The Disintermediation

The promise to cut out the middleman begins with an inviting subtraction. Remove the person who takes a portion of the payment, and the buyer should pay less while the seller keeps more. The promise becomes persuasive whenever a new means of communication lets the two sides reach one another directly. Yet reaching is only one part of dealing. Someone must establish what is offered, arrange the payment, deliver the thing, decide what happens when it disappoints, and bear whatever exposure remains between agreement and completion. Those functions explain why a middleman can return after an old one has been removed. The distinction worth preserving is between removing a particular organization and changing the work of exchange. A new technology can do both, but the disappearance of a name from the invoice does not tell us which costs have disappeared with it. To understand the saving, we have to follow the work after the old intermediary leaves.

The work can follow several routes, each with a different meaning for the spread. A producer can take orders directly and bring distribution inside its own business. A buyer can spend time comparing offers that a broker once compared on the buyer's behalf. Software can automate a check, or a change in the product can make an old handling operation unnecessary. A specialist can perform one remaining task for many producers, spreading its fixed costs across a larger volume. Only the last route necessarily creates a new independent intermediary. Calling every act of self-service another middleman would make the prophecy impossible to test: even a buyer doing the task unaided would count as evidence that the intermediary survived. What survives in that case is a function, with a cost in the buyer's time. The economic question is how the organization of that function has changed, and whether the complete arrangement provides the needed service at a lower cost. The count of firms standing in a row cannot answer it.¹

The same improvement that makes direct dealing cheaper can also make hiring a specialist cheaper. This was the central difficulty with the simple forecast of the intermediary's extinction. A network can reduce the cost of connecting a producer to a customer, while also reducing the cost of connecting that producer to a payment service, a delivery service or a source of product information. Which arrangement grows depends on the relative changes. Sarkar, Butler and Steinfield argued that electronic exchange could unbundle functions previously performed together: the physical journey of a product might become shorter even as the information and service relationships around it became more numerous. Their argument gives a mechanism for the return of the middleman without making his return a law of nature. A producer with sufficient scale may bring a function inside; another may purchase it from outside. A service that once required a local organization may be supplied remotely to many customers. The technological improvement belongs to the whole system of relationships, and the firm that first appears to lose from it may also acquire a cheaper way to perform its work.²

A smaller visible fee is therefore an incomplete measure of the saving, though it may be part of a real one. Suppose a good costs one hundred before distribution, and an intermediary charges twenty for a specified bundle of services. The buyer's payment is one hundred and twenty. A replacement charges eight but requires the buyer to undertake work worth five in time; with the good and all other consequences held fixed, the buyer's total cost is now one hundred and thirteen. The visible fee has fallen by twelve, while the saving after counting the transferred work is seven. If the producer can supply the same bundle internally for seven and charges that amount, the buyer's payment becomes one hundred and seven even though no separate intermediary appears. These deliberately simple accounts leave out quality changes, risks and the division of any saving between the parties. Their purpose is to make those omissions visible. A fee can disappear because a task became cheaper, because another party now performs it, or because the service is no longer supplied. Those outcomes have different consequences even when the invoice looks the same.³

The ledger that promises exchange without a trusted intermediary makes the distinction especially sharp. A shared protocol can allow participants to verify transfers of an asset native to that ledger without asking a central keeper to certify each one. That is a substantial change in how a particular task is performed. It leaves a different question when the record purports to describe something outside itself: whether a parcel contains the promised object, whether a physical asset exists, or whether an external event has occurred. A consistent record of an assertion does not establish the truth of the assertion at its point of entry. Catalini and Gans distinguish the cheap verification of digital state from this connection between the record and the world. The connection may require instruments, institutions or people whose reliability still has to be established. Their account also distinguishes verification from the incentives and resources needed to operate the network. The comparison is therefore between different arrangements of trust, work and control, each with its own failure conditions. Automating the record can remove a task from one institution while leaving several other tasks for somebody to perform.⁴

The return of an intermediary leaves the volume's question about power fully open. A specialist can save resources and still charge more than the service requires; an established firm can survive because customers find it useful, because changing providers is costly, or because it controls access to something they need. Conversely, removing a formerly useful firm need not expose its past income as pure toll. A later technology may have made its work redundant or allowed someone else to perform it more cheaply. These possibilities prevent the experiment of removal from delivering a complete verdict merely by succeeding or failing. To learn from it, we need the counterfactual: what service is provided, by whom, with what risks and access conditions, before and after the change? The previous plate, The Platform, examined the owner of the meeting-place. Disintermediation asks whether a new route actually frees the parties from that position or carries them into another one. The presence of a useful function explains why someone might be paid; it cannot establish the deserved size of the payment.⁵

The enduring achievement of disintermediation is to make an established arrangement answer to an alternative. Sometimes the alternative removes a toll; sometimes it improves a service; sometimes it moves the work and its cost beyond the edge of the old account. The buyer and seller benefit when the complete terms improve, whether the resulting system contains fewer firms or more. The useful discipline is to ask which function has disappeared, which has changed hands, and who can now command its price. That question preserves the force of the old suspicion without treating every intermediary as dispensable, and preserves the defense of real service without awarding every surviving middleman a certificate of merit. The next plate, The Commission, follows the work into the hands of a person engaged to represent one of the parties. Here the intermediary may find the counterparty, advise on the bargain and help conclude it. The difficulty becomes more intimate: the person trusted to improve the exchange has a payment of his own depending on how it ends.