The needs of commerce have been a primary factor in the evolution of information technology throughout history. Records of economic transactions on clay tablets, pictographic expressions of value, and physical tokens in media ranging from stone to plastic illustrate available technologies and states of economic development. Many of today's information infrastructure activities also deal with billing and payment. Most bills are mailed, faxed, and with decreasing frequency, presented in person; payments are generally made by cash, check, credit card, or money order. None of these methods, however, fully addresses the unique constraints or opportunities of doing business in an advanced networked environment. The purpose of this paper is to explore how, on tomorrow's networks, universal, fast, and secure payments can be made and received for services, physical goods and information.
Many of these payment mechanisms have already begun to adapt in response to the conduct of business over networks. But more remains to be done. Further, to ensure universal availability and acceptability, entirely new forms of electronic payment are needed.
Presently, for example, credit card charge information can be entered into a device screen and sent securely over the network from buyer to seller as an encrypted message (e.g., privacy-enhanced mail message). This practice, however, does not meet important requirements for an adequate financial system, such as nonrefutability, speed, safety, privacy, and security. To make a credit card transaction truly secure and nonrefutable, (1) a customer must present his or her credit card information (along with an authenticity signature) securely to the merchant; (2) which must validate that it is dealing with the true owner of the credit card account; then (3) relay the credit card charge information and signature to its bank; which (4) must relay the information to the customer's bank for authorization approval; and (5) the bank must return the credit card data, charge authentication, and authorization to the merchant. Then, and only then, can the actual goods, services, and funds flow.
If there can be a lapse in time between the charging for and the delivery of goods or services -- for example when an airline ticket is purchased well in advance of the date of travel -- the customer verification process can be simpler. In fact, all the relaying and authorizations can take place after the customer-merchant transaction is completed, unless the authorization request is denied. On the other hand, if the customer wants a digital airline ticket, which would be downloaded into a PC or other information appliance immediately at the time of purchase, many message relays and authorizations must take place in realtime while the customer waits. Such exchanges may require many sequence-specific operations such as staged encryption and decrypting and exchanges of cryptographic keys.
If this process is extended to all of the small-dollar services that may ultimately be available over the NII (e.g., $3 pay-per-view movies and $1 video game rentals), the overall processing load on key system components will likely become unmanageable or nonviable. Providing this processing service for numerous $1 and $3 transactions may not be as financially attractive as it is now, when the average credit card transaction is about $60.
Today's electronic transactions systems can be modified and refined to work over the NII. They can also be extended to take advantage of new technological capabilities. For this reason, alternative mechanisms for managing digital cash are to be expected. Some requirements are illustrated in the following scenario set in the NII.
A student is doing research in the electronic library. Using a public communications port on the NII's equivalent of today's pay phone, she launches an inquiry using a knowledge-gathering software agent which roams the world's networks and identifies relevant studies. Some are located through a commercial archival research service with which she has never before dealt. She would like to retrieve copies of these studies to review on her portable knowledge appliance. The archive service informs her that there will be an $8 charge for the copies. The student can't buy on credit since she doesn't have an account with this service and the service doesn't accept credit cards for charges under $10. So she places her university's student smart card containing electronic cash provided as part of a scholarship into her appliance. She transfers $8 worth of these electronic tokens to the research service. The service validates the tokens as authentic and sends the reports to the student. Upon receipt of the requested reports, the student in turn completes the transaction by transferring ownership of $8 in tokens to the service. Later, browsing on an entertainment network, the student sees an advertisement about a movie. She sends a request to the advertising distributor to have the movie presented on her appliance. The distributor notifies her that the charge will be $3.99. She places her university card into her knowledge appliance and transfers $3.99 in electronic tokens. The movie service attempts to validate these tokens and discovers that they are earmarked for specific scholarship purposes, which do not include purchasing entertainment movies. The service returns the tokens to the student with an explanation.
This paper focuses on the concept of electronic tokens. Digital cash is one form of electronic tokens. However, electronic tokens can also be designed as electronic analogues of other forms of payment as well, including checks and credit backed by a bank or financial institution. These latter alternatives are designed to accommodate the many individuals and entities that might prefer to pay on credit or through some mechanism other than cash. The availability of credit is an important stimulant to commerce.
In an electronic token system, tokens can be stored on a user's card or computer, and can be exchanged directly between remote transacting parties. This exchange does not require a fixed network infrastructure, and can be accomplished through an intermittent network connection, e.g., via mobile appliances that use wireless networks. Other electronic payment mechanisms have been proposed that require on-line third-party payment servers to process transactions. These mechanisms can be designed with any of the attributes of electronic tokens, including anonymity. They differ from electronic token systems in that (1) they depend on a network infrastructure and (2) they require the on-line involvement of at least one additional party and, in some cases, multiple parties. Because requiring an on-line third-party connection for each transaction could lead to processing bottlenecks and potentially undermine the goal of reliable use, this paper does not discuss third-party systems in any detail.