Multiple economic agents play a role in any given transaction. Digital coordination of these agents is the role of the platform or online intermediary in this multi-sided markets model. The user or buyer is on one side of the platform, the content provider or seller is on another and the ultimate payee, perhaps the advertiser, is on the third. The platform is a passive intermediary, a bazaar for buyers and sellers, who themselves originate and consummate the transaction. The platform itself requires no deep knowledge of the characteristics or motives of the various economic agents. The platform model captures the essence of a network economy, where the platform now replaces multiple participants along the linear supply chain. This design feature breaks the market into its modular components allowing direct buyer-seller interaction without intermediaries.
By contrast, the traditional economy had active intermediaries who, using historical knowledge of market participants’ profiles, systematically matched buyers and sellers. Economists simplify markets by ascribing the matching function to impersonal prices - markets comprise buyers and sellers who respond to prices determined by some invisible mechanism. The price mechanism is insufficient when the interaction consists of “a two-sided matching market that involves searching and wooing on both sides. A market involves matching whenever price isn’t the only determinant of who gets what” . Market design is the science of matchmaking, configuring appropriate rules for various markets. The criteria for a successful matching are thickness (multiple agents), lack of congestion (resolving time sensitive issues such as exploding job offers when the offer will vanish after a certain date), and simplicity. The more intensive the care in designing and implementing rules for a given market, the more important are the intermediaries who, in fact, are the actual designers. Al Roth puts it succinctly:
Not all markets grow like weeds; some, like hot-house orchids, need to be
nurtured. And some carefully nurtured marketplaces on the Internet are
now among the world’s biggest and fastest growing businesses. 
Intermediaries, then, are active participants in some matching markets while platforms are passive bazaars.3 Disintermediation refers to the elimination of active agents intervening in markets, letting buyers and sellers interact directly. Platforms are ubiquitous on the Internet, providing a space for economic agents to transact.
The revenue or business model in multi-sided markets is complex. Consider the media industry, where information content is the product. There are two basic types of revenue models in this industry. Model A where content is free and revenue is based on advertising; model B has the consumer paying for content and there is no advertising; and model C, a combination of A and B, involves both a subscription fee paid by consumers and advertising. Model A is the basis for the Internet radio station, Spotify; Spotify’s premium version, however, follows model B. The digital versions of The New Yorker, Economist and The New York Times are examples of model C, where content is provided by the platform for a fee, but is supplemented by advertising.
The strategic management of all sides of this platform is even more complicated. Consider the transportation and logistics platform, Uber. It has passengers on one side of the market, driver-partners on the other and the Uber platform managing the entire operation. To be successful, Uber needs a critical mass of customers and drivers or it would have the penguin or coordination problem. Hungry penguins crowd the edge of an ice flow but fear being the first to dive in, lest there be a predator in the water. Similarly, in platform markets there is a coordination problem arising from the risk of backing the wrong horse. On the other hand, powerful network effects arise if current users recruit new users and the value to users is a function of the size of the network of all users. When more passengers use Uber, more drivers sign up, and it quickly becomes a self-reinforcing cycle. Network effects, defined later in this chapter, can reinforce first-mover advantages so that the initial firm entering the market may have a strategic advantage.
Attaining this critical mass is the major hurdle to success. Do you price low? Do you target gatekeeper nodes? Which side of the multi-sided market should you first address - the customers or drivers/advertisers? Each side fears being stranded, in the holdup problem, if the other side (i) becomes too successful and can extract rents, (ii) exits the market, stranding individuals who have developed a dependence on the product or service, (iii) doesn’t make an investment in enriching the platform, or (iv) if the platform provider exercises monopoly power by limiting access to certain parties or by adversely impacting the terms of trade. The debate over network neutrality, where Internet Service Providers exercise discretion over content flowing through their distribution network, is an illustration of this latter point.
More generally, in multi-sided markets, should the platform be the center node in a network of suppliers, sales channels and R&D partners? Is the ideal topology a star network, in which other nodes do not have links with each other? Apple and Samsung were both, until recently, examples of a star network, obtaining information from all other nodes and innovating, without the risk that this information is transferred between them . Subsequently, Apple launched its App Store in 2008, allowing app developers to engage directly with one another via the platform. A firm whose network is more embedded can receive help in an emergency or when manufacturing problems arise, but the chance of radical innovation may be lower. Embeddedness creates a more inward-looking network with few links to distinct outside sources of information.