2 networks, 2 philosophies.

IN 1969, when Eastern States Bankcard Association was putting its authorization system in place, the telephone company crossed its lines with those of a Long Island bookmaker.

"The bookie was authorizing credit card purchases, and we were getting calls about horses," said Philip Verdi, who worked for the pioneering credit card processor, and who is now executive vice president in charge of electronic services at MasterCard International.

Veterans of the early days of credit card technology have many such tales of missteps and mistakes, but out of them grew two of the fastest, most sophisticated data communications networks in the world.

Typical of their intense rivalry, MasterCard and Visa don't see eye-to-eye on communications strategy.

MasterCard has embraced the distributed approach, saying its system allows data to flow more smoothly, avoids bottlenecks, and is less expensive to run. Visa says its more centralized "star" configuration is more reliable and powerful and capable of accommodating changes and upgrades more efficiently.

The roots of the opposing strategies were planted in the early 1970s, when both organizations began their automation planning.

They started at the same lowtech place. Everything was done by telephone--a merchant called his bank or a processing company, which in turn would get an authorization by calling the cardholder's bank. They did well to complete the process in five to 10 minutes.

A California shop owner might be told on a Friday afternoon to call back on Monday to get an authorization from a New York bank that had closed for the day.

Clearing and settlement were equally difficult. Merchants sent bundles of paper sales drafts to their banks, and the banks either sorted the drafts sent them out to a company like Eastern States. The tickets were sorted by hand or machine for mailing to the appropriate card-issuing banks.

A purchase might not show up on a cardholder's statement for two to three months.

As losses from fraud and inefficiency mounted, it was clear that authorizations and settlement had to be speeded up. Authorizations were the first priority, and thus was born the troublesome concept of the floor limit: to shorten customer waiting times, issuing banks allowed purchases of up to $150 to go through without authorizing them.

"The credit limits weren't working," said Irwin H. Derman, one of Visa's first employees, currently vice president of VisaNet market development. "Customers were running over their credit limits, stolen cards were being authorized, and banks were writing off millions of dollars."

In 1972, the fledgling card associations discussed setting up a universal authorization system, Mr. Derman said. They even had a proposal from American Express Co. to operate the bank card payment system.

Visa--then National BankAmericard Inc.--quickly decided to go its own way.

BankAmericard had only 45% of the bank card market, then under $4 billion in outstandings. Dee W. Hock, the group's first president, argued that it had to differentiate itself from MasterCard, or it had no future. Mr. Hock got approval to spend $3 million on the authorization system that became known as BASE I. It was up and running by April 1973.

NBI built the system using the centralized, mainframe-based computer technology that was the most reliable one available at the time. BASE I the initials meant BankAmerica Service Exchange-made the cumbersome verbal communication unnecessary in authorizing a credit card purchase.

When a merchant's bank called the card-issuing bank, the call was routed to a host computer in the card association data center in San Mateo, Calif. The system would block transactions from cards that had been reported lost or stolen.

Banks were able to set certain limits on card transactions. They could say that only a certain number of transactions or dollar amount could be authorized within a specified period, between one and four days.

Interbank Card Association, the group then doing business with Master Charge, followed with its Interbank National Authorization System, or INAS. It was, like BASE I, mainframe-based, and it obviated the need for voice calls to bank clerks. The central processor was in St. Louis.

INAS linked over 50 locations, including card-processing associations like Eastern States, independents like First Data Resources, and big issuers like Citibank that did their own processing.

Mr. Verdi said the real systems differences between Master Charge/MasterCard

and BankAmericard/Visa did not emerge until 1984, when MasterCard built Banknet, a payment network built around packet-switching technology.

Banknet, still at the core of MasterCard's operation, is an authorization, clearing, and settlement network in one.

MasterCard chose distributed, packet-switching systems--the name refers to the bundles of data that accumulate before being transmitted in "packets"--simply because they were a hot item.

