There is a standard explanation for why markets fail to form. The textbook version involves transaction costs, information asymmetry, and trust deficits. These are real, but they miss something more fundamental.
In a thin market, the thing being exchanged resists description.
This is not an accident. It is the definition. A grain shipment with standard specs, standard delivery terms, and standard pricing is a commodity — it trades in a thick market, with many buyers, many sellers, and price discovery that happens in real time. The moment those specifications become unusual — a specific protein content, a particular origin corridor, a storage history that matters for end use — the market thins. The more specific the requirement, the fewer parties can fill it. The fewer parties, the harder it is to find each other and agree.
The complexity is the value. And the complexity is the problem.
The Stranger Problem
Human beings have been navigating complex exchanges with known counterparties for millennia. Business relationships built on repeated interaction, shared networks, common professional vocabulary, and mutual accountability produce surprisingly efficient outcomes even without formal contracts or regulatory oversight. The grain merchant who buys from the same cooperative every harvest, the manufacturer who has sourced specialty components from the same family shop for thirty years — these are relationships that encode enormous amounts of mutual knowledge without anyone writing it down.
Thin markets destroy this comfort. By definition, a thin market is one where there are not enough known counterparties to sustain the exchange. The buyer who needs a specific product or service has to find a seller they have never dealt with. The seller has to trust a buyer whose reliability is unproven. Both have to communicate about something complex — something that resists the kind of shorthand that familiarity normally provides.
Two parties, barely known to each other, trying to negotiate something neither can describe in simple terms. This is the structure of every thin market transaction.
What Tribal Knowledge Does
The term “tribal knowledge” captures something specific: the domain understanding that circulates within a professional community and is never written down. It includes the vocabulary that practitioners use among themselves, the standards that are technically optional but practically mandatory, the difference between what a contract says and what it means in practice, the judgment about which specifications matter and which ones are pro forma.
In thick markets, this knowledge is distributed across enough participants that most parties already have it, or can acquire it cheaply. In thin markets, it is concentrated in the hands of a few — usually the brokers and intermediaries who have spent careers working the specific corridor, vertical, or domain where the exchange happens.
This creates a dependency that is productive but fragile.
Productive, because a good broker doesn’t just find counterparties — they translate between them. They know that a buyer in Southeast Asia asking for “high-protein wheat” means something different than a European mill asking for the same thing. They know which contract template applies in which corridor, which quality inspectors are trusted by which ports, which regulatory requirements are enforced strictly and which are routinely waived. They make complex exchanges intelligible to parties who do not share a common professional vocabulary.
Fragile, because this knowledge is personal. It lives in the broker’s memory. It does not transfer reliably when brokers retire, change firms, or simply move on. And it has a darker dimension: knowledge that cannot easily be acquired elsewhere is knowledge that can be rationed. A broker who controls access to a market has leverage — leverage that is generally benign when times are good, and that becomes something else when an industry comes under pressure.
The Bar Keeps Rising
The broker problem is not static. Both globalization and the advance of technology are continuously increasing the information requirements of every transaction — raising the complexity ceiling that brokers must clear to make a deal happen.
Globalization multiplies regulatory regimes. A grain shipment that crosses three borders now touches the phytosanitary requirements of each destination country, the export licensing rules of the origin, the customs documentation of the transit ports, and a growing stack of sustainability and provenance certifications — the EU Deforestation Regulation, carbon border adjustment mechanisms, supply chain due diligence frameworks. Each regulation is a new body of specialized knowledge. Every new trade corridor a broker might work adds another layer.
Technology advances do the same thing in the opposite direction. Precision instruments, IoT sensors, and digital quality-assurance systems generate richer, more granular product specifications than were possible a generation ago. A steel buyer who once specified a grade and a yield strength now specifies surface roughness to the micron, traceability to the heat batch, and certification to an updated version of a standard that was revised eighteen months ago. The information density of the transaction increases with every product cycle.
The cumulative effect is a world where the domain expertise required to navigate a single transaction keeps growing — and the population of humans who hold that expertise does not grow at the same rate. Brokers are not becoming more scarce because the profession is declining. They are becoming more scarce because the knowledge load per transaction is increasing faster than individuals can absorb it.
This has a specific economic consequence that is rarely named: there is a complexity threshold above which human brokerage becomes structurally uneconomic. A transaction complex enough to require deep expert intermediation but not large enough to justify its cost simply does not happen. The deal that would generate $40,000 of value on each side dies because the broker who could navigate it needs to charge $60,000 to make it worth their time. Both parties would have traded. Neither walks away better off.
This is not a marginal failure case. In many thin market verticals — specialty agricultural commodities, precision industrial subcontracting, licensed professional services across jurisdictions — the proportion of potential transactions that fall below the cost-of-intermediation threshold is probably large and growing. The unrealized trades are invisible, because markets record what happens, not what doesn’t.
The Two Exits
When a transaction is too complex to navigate without expert intermediation and too small to justify its cost, there are historically only two exits.
The first is standardization. Reduce the specification until the product is simple enough to trade without a guide. This works. Commodity markets are extraordinarily efficient. What it costs is the information that made the goods specifically valuable. The grain buyer who needed a particular protein profile and storage history accepts a standard grade instead. The value that would have been captured by a more precise match is abandoned.
The second exit — which most markets take without acknowledging it — is that the deal simply does not happen. The buyer and seller who could have traded don’t find each other, or find each other and cannot bridge the complexity gap, and part ways. No one records this outcome. It is the default failure mode of thin markets everywhere.
AI-augmented brokerage — and potentially, complete AI brokerage for categories of transactions that are well-defined enough — represents a third exit that has not previously existed. A system that can hold and apply the relevant domain knowledge for a specific vertical at a fraction of the marginal cost of a human expert changes the economics of intermediation. Transactions that were too complex for unassisted negotiation and too small for professional brokerage might fall within reach.
Whether that is actually achievable is still a hypothesis. The knowledge that a veteran broker carries is not just a database of regulations and contract templates. It includes judgment about when rules are enforced, whom to trust in ambiguous situations, and how to navigate the human dimensions of a negotiation that is technically stalled. Encoding that judgment reliably enough to stake a deal on it is a genuinely hard problem.
DeeperPoint is building the scaffolding that allows sponsors of specific thin market platforms to test it rigorously. Not a pitch — a working set of tools that a sponsor can deploy in a real market vertical, with real participants, to find out whether the hypothesis holds where it matters: in the field, on transactions that were previously not happening.
This post introduces a series on why thin markets fail to close even after parties have found each other. The next two posts go deeper into the architecture: why a semantic match isn’t enough to close a deal, and how a metadata schema makes domain knowledge retrievable at the right level of specificity.