious software. Cloud service pricing reflects the real cost of infrastructure provision. Compliance requirements respond to genuine regulatory obligations. But each requirement, however individually reasonable, adds a layer to the formal system that the extralegal builder must navigate without the institutional support — the legal teams, the compliance officers, the financial resources — that formal-sector participants take for granted.

The fourth component is financial infrastructure. The mechanism by which software generates revenue — payment processing, subscription management, invoicing, tax compliance — is as essential to the conversion of code into capital as the mechanism by which land generates rent. Stripe, PayPal, and their competitors provide the payment rails along which the software economy's revenue flows. These rails are built for the formal economy. They require bank accounts in supported jurisdictions, identity verification that assumes government-issued documentation, and compliance with know-your-customer regulations that vary by country and change frequently. For a developer in a supported jurisdiction with a bank account and a government ID, onboarding to a payment platform is a minor administrative task. For a developer in an unsupported jurisdiction, or in a supported jurisdiction without a bank account, or with documentation that does not conform to the platform's verification requirements, the payment rails are inaccessible. The software works. The money cannot flow.

The fifth component is the network of trust and credentialing that de Soto identified as the mechanism by which formal property systems connect strangers. In the formal technology economy, this network is maintained through a combination of institutional affiliations, professional credentials, and reputational signals. A developer who works for a recognized company carries that company's reputation. A graduate of a recognized program carries that program's credential. A contributor to a prominent open-source project carries that project's social capital. These signals allow strangers to assess competence and trustworthiness without personal acquaintance — the same function that a property title performs when it allows a bank to extend a mortgage to a borrower it has never met.

Extralegal builders lack these institutional signals. Their competence may be real — demonstrated in the code they produce, the products they ship, the users they serve. But competence that is not represented in the formal signaling system is invisible to the formal economy's evaluative mechanisms. The investor who evaluates startups, the enterprise customer who evaluates vendors, the partner who evaluates collaborators — all rely on institutional signals that the extralegal builder cannot provide. Not because she lacks competence, but because she lacks the representational infrastructure that makes competence legible to strangers.

The cumulative effect of these five components is the formal system that turns code into capital. The system does not create the code. The system does not improve the code. The system does not make the code more functional or more elegant or more useful. The system does something that is, in economic terms, more important than any of those things: it allows the code to lead a parallel life. To serve simultaneously as a product and as a claim. To be not only functional but recognizable, not only useful but tradeable, not only valuable but leverageable.

Without the system, the code is dead intelligence. With the system, the code is capital — an asset that can attract investment, generate sustainable revenue, compound over time, and create the economic foundations for further building.

The developer in Lagos who builds a working product with Claude Code has produced an asset. The question is not whether the asset is real. It is whether the formal system exists that can convert it. In most of the world, for most of the world's builders, the answer remains no. And that answer is not a technology problem.

It is a property rights problem, recognizable to anyone who has studied why the poor remain poor despite possessing trillions of dollars in assets. The assets are there. The system is not.

---

Chapter 7: The Dogs That Bark: Why Local Knowledge Matters in the AI Age

When de Soto's researchers arrived in the informal settlements of Lima to map property boundaries, they discovered something that confounded their surveying instruments. The boundaries were already known. Not formally — there were no deeds, no registry entries, no cadastral maps. But informally, with a precision that rivaled anything the formal system could produce, the community knew exactly who owned what. The knowledge was maintained not in documents but in social memory — in the collective awareness of neighbors who had watched each house being built, room by room, over years, who knew which family had cleared which plot, who remembered the handshake agreements and the verbal contracts and the informal transactions that had, over decades, created a property system as detailed as any county recorder's office, but entirely invisible to the formal economy.

De Soto captured this phenomenon with a metaphor that has become one of his most cited observations. In the informal settlements, the dogs barked. When a stranger crossed an invisible boundary — a boundary that appeared on no map but was understood by everyone who lived there — the neighborhood dogs announced the intrusion. The dogs knew the territory. The community knew the territory. The formal system did not.

This was not a quaint detail. It was the central insight that separated successful property rights reform from the long, expensive history of failed reform. Every attempt to formalize property in the developing world that ignored local knowledge — that imposed a registration system designed in a government ministry without consulting the communities whose property it purported to register — had failed. The failures were not failures of will or funding. They were failures of representation. The formal system did not capture what the community already knew, and a formal system that does not represent reality is worse than useless: it creates a second, competing version of reality that the community neither trusts nor obeys.

