The middleman of this plate builds the meeting-place itself. He gathers the two sides who might trade, supplies a means for them to find and deal with one another, and charges for passage through the venue. The platform can be a marketplace, a payment system or a network connecting people who need different things from the same encounter. Its service is partly the gathering: a seller values access to buyers, and a buyer values access to suitable sellers. An empty venue cannot offer either side much of that service, however well its machinery runs. The owner must therefore create a reason for each side to arrive before the other side's presence can be taken for granted. Once the gathering succeeds, the same owner sets terms for access to something the participants have helped make valuable. This is the volume's familiar seam, drawn through the marketplace rather than through a single trade. The network can be both a useful achievement and a source of bargaining power. The fee asks how much the owner can keep for bringing the parties together and continuing to stand between them.
The take rate gives that fee a convenient form: the portion of the value transacted that the platform retains. If a specified hundred-unit sale carries a ten-unit platform charge, the take on that transaction is ten percent. The calculation becomes less simple when the venue also charges for entry, subscriptions, placement or other services, or pays incentives to attract participants. One must say which receipts and which transactions enter the measure. Nor is the fraction a profit margin. Maintaining the service, processing payments and meeting the obligations attached to a transaction can consume part of what is collected. The service bundle differs across venues: some arrange a match, some settle a payment, some undertake further work after the parties agree. A comparison of two percentages can therefore compare different things. The quoted rate tells the participant what a particular route charges under specified terms. To understand the platform's business, one must also follow the charges and benefits across the other side of the meeting.1
The price has a structure as well as a level. Jean-Charles Rochet and Jean Tirole showed why a platform must consider which side pays, because attracting one side changes the service offered to the other. A simple illustration makes the point. Suppose that using a venue gives a buyer an additional benefit of three and a seller an additional benefit of twelve, before platform fees. Hold the price of the underlying good fixed, and assume that the parties cannot negotiate another transfer. Charging each side five produces a total fee of ten, but the buyer will decline. Charging the buyer nothing and the seller ten leaves both with a positive benefit, so the same total charge can support a transaction.2 The cheap side is not necessarily a neglected source of revenue. Its participation can be what makes the other side willing to pay. The qualification about transfers matters: if the seller could freely compensate the buyer, the parties might undo the difference between the two fee structures. A two-sided pricing problem is strongest where their own bargaining does not fully neutralize the platform's allocation of charges. Merely having a buyer and a seller is not enough to establish it.
The gathering can reinforce itself. More suitable sellers can make a venue more attractive to buyers; more buyers can give sellers a stronger reason to join. A challenger must offer more than an empty room at a lower rent. It may need to attract enough participants on one side to make the other side's arrival worthwhile, and to bear the cost of doing so before the network is established. That difficulty can protect an incumbent, but it does not make a single permanent winner inevitable. Caillaud and Jullien's model of competing matchmakers produces different market structures depending on the cost and effectiveness of matching and the ways users can join. A single venue can be efficient when bringing the participants together avoids costly fragmentation; several venues can be useful when their matching opportunities complement one another. Even a concentrated outcome can face a serious threat of entry.3 Network effects explain why the choices are interdependent. To infer how much power follows, one still needs to know what a rival can offer, what the participants can change, and which costs prevent them from doing so.
Using more than one platform changes the competition, without supplying a universal cure for power. A seller who can reach the same buyers through another venue may be able to direct a transaction toward the cheaper route. Rochet and Tirole explain how participation in several networks on one side can intensify the contest for the other side. But Caillaud and Jullien also find arrangements in which nonexclusive services soften rivalry and sustain positive intermediary profits. The implications depend on which side joins several venues, how the matches overlap, and whether the venue charges for registering or for completing the transaction.4 The relevant alternative is therefore more concrete than another application installed on a device. It is another feasible way to reach the desired counterparty on acceptable terms. A participant may have many nominal options and few useful substitutes; a large venue may face a smaller rival that provides an effective route for an important set of transactions. The analysis follows those routes, rather than treating the number of names in a market as a sufficient measure of escape.
The fee consequently carries more than one question. How costly is the service? How much does each side value the access? How readily can a rival supply a comparable route? A zero price on one side does not establish that the whole arrangement is free, just as a substantial charge on the other does not by itself identify the amount of rent. The chosen balance can enlarge participation and still differ from the balance that would maximize the participants' combined benefit. Rochet and Tirole's comparison of a profit-seeking platform with a budget-balanced welfare benchmark makes that distinction explicit: the platform's interest in transaction volume is not identical to its users' interest in the surplus those transactions create.5 The network is a reason the service can be valuable and a reason its replacement can be difficult. Those facts can coexist without fixing the proper size of the fee. A useful venue can charge more than a rival would require to provide it; a costly venue can also need a substantial payment simply to continue. The percentage alone cannot separate those cases.
When the marketplace is owned, the owner can charge for a meeting whose value comes partly from the people who attend it. That is the platform's distinctive position. Its achievement may be to make the meeting possible, cheaper or more reliable; its advantage may also be that the parties have come to depend on finding one another there. The task is to distinguish the service from the limits on alternatives without pretending that they occupy separate corners of the same network. Better access can increase value and alter power at once. The inquiry must follow the conditions of joining, trading and leaving, and the consequences of replacing the arrangement with another that could actually work. The next plate, The Disintermediation, begins with precisely that attempted replacement. It asks what happens when a new route promises to cut out the middleman. If the meeting-place disappears, the parties must still find each other; if another owner supplies it, the question of the cut has acquired a new address.