Plate vii in print

The Forum

the private marketplace; the platform that owns the market it hosts, and writes the rules for those who trade within it.

The owner of the marketplace governs more than the goods it sells.

— the argument of this plate, after the earlier Platform and Rochet and Tirole

The marketplace can be private property even while the trades inside it belong to other people. Its owner supplies the room in which buyers and sellers find one another, establishes the admission rules, orders what appears first, and decides how disputes interrupt a transaction. The earlier Platform followed the power created by that position. Here the volume asks how the position is governed. A merchant may own her stock and choose her price while depending on another firm's permission to reach her customers. A buyer may choose freely among the offers displayed while having little knowledge of the rules that selected those offers. The forum makes exchange easier and, through the same machinery, gives its operator a continuing part in the decisions of those who exchange.

The operator's first problem is to attract participants whose presence makes participation valuable to others. Buyers want sellers worth visiting; sellers want buyers worth serving. Rochet and Tirole sharpened the economic meaning of this relationship: in a two-sided market, the allocation of charges between the sides can affect transactions even when the total charge is unchanged. Consider a hypothetical service in which a user obtains two units of benefit and a merchant obtains eight from a completed interaction, with no way to renegotiate the fees between them. A four-unit charge on the user prevents the interaction. Charging the merchant the same four leaves both willing to participate. The arithmetic explains why a free service on one side can support a viable market. It does not by itself establish how competitive or fair the resulting arrangement is.1

1. Rochet and Tirole, Two-Sided Markets: A Progress Report, November 29, 2005 working-paper version, introduction. The invented example assumes that users and merchants cannot undo the fee allocation by bargaining with one another.

Governing admission and conduct can create real value for everyone using the forum. A marketplace that cannot distinguish a genuine seller from a disappearing fraudster offers a cheap doorway into an expensive disappointment. Verification, payment protection, dispute handling, and limits on harmful conduct may therefore enlarge the set of trades people are willing to make. Suppose an exchange produces a combined expected benefit of ten, but an unaddressed risk costs the participants six. A safeguard that costs two and reduces the expected loss to one raises the net benefit from four to seven. The operator can earn a return while improving the transaction. The question is whether a particular rule supplies that service, at what cost, and with what errors imposed on participants who are excluded or delayed.2

2. The illustrative safeguard changes total expected benefit from 10 − 6 = 4 to 10 − 1 − 2 = 7. The gain depends on the assumed effectiveness and resource cost, and says nothing about the desirability of an unrelated admission restriction.
⁂

The same authority can favor the operator's own interests when it also competes within the market. Imagine a forum that earns a commission of two on an independent merchant's sale and a margin of five on its own otherwise comparable offer. Directing the next sale to itself adds three to its return. A ranking rule that does so may be profitable even if the buyer would have preferred the merchant's offer. This conditional incentive is enough to explain why the criteria matter; it is not proof that every ranking change is abusive. Delivery quality, fraud risk, price, relevance, and paid promotion can all affect what should appear first. Participants need to know which kind of decision the system is making and whether the operator's competing role changes the standard applied to rivals.3

3. The assumed operator return rises from 2 to 5, a difference of 3. This identifies a possible conflict of interest; determining the effect of an actual ranking rule requires its criteria and evidence about alternatives.

Exit disciplines private government only to the extent that an alternative can carry the activity away. A seller may be able to open an account elsewhere while being unable to move the reviews, customer relationships, or compatible software that made the original account valuable. Buyers may follow the range of sellers, and sellers may wait for the buyers. Yet several forums can coexist where participants use more than one, specialize, or exchange through interoperable arrangements. Dependence is consequently a matter to investigate, not a conclusion contained in the word platform. The costs of joining, leaving, and using several services help determine how much freedom the operator has to revise fees or terms after participants have invested in being there.4

4. This adaptation draws on The Platform. Network effects, accumulated data and switching costs can strengthen dependence; their strength and the ability to use multiple services vary by market.

A workable private constitution makes the exercise of that freedom intelligible and contestable. It states the material rules, gives useful notice of changes, identifies the reason for a consequential exclusion, and provides a route to correct an error. Different settings require different procedures: an urgent fraud response cannot always wait for a prolonged hearing, while an unexplained permanent exclusion can destroy a business before its account is heard. Portability and interoperability can strengthen alternatives, but must be designed around privacy, security, and the actual function being moved. External public rules can also constrain the forum, and the forum's own remedies cannot settle every public question. The institutional task is to connect its operational power to reasons, review, and responsibility.5

5. The governance discussion states design questions rather than the requirements of a particular jurisdiction. It does not reproduce the earlier edition’s dated account of pending litigation or legislation as a current status report.

The forum reveals how a market can contain many competing sellers and still be governed by one concentrated source of authority. Its owner may create the conditions that make trade possible and retain the ability to alter those conditions after others have come to rely on them. Both facts belong in the account. A useful marketplace deserves payment for the service it supplies; the dependence it creates gives participants a reason to examine how that service is governed. The Architect now widens the frame from the rules of a single place of exchange to the institutions of an economy. The question travels with it: who can enter, who may decide, who can challenge a decision, and who bears the cost when the rules are changed?