De Soto's solution was to work from the informal system outward — to begin with what the community already knew and build the formal representation on top of it. This meant sending researchers not into government archives but into the settlements themselves, interviewing families, tracing the history of each parcel, documenting the informal transactions and verbal agreements that constituted the community's property system. The formal title, when it was eventually issued, represented not an imposition from above but a recognition from below — a formalization of what already existed.

This principle — that institutional infrastructure must be built with local knowledge, not imposed upon local contexts — applies to AI democratization with a directness that the technology industry, with its structural preference for universal, scalable solutions, is poorly equipped to appreciate.

The AI tools that are reshaping the global economy were built in a specific context. They were built in the United States, by companies headquartered in San Francisco and its adjacent research institutions, by teams whose composition, training, and daily experience reflect the assumptions and priorities of the American technology sector. The training data that shaped these tools is predominantly English-language text drawn from the internet — a corpus that, despite its vastness, represents a specific and partial slice of human knowledge, weighted heavily toward the concerns, vocabularies, and problem-framing conventions of the English-speaking world.

This is not a moral failing. It is a structural fact. Training a large language model requires data, and the data that exists in accessible, digitized, processable form reflects the populations and institutions that have had the resources to digitize their knowledge. The model learns what the data teaches. And the data teaches, disproportionately, the problems, solutions, and frameworks of the developed world.

The consequences are subtle but consequential. When a developer in Lagos describes a problem to Claude Code, the tool responds with extraordinary capability — but the response is shaped by a training corpus that may not include the specific context of her problem. The local payment systems her users rely on, the regulatory requirements of her jurisdiction, the infrastructure constraints that shape her deployment options, the cultural conventions that determine what her users expect from a digital interface — all of these contextual factors may be underrepresented or entirely absent from the model's training data.

The tool is not broken. It is decontextualized. It produces responses that are technically competent and contextually incomplete — responses that reflect the problems and solutions of the environments represented in the training data rather than the environments where the developer actually works. This decontextualization is the AI equivalent of a property registration system designed in a government ministry without consulting the communities it purports to serve. It works on its own terms. It fails on the community's terms.

The problem extends beyond training data to the design assumptions embedded in the tools themselves. AI coding assistants are optimized for workflows that assume reliable broadband, continuous connectivity, desktop or laptop hardware, and the development environments standard in Western technology companies. A developer who works primarily on a mobile device — a reality for a significant proportion of developers in Africa and South Asia — encounters tools that were not designed for her context. A developer whose internet connection is intermittent faces tools that assume the ability to make continuous API calls to cloud-based models. A developer who thinks and communicates primarily in a language other than English faces tools whose capabilities degrade outside the languages best represented in the training data.

Each of these mismatches is individually surmountable. Developers in constrained environments are, by necessity and ingenuity, expert at working around limitations. But the cumulative effect of the mismatches is a tax — a friction surcharge imposed on every builder whose context diverges from the context the tools were designed for. And this tax, invisible to those who do not pay it, is a component of the bell jar that separates the world's builders into those for whom AI tools work seamlessly and those for whom the tools work, but at a cost.

The dogs-that-bark principle prescribes a specific remedy: build the institutional infrastructure with local knowledge, not upon local contexts. In the domain of AI tools, this means several things that the technology industry is beginning to recognize but has not yet acted on at scale.

It means training data that represents the problems, languages, and contexts of the developing world — not as an afterthought or a diversity initiative but as a core component of the model's capability. The problems a developer in Nairobi needs to solve — mobile money integration, low-bandwidth optimization, localized user interface conventions, the specific regulatory and cultural contexts of East African commerce — are not edge cases. They are the problems of the majority of the world's population. A model that cannot address them competently is not a universal tool. It is a local tool with global distribution.

It means interfaces designed for the hardware and connectivity environments where the majority of the world's developers actually work. Mobile-first development environments. Offline-capable coding assistants that can function during the intermittent connectivity that characterizes internet access in much of the developing world. Lightweight models that can run on consumer hardware without requiring continuous cloud access. These are not accommodations for marginal users. They are design decisions that determine whether the tool serves the majority or the minority of the world's builders.

It means community-driven customization — mechanisms that allow local developer communities to adapt, fine-tune, and extend the tools for their specific contexts. De Soto's property reform succeeded when it incorporated the community's existing knowledge into the formal system. AI tool development will succeed in reaching the global majority only when it incorporates the global majority's existing knowledge into the tools. This requires not merely open-source models, though open source is a necessary condition, but the infrastructure — compute resources, training pipelines, documentation, and support — that allows communities to make open-source models their own.