"I'm glad, now, that we have a distributed system," he said. "We don't have the overhead of big processing centers, and if I lose one switch, everything is routed automatically to another switch."

Even with the problem of 24-hour authorizations solved, banks were still transferring huge piles of paper for clearing and settlement, and banks were "always out of balance," Mr. Derman said.

National BankAmericard spent $7 million to launch the BASE II settlement network in 1974, and Master Charge followed the next year with Interbank Network for Electronic Transfer, or INET. They replaced paper with electronic transmissions.

Both associations developed techniques for dealing with fraud. Visa in 1982 introduced dial terminals, which allowed cards to be checked automatically against electronic data bases, rather than being compared against printed hot-card bulletins.

In 1984, MasterCard mandated holograms on the front of credit cards, which helped reduce counterfeiting. Later antifraud techniques involved putting verification codes on the magnetic stripe.

Throughout the 1980s, Visa had a reputation for focusing more on technology, while MasterCard's strong suit was seen as marketing. Although Visa will not say how much it spends on technology each year, it said that its current facilities would cost over $350 million to duplicate.

Visa is still into centralization. Transactions are funneled from around the world to IBM mainframes in one of the two "supercenters" in McLean, Va., and Basingstoke, England, or to one of two regional centers in Asia and California. These four data centers handle all the hundreds of billions of dollars in sales handled by Visa in a year.

MasterCard relies on 14 switching points around the world. Transactions are routed around the network any number of ways, depending on what will be the fastest way at any given time.

As the two associations evolve, their networking differences are blurting. Visa is distributing more of its processing, in part to be closer to new and growing markets. Visa set up a regional processor in Japan, which avoids having to send all transactions to one of the supercenters, and it addresses Japanese banks' concerus about their data flowing outside the country.

Visa may set up more regional processors if nations require it, said William Chenevich, group executive vice president for information technology.

Visa also is distributing computing power to give its banks more information. One example is the VisaVue product, which gives banks access to mainframe data about credit card spending patterns that the banks can analyze and use to do target marketing.

Technologists say valid arguments can be made in favor of either technology approach, and the associations are not shy about accentuating their differences.

"Our goal is to be the best payment service, not the biggest," said Mr. Verdi. "We don't need to process every transaction in the world."

Visa is aggressively going after transactions that don't fit the traditional credit or debit card mold, such as automated clearing house items. VisaNet's costs can come down by utilizing more of its capacity.

Visa's recent upgrade, Payment Service 2000, was aimed at giving merchants and banks better service and access to information. A typical application let hotels automatically authorize payments from good customers without having to physically place the customer's card in a reader.

The emphasis is moving toward new products, Mr. Chenevich said. Visa plans to offer a magnetic stripe card with prepaid value, in effect an "electronic travelers check." As a cardholder draws funds from teller machines around the world, VisaNet will tell the host computer to update the remaining value, which is available in several currencies.

As new products come one and telecommunications becomes increasingly sophisticated, "you don't need to eliminate the infrastructure behind VisaNet,"

Mr. Chenevich said.

MasterCard is on a campaign with its European partner, Europay International, called Programme Global, aimed at "taking advantage of tools that didn't exist 10 years ago," Mr. Verdi said.

MasterCard also is considering using a "virtual" network worldwide that would deliver a more flexible telecommunications arrangement. And there would be new schedules that would allow it to settle transactions in a given region of the world at the close of its business day, rather than working around the close of business in New York.

Visa is experimenting with using the Internet. It is also looking at Asynchronous Transfer Mode technology that can process huge quantities of voices, data, and video at the same time.

The big question for the future is how to let technology span the world, especially areas with lesser developed infrastructures. Mr. Chenevich said, "The challenge is balancing out the demands brought to us by the increasing pace of technological change, and by the emerging markets. Do we ride the horse of technology, or do we try to adapt our system to the emerging markets."

For reprint and licensing requests for this article, click here.
MORE FROM AMERICAN BANKER