The gate may be free to enter and expensive to pass through. A reader pays nothing to join, a seller pays to be found, an advertiser pays to appear, and the platform that brings them together decides which encounters will occur and on what terms. The familiar question—how far has the seller raised the buyer's price?—has acquired an incomplete answer. One participant sees a zero; another sees a bill; both depend on an arrangement neither controls. Yet the zero itself proves neither generosity nor abuse. A platform may sensibly subsidize the group whose presence attracts the other, and the resulting service may be worth far more than either group could have organized alone. The inquiry into power begins where this useful act of bringing people together gives its organizer authority over their access to one another.
The price of a platform has to be read across the sides it joins. A marketplace with many prospective buyers is more useful to its sellers; more sellers can make it more useful to buyers. Charging each group what it costs to serve that group separately can overlook these cross-effects. Reducing one charge may bring in participants whose presence makes the other side willing to pay more. This does not mean that any business with two kinds of customers has escaped ordinary economic reasoning. The distinctive issue is whether the allocation of charges between the groups changes participation and transactions, rather than being neutralized by payments they can freely rearrange among themselves. Where they cannot readily make those adjustments, the platform chooses a structure of prices as well as their level. Examining one fee in isolation can mistake the very mechanism that makes the meeting possible.
A platform can become a bottleneck when the people on one side can be reached nowhere else. Suppose users choose a single service while businesses can join several. The business that wants a particular service's users must deal with that service, even though it can also buy access elsewhere. Rival platforms may compete hard to attract those exclusive users and spend much of the resulting access revenue in doing so. Profits can consequently be modest while the terms of access remain powerful. The possibility depends on the actual pattern of use. If users visit several services, if sellers can reach them directly, or if the desired audience can be carried to a rival, the bottleneck weakens. The platform's size is evidence about its reach; the exclusivity of that reach is a separate question about the alternatives available to those who need it.
The network makes this position durable when leaving means losing the people who make the service valuable. A better interface is then an incomplete offer: the rival must also make it plausible that the relevant others will be there. Expectations can reinforce an established lead, but the reinforcement has limits. Participants may value different communities, dislike crowding, seek specialized services or use several networks at once. Connections may be transferable; communication may cross the boundary. The claim that a network tends to favor a large operator therefore needs its missing clauses restored. Which participants benefit from which others? Over what range? Can a newcomer assemble a useful group before it has assembled everyone? A whole industry does not acquire a natural destiny merely because one additional connection is valuable.
The gatekeeper can exercise its authority through terms that never appear as a positive price to the user. Visibility, waiting, access to functions, the amount of attention demanded and the conditions attached to participation can all alter what the service offers. Businesses on the other side may pay more for the same audience, or find that reaching it requires a further service supplied by the gatekeeper. These possibilities do not establish that any particular change is harmful: a new rule may improve quality, protect users or reduce a real cost. They explain why an investigation confined to the zero in the consumer's bill can miss the dispute. The relevant comparison is between the complete terms that feasible alternatives would supply. Quality, privacy and access become part of that comparison because they are part of the exchange, not because a zero-price service must secretly be worthless.
Information can deepen an advantage only through the work it actually does. A history of transactions may improve matching, help identify fraud or make a supplier more useful to its existing customers. For the advantage to protect a position, however, it must be difficult enough to reproduce, substitute for or carry away. A larger store of records is not by itself a measure of a larger barrier. Their relevance, accuracy and permitted uses matter; so does the speed with which a newcomer can obtain useful information of its own. The same practical distinction governs portability. Moving a pile of records will not necessarily recreate the relationships, reputation or live access that gave them value. The question is what a person needs in order to make an alternative service workable, and which of those needs the incumbent alone can satisfy.
Taxing the platform requires deciding which part of this arrangement the tax is meant to reach. A charge on a transaction raises the cost of making that transaction. A charge on gross receipts can reach payments that are needed to cover the service, the subsidy to another group or the ordinary return to investment. A proportional tax on pure economic profit has a different ideal logic: if it applies to all the relevant profit, allows the full associated costs and treats losses symmetrically, it scales the return to every available choice without changing their ranking. That observation concerns a deliberately clean tax base, not the ease of measuring it. Allocating shared costs, identifying the normal return and dealing with losses and investment can reopen the problem. Even a perfectly measured tax that takes part of a rent need not make the gatekeeper's access rules more open.
A small invented market shows why the distinction matters. Hold the audience and quality fixed, let the price paid by businesses be p = 10 − q, and let each unit of access cost two to supply, with no fixed cost. The platform maximizes (8 − q)q by selling four units at six, earning sixteen. A tax of one quarter of that economic profit collects four and leaves the same quantity and price, with twelve retained. A tax of two per unit changes the objective to (6 − q)q: three units are sold at seven, the tax collects six and profit after tax is nine. The customer price rises by one rather than by the entire two-unit tax. The example is intentionally spare. Once audience participation responds, the platform can also change the subsidy, quality or conditions on that side. The burden then travels through the whole arrangement, and the statutory address on the tax bill cannot tell us where it finally rests.
Breaking the platform likewise requires specifying what is to be separated. Splitting one network into disconnected fragments can reduce the encounters each fragment makes possible. Connecting the fragments under workable access rules can preserve more of that benefit while permitting rivalry in the services offered over them. Separating the gatekeeper from a business that sells through the gate addresses a different concern: the incentive and opportunity to favor its own downstream activity. It does not automatically remove the gatekeeper's ability to charge for access. Nor does a new corporate boundary automatically reproduce the coordination previously performed within the firm. The remedy has to describe who can connect, on what terms, with what responsibilities for security, quality and payment. A count of newly independent companies is no more a complete account of the result than the old count of one.
Interoperability is promising when it makes the audience reachable without requiring every customer to abandon the service at once. It can loosen the coordination problem that protects an established network. But a duty to exchange messages is different from a duty to share a reputation score, a recommendation system or a commercial relationship, and each raises different costs and responsibilities. The useful demand is precise enough to be performed and checked. Portability has a related purpose: to reduce the investment a customer must sacrifice on leaving. Neither word should stand in for the actual arrangement. The reform succeeds economically only insofar as it creates a usable alternative while accounting for the qualities that made the original connection valuable. It can otherwise leave the gate exactly where it was, furnished with a new form to fill out on the way through.
The platform's achievement and its authority must be examined together. Bringing the two sides into contact can create a market that would otherwise be smaller or absent. Once that contact is organized, the terms for continuing it can become a source of dependence. Competition for the audience, competition for access to it and competition in the services offered through it are related contests, but they need not have the same strength. A low price in the first can coexist with restricted options in the second. A successful intervention must identify which contest it expects to improve, what participant will acquire a usable alternative and how the other participants are likely to respond. The same discipline applies to the defense of the existing arrangement. The fact that a service is useful does not settle whether its present terms are the best available way of providing it.
The service may cost nothing to enter while the right to reach its users becomes the thing for sale.
Where, then, has the power gone when the visible price is zero? It has not necessarily gone anywhere. It may reside in the paid side, in the terms of access, in the cost of moving or in the authority to decide which encounters are allowed. It may also be constrained more sharply than the size of the platform suggests. Finding out requires tracing alternatives through the network, rather than treating free admission as the end of the inquiry. Taxation can address the return, separation can alter the incentives, and access rules can change the possibilities of entry; each answers a different part of the question. The ideal of competition returns in the final plate as a demand to say what would make those alternatives real—and what, even if made real, they could be expected to accomplish.