The technology industry's instinct is to build universal solutions. A single model, trained on a single corpus, deployed through a single interface, serving the entire world. This instinct is understandable — scale is how technology companies generate returns, and fragmentation is the enemy of scale. But de Soto's forty years of fieldwork demonstrate that universal solutions imposed without local knowledge reproduce the exclusions they claim to remedy. The property registration system that works in the government ministry fails in the settlement because it does not represent what the community already knows. The AI tool that works in San Francisco underperforms in Lagos because it does not represent what the developer in Lagos already needs.

The dogs bark in the informal AI economy, too. The local developer communities know their contexts — their infrastructure constraints, their user expectations, their regulatory environments, their linguistic and cultural frameworks — with a precision that no San Francisco design team can replicate from a distance. The institutional infrastructure for the AI age must be built from these local knowledges outward, not from Silicon Valley assumptions inward.

This is not a plea for fragmentation. It is a plea for the kind of institutional design that de Soto demonstrated works: design that begins with what already exists, that respects the knowledge embedded in informal systems, and that builds formal infrastructure on a foundation of local reality rather than external assumption. The dogs know the boundaries. The formal system must learn to listen to the dogs.

---

Chapter 8: The Fishbowl of the Developed World

The fishbowl, as The Orange Pill defines it, is the set of assumptions so familiar that one has stopped noticing them. The water one breathes. The glass that shapes what one sees. Everyone is in one. The powerful think theirs is bigger. Sometimes it is. It is still a fishbowl.

De Soto's career can be understood as a sustained effort to make visible the fishbowl of Western capitalism — to show the inhabitants of the developed world the water they have been breathing without noticing. The formal property system, the business registration system, the contract enforcement system, the credit system — these are the water. They are so deeply embedded in the institutional fabric of Western economies that their inhabitants cannot see them any more than a fish can see the ocean. They form the invisible substrate on which every economic transaction rests, and their invisibility is precisely what makes them powerful and precisely what makes their absence in the developing world so difficult for Western economists and policymakers to comprehend.

When a Western economist looks at the developing world and sees poverty, the economist's fishbowl shapes what the economist thinks the problem is. The problem appears to be a lack of resources — a lack of capital, a lack of technology, a lack of education, a lack of the inputs that the developed world possesses in abundance. The standard prescription follows logically from the diagnosis: provide the resources. Send aid. Transfer technology. Build schools. The assumption is that the developing world lacks the things and that providing the things will produce the outcomes.

De Soto demonstrated that this diagnosis is wrong. The developing world does not lack assets. It possesses trillions of dollars in assets. What it lacks is the institutional infrastructure that allows assets to generate capital — the representational system that the developed world built over centuries and has since forgotten it built. The prescription that follows from this revised diagnosis is categorically different from the standard one: do not provide things. Build systems. Do not transfer assets. Construct the institutional infrastructure that allows existing assets to generate economic life.

The AI democratization narrative is operating inside the same fishbowl, and making the same diagnostic error, with remarkable precision.

The narrative goes like this: AI tools have collapsed the imagination-to-artifact ratio. Anyone with an idea and the ability to describe it can now build a working prototype. The developer in Lagos has access to the same tools as the engineer in San Francisco. Therefore, the barrier to participation in the technology economy has been dramatically lowered. The democratization is real. The future is bright.

Each individual statement in this narrative is true. The conclusion drawn from them is incomplete in ways that de Soto's framework makes visible.

The narrative measures the tool. It does not measure the system. It sees the capability — correctly — and assumes the capability is the relevant variable. It does not see the institutional infrastructure that converts capability into capital, because that infrastructure is, for the narrators, invisible. It is the water they breathe.

Consider the specific assumptions that the AI democratization narrative carries without examining.

Assumption one: reliable electricity. The narrative assumes that the builder has consistent access to the electrical power required to run a computer, maintain an internet connection, and interact with cloud-based AI tools for sustained periods. For builders in the developed world, this assumption is so thoroughly satisfied that it does not register as an assumption at all. For builders in large parts of Sub-Saharan Africa, South Asia, and other regions, electricity is intermittent, expensive, and unreliable. A coding session interrupted by a power outage is not a minor inconvenience. It is a loss of context, a disruption of flow state, and, if work is not saved, a loss of output. The assumption of reliable electricity is invisible to those who have it and constitutive of the experience for those who do not.

Assumption two: affordable broadband. AI tools, particularly cloud-based large language models, require internet connectivity of sufficient bandwidth and reliability to sustain real-time interaction. The cost of broadband, measured not in absolute terms but relative to local income, varies by orders of magnitude across the world. A hundred-megabit connection that costs fifty dollars per month in the United States represents a trivial expense for a technology worker earning a median salary. The same connection in Nigeria, where it may cost a comparable absolute amount but where the median income is a fraction of the American figure, represents a substantial investment. And in many regions, the connection is not merely expensive but unavailable — the physical infrastructure does not extend to the builder's location.

Assumption three: financial infrastructure. The narrative assumes that the builder can pay for AI tools, cloud services, and other inputs through the financial systems that the technology economy relies on. Credit cards, bank transfers, subscription billing — these are the mechanisms through which the formal technology economy transacts. They require a bank account in a supported jurisdiction, a payment method recognized by the platform, and compliance with identity verification requirements that assume government-issued documentation of a specific type. The approximately 1.4 billion adults worldwide who remain unbanked, according to the World Bank's most recent estimates, are excluded from these mechanisms not by choice but by the absence of the institutional infrastructure that banking requires.

Assumption four: intellectual property protection. The narrative assumes that the builder's work, once created, is protected by a legal framework that recognizes her ownership and provides recourse if her work is copied. Intellectual property law exists, in some form, in nearly every country. Enforcement of intellectual property law is a different matter entirely. In countries where the judicial system is slow, expensive, or unreliable, the practical protection available to an individual builder — particularly one without access to legal counsel — approaches zero. The protection exists on paper. It does not exist in practice.

Assumption five: English-language fluency. The tools are built by American companies. The documentation is in English. The community forums, tutorials, and support resources are predominantly English-language. The training data is weighted toward English. A builder who is fluent in English navigates this environment without friction. A builder who is not fluent in English encounters a language barrier at every turn — not an insuperable barrier, but a tax, a friction surcharge that adds to the cumulative cost of participation.

Assumption six: social capital. The narrative assumes that the builder, having created a product, can reach users, attract attention, and build the relationships that convert a prototype into a business. But the channels through which attention flows in the technology economy — venture capital networks, technology media, conference circuits, professional communities — are concentrated in a small number of geographic and institutional centers. A builder who is not embedded in these networks faces a discovery problem that no amount of product quality can solve. The product may be excellent. If no one with the institutional power to amplify it knows it exists, the product remains invisible.

Each of these assumptions is a pane of the fishbowl. Each is invisible from the inside. Each is a barrier from the outside. Together, they constitute the analytical blind spot that de Soto identified in Western economic thinking about the developing world: the inability to see one's own institutional infrastructure, and therefore the inability to understand what its absence means for those who lack it.

The Orange Pill's Foreword includes a passage of admirable honesty: "We are all swimming in fishbowls. The set of assumptions so familiar you've stopped noticing them." The passage acknowledges the fishbowl without breaking out of it. The democratization argument that follows, despite its genuine moral seriousness, remains inside the glass. It measures the tool — correctly — and underweights the system. It sees the capability — correctly — and underweights the infrastructure. It celebrates the floor rising — correctly — without fully reckoning with the distance between the floor and the ceiling that the institutional infrastructure determines.

De Soto's contribution to the AI discourse is not a counter-argument. It is a lens adjustment. The tool is real. The capability is real. The democratization, measured by the imagination-to-artifact ratio, is real. But the fishbowl of the developed world — the invisible institutional infrastructure that converts artifacts into capital — is also real. And until the inhabitants of that fishbowl can see it, they will continue to mistake tool access for economic inclusion, capability for capital, and the collapse of the production barrier for the collapse of the institutional barriers that actually determine who prospers and who remains outside.

The fishbowl is not malicious. It is architectural. It was built over centuries, through institutional innovations that the inhabitants have forgotten. De Soto's work is, at its core, a project of remembering — of reconstructing the history of institutional infrastructure so that its principles can be applied to the places, and the people, that the current system excludes.

The AI economy needs the same project of remembering. The formal technology ecosystem was not always formal. Silicon Valley was itself, at its inception, an extralegal economy — a collection of garage-based builders operating outside the established institutional structures of the electronics industry. The infrastructure that now sustains it — the venture capital ecosystem, the intellectual property framework, the deployment platforms, the professional networks — was built over decades, through trial and error, through the gradual formalization of practices that had previously been informal. The Valley has forgotten this history, just as the West has forgotten the history of its property systems.

Remembering is the first step toward building. The fishbowl can be made visible. The assumptions can be examined. The institutional infrastructure can be extended. But only if the inhabitants of the fishbowl can see the glass.

Chapter 9: Dead Capital and the Developer in Lagos

She appears three times in The Orange Pill. First in Chapter 14, as the moral anchor of the democratization argument — a developer with intelligence, ambition, and ideas who previously lacked the tools to build and who now possesses those tools at affordable cost. Then again in Chapter 18, where the practical implications of the AI revolution are addressed and where her situation is acknowledged as structurally different from that of her counterpart in San Francisco. And once more, implicitly, in the book's closing meditation on what it means to be worth amplifying.

She is a composite figure, not a specific individual. But the composite is grounded in demographic reality. Nigeria has the largest technology workforce in Africa. Lagos is its center — a city of approximately twenty-two million people, growing at a rate that makes it, by some projections, the world's largest city by 2100. The developer population in Sub-Saharan Africa is growing faster than in any other region, driven by young demographics, expanding internet access, and a generation of entrepreneurs who have watched the global technology economy from outside and are determined to participate in it.

De Soto's framework requires that we examine her situation with the same granular specificity that his researchers brought to the informal settlements of Lima. What, precisely, does the AI economy look like from her position? What institutional barriers stand between her intelligence, now amplified by tools of extraordinary power, and the capital that intelligence could generate?

Begin with what she has. She has a smartphone — the primary computing device for the majority of developers in Sub-Saharan Africa. She may have a laptop, though laptops capable of running modern development environments represent a larger investment relative to her income than they do for developers in the West. She has internet access, though the access is likely mobile-data-based, metered by the gigabyte, and subject to the congestion and outages that characterize mobile networks in dense urban environments. She has AI tools — Claude Code, ChatGPT, open-source models that can run on consumer hardware — available at prices that range from free to approximately one hundred dollars per month.

She has education, though the character of her education differs from the standardized pathway that Western technology companies use as a filtering mechanism. She may have attended a university with a computer science program, or she may have learned through the informal education networks — bootcamps, online courses, YouTube tutorials, community mentorship — that have produced a substantial proportion of Africa's developer workforce. Her technical competence is real and demonstrable. The code she writes with AI assistance compiles, runs, and performs its intended function.

She has ideas. This is the element that The Orange Pill most celebrates, and correctly so. The imagination-to-artifact ratio has collapsed for her. She can describe a problem in natural language and receive a working solution. She can prototype an application in hours that would have required weeks or months of development under previous constraints. The barrier between her intelligence and an artifact has been reduced to the cost of a conversation.

Now examine what she does not have, and examine it with the specificity that de Soto's framework demands.

She does not have reliable electricity. Lagos experiences power outages that are measured not in minutes per year but in hours per day. The national grid provides electricity for an average of approximately ten to twelve hours daily in urban areas, less in rural ones. The remainder is supplied by diesel generators, which are expensive to operate and which most individual developers cannot afford. A coding session interrupted by a power outage is not merely an inconvenience. It is a disruption that forces the developer to reconstruct context — the mental state, the working memory, the flow of the development process — from scratch. The cumulative cost of these disruptions, measured in lost productivity and cognitive fatigue, is a tax that no developer in the developed world pays.

She does not have affordable, reliable broadband. Mobile data in Nigeria costs approximately one to two percent of gross national income per capita for a basic broadband subscription, according to the International Telecommunication Union's affordability metrics. This sounds modest until one considers the data consumption patterns of AI-assisted development. A sustained Claude Code session involves continuous back-and-forth with a cloud-based model, consuming data at rates that can exceed several gigabytes per day for intensive use. At Nigerian mobile data rates, a month of serious AI-assisted development can cost more than many developers earn in that period. The tool is affordable. The data required to use the tool is not.

She does not have financial infrastructure that connects her to the global economy. Nigeria has a vibrant mobile money ecosystem — services like OPay, Kuda, and PalmPay that provide basic financial services to millions of unbanked Nigerians. But these services are domestic. Connecting them to the international payment systems through which the global technology economy transacts — Stripe, PayPal, the credit card networks — requires crossing institutional boundaries that impose friction at every point. Currency conversion fees, regulatory compliance requirements, identity verification processes designed for Western documentation norms, and the periodic freezes and restrictions that Nigerian financial regulators impose on cross-border transactions all constitute barriers between the developer's product and the revenue it could generate.

She does not have intellectual property protection in any practical sense. Nigeria's intellectual property framework exists on paper. The Nigerian Copyright Commission administers copyright law. The Patents and Designs Act provides for patent protection. But enforcement is slow, expensive, and uncertain. A developer who discovers that a larger, better-resourced company has replicated her product faces a choice between legal action she cannot afford and acquiescence she cannot prevent. The nominal protection exists. The practical protection — the institutional machinery that converts a legal right into an enforceable boundary — does not.

She does not have access to the networks that convert products into businesses. Venture capital investment in Nigeria, while growing, remains a fraction of the total available to developers in the United States, Europe, or Asia. The networks through which investment flows — the personal introductions, the pitch events, the alumni networks of specific universities and accelerators — are geographically and institutionally concentrated. A developer in Lagos who builds a brilliant product faces a discovery problem that her counterpart in San Francisco does not: the people with the institutional power to invest in her, partner with her, or distribute her product do not know she exists, and the channels through which they discover new products are not designed to reach her.

Each of these barriers is a component of the bell jar. Each is invisible from the inside of the developed world's fishbowl. Each is the defining reality of the developer's daily experience.

De Soto's researchers in Lima discovered that the total value of informally held assets in the developing world exceeded the total value of all foreign aid, all foreign direct investment, and all development lending combined. The implication was devastating: the resources were already there. The institutional infrastructure to convert them was not.

The same structure obtains in the AI economy. The intelligence is already there. The tools are already there. The motivation — the drive to build, to create, to participate in the global technology economy — is abundantly present. What is absent is the institutional infrastructure that would allow the developer's amplified intelligence to generate capital.

This is not an argument against AI tools. It is an argument for the institutional complement that AI tools require. The tool democratizes production. The institution democratizes capitalization. Without both, the democratization is real in its promise and hollow in its delivery — a revolution that changes who can build without changing who can prosper from building.

The developer in Lagos is not a metaphor. She is a demographic reality — one of millions of builders worldwide who possess the intelligence, the tools, and the ambition to participate in the AI economy, and who are separated from that participation not by a lack of capability but by the absence of the institutional infrastructure that de Soto has spent forty years arguing is the single most important determinant of economic inclusion.

Her intelligence is not dead because she lacks ideas. Her intelligence is dead because the system — the representational infrastructure that would allow her ideas, now buildable thanks to AI, to lead a parallel life as capital — does not extend to where she stands. The bell jar is transparent. She can see the formal economy operating on the other side. She can produce artifacts that match its quality. She cannot reach through.

Building the institutional bridges that allow her to reach through is not charity. It is not development aid. It is the construction of infrastructure that converts dead intelligence into living capital — infrastructure that serves not just the developer in Lagos but the global economy that her intelligence, once formalized, would enrich. De Soto demonstrated that formalizing the assets of the poor is not a redistribution of wealth from rich to poor. It is an expansion of wealth — the creation of new capital that did not previously exist in the formal economy. The same logic applies to intelligence. Formalizing the intelligence of the world's extralegal builders does not diminish the capital of the formal economy. It expands it. The developer's intelligence, once formalized, becomes a new node in the global network of economic production — a node that generates value, attracts investment, creates employment, and contributes to the tax base that funds public services.

The developer in Lagos is the test case for the AI revolution's claim to democratization. If the revolution reaches her — not merely with tools, but with the institutional infrastructure that allows tools to generate capital — then the claim is vindicated. If it does not, then the revolution is a reproduction, in technological form, of the oldest structural exclusion in the history of capitalism: the exclusion of the productive majority from the institutional systems that convert productivity into wealth.

---

Chapter 10: Building the Representational System for the Next Billion Builders

De Soto's career has been defined not by diagnosis alone but by prescription. His analysis of dead capital was always oriented toward a practical question: what, specifically, must be built? The identification of the problem — the absence of representational infrastructure that converts assets into capital — was valuable because it pointed toward a constructible solution. Not a theory. Not a framework for further analysis. A building project.

The building project for the AI age can be specified with the same concreteness that de Soto brought to property rights reform. The components are identifiable. The precedents are instructive. The obstacles are real but not insuperable. And the stakes — measured in billions of lives and trillions of dollars of dead intelligence — are as large as any institutional challenge the global economy has faced.

The first component is digital property rights — formal systems for documenting, registering, and protecting the intellectual property of individuals who currently operate outside the formal economy. This does not mean extending Western intellectual property law wholesale to developing contexts. De Soto's central methodological insight was that institutional infrastructure cannot be imposed from above; it must be built from below, incorporating the customs, practices, and knowledge systems of the communities it serves. Property titling programs that ignored local customs failed. Digital property rights systems that ignore the realities of informal AI creation will fail in the same way and for the same reasons.

What is required is a system that meets the extralegal builder where she is — that recognizes the forms of creation, documentation, and distribution that already exist in the informal AI economy and provides a formal representation that the broader economy can engage with. De Soto's own work has pointed in this direction. His recent advocacy for blockchain-based property registries — immutable, distributed, verifiable records of ownership that do not depend on centralized government institutions — represents a technological approach to the representational problem. At LABITCONF 2024 in Buenos Aires, de Soto articulated a vision in which blockchain, artificial intelligence, and digital infrastructure converge to create the institutional foundation for economic inclusion that the developing world has lacked.

Blockchain-based registries for digital assets, including software products built with AI tools, could perform the six functions de Soto identified for formal property systems: fixing economic potential, integrating dispersed information, creating accountability, enabling fungibility, networking people, and protecting transactions. A developer who registers her product on a blockchain-based registry creates a timestamped, verifiable, tamper-proof record of ownership — a digital title deed that the formal economy can recognize and act upon. The registry does not change the product. It changes what the product can do. It allows the product to lead a parallel life as capital.

The technology exists. The institutional design that would make it effective — the standards, the governance, the dispute resolution mechanisms, the integration with existing legal frameworks — requires the same patient, context-sensitive construction that de Soto's property rights reforms required. It cannot be built in a hackathon. It must be built with the participation and knowledge of the communities it serves.

The second component is financial infrastructure — mechanisms that allow extralegal builders to convert their products into sustainable revenue. The most promising approaches combine mobile money systems, which are already widely adopted in the developing world, with international payment rails that connect local transactions to the global economy. Mobile money platforms like M-Pesa in Kenya have demonstrated that financial infrastructure can reach populations that traditional banking cannot — the key insight being that the infrastructure was designed for the population it serves, not adapted from a system designed for someone else.

The equivalent for the AI economy is a payment infrastructure designed for the extralegal builder: one that accepts local payment methods, handles currency conversion at rates that do not consume the builder's margin, complies with local regulatory requirements, and connects to the international payment systems through which the global technology economy transacts. Such infrastructure would not merely enable transactions. It would formalize them — creating the revenue records, the tax documentation, and the financial history that are prerequisites for access to credit, investment, and the other mechanisms through which the formal economy compounds wealth.

Microfinance, the innovation most associated with the principle that small-scale financial infrastructure can unlock economic participation, offers instructive precedents and cautionary lessons. Muhammad Yunus's Grameen Bank demonstrated that the poor are creditworthy when the lending infrastructure is designed for their context — small loans, community-based repayment structures, lending criteria adapted to the realities of informal enterprise. The model generated enormous optimism and was widely replicated. It also generated criticism, particularly when microfinance institutions pursued scale and profit at the expense of the communities they were designed to serve. The lesson is not that microfinance failed. It is that financial infrastructure for the excluded must be governed by the interests of the excluded, not by the return expectations of outside investors. The same principle applies to financial infrastructure for extralegal AI builders.

The third component is localized tools — AI systems designed for the contexts in which the majority of the world's builders actually work. This means training data that represents the problems, languages, and institutional realities of the developing world. It means interfaces optimized for mobile devices and intermittent connectivity. It means models that can run locally on consumer hardware when cloud access is unavailable or unaffordable. And it means community-driven customization — mechanisms that allow local developer communities to fine-tune and extend the tools for their specific contexts, incorporating the local knowledge that de Soto's framework identifies as essential to effective institutional design.

The open-source movement provides the foundation for this component but not the complete structure. Open-source models are necessary because they allow communities to adapt tools to local contexts without dependence on the decisions of distant corporations. But open source alone is insufficient, because the compute resources, training pipelines, and technical expertise required to fine-tune large language models are themselves unevenly distributed. The infrastructure must include not only the models but the capacity to modify them — community compute resources, training data cooperatives, and the educational programs that de Soto has advocated for integrating AI and related technologies into primary and secondary curricula, creating a generation that understands and leverages emerging technologies from an early age.

The fourth component is marketplace infrastructure — platforms that connect extralegal builders to customers, collaborators, and investors globally, with the institutional protections that builders in the formal economy take for granted. This means discovery mechanisms that are not biased toward the products of the developed world. It means reputation systems that allow builders to accumulate credible track records outside the traditional credentialing pathways. It means escrow and dispute resolution services that protect both buyers and sellers in transactions between parties who do not know each other and may be separated by continents, cultures, and legal jurisdictions.

Elinor Ostrom's research on collective management of shared resources offers a governance model for these marketplaces. Ostrom demonstrated that communities can manage common-pool resources effectively when the governance structures are designed by the participants, adapted to local conditions, and enforceable through mechanisms the community recognizes as legitimate. Platform cooperatives — marketplaces owned and governed by their participants — represent one institutional form that could apply Ostrom's principles to the digital economy. The platform creates the infrastructure. The community governs it. The value flows to the builders.

The fifth component is what de Soto called "political awareness" — the understanding, among policymakers and political leaders, of what is happening in the AI economy and what institutional responses it requires. The political class in most countries has not yet grasped the structural nature of the AI transition. Policy responses have focused on the supply side — regulating what AI companies may build, requiring disclosures, assessing risks. The demand side — the institutional infrastructure that citizens, workers, and builders need to participate in the AI economy — remains largely unaddressed. De Soto's career demonstrates that institutional reform is a political project, not merely a technical one. Building the representational system for the next billion builders requires political will, legislative action, and the sustained attention of governments that are, at present, focused elsewhere.

The Orange Pill concludes with a call to stewardship — to building dams that redirect the river of intelligence toward life rather than destruction. De Soto's framework specifies what those dams must be made of. Not goodwill. Not aspiration. Not the earnest hope that capability will translate into prosperity. The dams must be made of institutions — property registries, financial infrastructure, localized tools, marketplace platforms, and the governance structures that maintain them. These are not abstract desiderata. They are constructible systems, analogous to the property systems that the developed world built over centuries and that de Soto has advocated extending to the developing world for forty years.

The construction will be difficult. It will require resources, patience, and the political will to prioritize institutional infrastructure over more visible investments. It will require the humility to build from local knowledge outward rather than from Western assumptions inward. It will require the recognition that the extralegal builders of the AI economy are not a marginal population to be accommodated but the primary constituency to be served — the majority of the world's intelligence, the majority of the world's creative energy, the majority of the world's potential capital, currently locked outside the system that would allow it to generate the economic life that billions of people and their societies need.

The imagination-to-artifact ratio has collapsed. This is the technological achievement of the age. The task now is to build the artifact-to-capital infrastructure that allows the collapse to produce not a new form of the old exclusion but a genuinely inclusive expansion of who gets to build, who gets to prosper, and whose intelligence gets to live.

De Soto has demonstrated, across four continents and four decades, that dead capital can be brought to life through institutional construction. The dead intelligence of the AI age is waiting for the same construction. The blueprints exist. The materials are available. The builders — billions of them — are ready. What remains is the will to build.

---

Epilogue

The nine-point-three trillion dollars changed everything for me.

Not the number itself — numbers that large stop meaning anything; they become decoration. What changed was the inversion it demanded. De Soto's researchers counted the assets of the world's poor and arrived at a figure larger than the combined stock market capitalization of the twenty wealthiest nations. And the conclusion was not that the poor were secretly rich. The conclusion was that the system was secretly broken — that the machinery designed to convert assets into capital had a hole in it large enough for the majority of humanity to fall through.

When I wrote The Orange Pill, I wrote about the developer in Lagos as proof that the floor was rising. I believed it then, and I believe it now. The tools are real. The capability is real. The imagination-to-artifact ratio has collapsed in a way that I have felt in my own hands, on my own screen, in the work I do every day with Claude. That collapse is not rhetoric. It is the lived experience of millions of builders.

But de Soto forced a question I had not asked with sufficient precision. I had asked: Can she build? The answer is yes — demonstrably, measurably, beautifully yes. I had not asked with equal rigor: Can she capitalize? Can she deploy on infrastructure she can afford? Can she reach customers through channels that see her? Can she protect what she has made? Can she sustain what she has started? Can her work compound?

Those are not the same question. The distance between them is the distance between a working prototype and a livelihood, between an artifact and a life. And that distance is institutional, not technological. The tools crossed the gap. The institutions have not.

What de Soto's framework taught me is that the river metaphor I built through The Orange Pill was right but incomplete. The river of intelligence is real. It flows. It is widening. But a river without institutional channels does not irrigate. It floods. And the difference between irrigation and flood is not the volume of the water. It is the infrastructure on the ground.

I think about my engineers in Trivandrum — the twenty-fold productivity multiplier, the excitement in the room, the future we were building together. And then I think about the identical capability, the identical tools, deployed in a context where the electricity cuts out, where the data costs more than the subscription, where the payment rails do not reach, where the intellectual property framework exists on paper and nowhere else. The same tools. The same intelligence. A radically different institutional reality.

The hardest lesson in this book is that democratizing the tool is necessary and insufficient. It is the easier half of the project. The harder half — the construction of representational infrastructure that allows intelligence to lead a parallel life as capital — is institutional work, political work, slow and unglamorous work that does not produce the kind of exhilaration I described standing on the CES floor watching Station come alive. It is beaver work, in the precise sense I meant: sticks and mud and teeth, placed carefully, maintained constantly, serving an ecosystem that extends far beyond the builder.

De Soto has been doing that work for forty years. The rest of us are just starting to understand what the AI age requires of us — not just better tools, but better systems. Not just access, but infrastructure. Not just the right to build, but the institutional architecture that allows building to become something more than an impressive, functional, economically invisible act.

The next billion builders are ready. The question is whether we will build the systems worthy of their intelligence.

-- Edo Segal