Sabre (travel reservation system)
Based on Wikipedia: Sabre (travel reservation system)
The Chance Encounter That Changed How We Book Flights
In 1953, two men named Smith sat down next to each other on an American Airlines flight from Los Angeles to New York. One was Blair Smith, a high-ranking IBM salesman. The other was C. R. Smith, the president of American Airlines. They discovered they shared more than a surname—they shared a fascination with what computers might be able to do for the chaotic business of booking airplane seats.
That conversation would eventually produce SABRE, the Semi-Automated Business Research Environment. Before SABRE launched in 1964, booking a single airline reservation took an average of ninety minutes. Afterward, it took seconds.
This is the story of how the travel industry went from handwritten tickets and shouted phone calls to the instant global booking systems we take for granted today.
The Maddening Reality of 1950s Air Travel
Imagine trying to book a flight in 1935. You call your local airline sales office. The agent there picks up the phone and calls another agent—the booking agent—stationed in whatever city your flight departs from. This second agent is standing around a rotating file cabinet with seven colleagues. Eight people total. That's the maximum who can physically fit around this piece of furniture, and it represents the hard upper limit on how many reservations the airline can process at once.
The booking agent spins the file, finds the card for your flight, checks if seats are available, marks one as taken, and then manually writes out your ticket. Ninety minutes later, if you're lucky, you have a confirmed reservation.
American Airlines called this the "request and reply" system. It was unbearably slow.
In 1939, someone in American's Boston office had a revelation. What if sales agents just sold seats whenever customers asked, without waiting for confirmation? They could report what they'd done later. This "sell and report" approach was faster, but it created a new problem: overselling. Once a flight got nearly full, headquarters would blast out a "stop sale" message to all locations. But by the time that message arrived, some agents had already sold seats that didn't exist. Customers would show up at the airport only to learn their confirmed reservations were worthless.
By 1952, American had deployed something called the Magnetronic Reservisor—an electromechanical computer with a single magnetic drum. Each memory location stored how many seats remained on a particular flight. Multiple operators could check seat counts simultaneously. Progress! But the system still required a human at each end of every phone call, and ticket-writing remained entirely manual.
The jet age was coming. Boeing was developing the 707. Airlines would soon be filling aircraft with hundreds of passengers instead of dozens. American's reservation system was going to collapse under the weight.
Two Smiths, One Vision
Blair Smith and C. R. Smith talked the entire flight from Los Angeles to New York. The IBM salesman learned that American's president wanted something more ambitious than a seat counter. He wanted a system that could track everything about a passenger's reservation—not just whether seat 14A was occupied, but who was in it, where they were connecting, what fare they'd paid.
C. R. Smith gave Blair Smith an assignment: tour American's reservation center at LaGuardia Airport, then write a letter recommending what IBM should build. When Blair Smith returned to IBM headquarters, he briefed Thomas J. Watson Jr., IBM's president. Watson told him to take the tour seriously and send a copy of his recommendation letter directly to Watson himself.
IBM was already working on something that would prove remarkably relevant: the Semi-Automatic Ground Environment, or SAGE, a massive Cold War project for the United States Air Force. SAGE used networked computers to coordinate radar sites and fighter interceptors across the continent. Teleprinters around the world fed information into the system, which processed it and sent orders back out. It was one of the first online systems ever built—computers talking to each other and to humans in real time.
The two Smiths realized SAGE's architecture could revolutionize airline reservations. Instead of phone calls, teleprinter machines would sit in every ticket office. Agents would type requests directly into the network and receive responses on the same machine. No booking agents spinning file cabinets. No shouted conversations over crackling phone lines. Just terminals connected to a central computer that knew everything about every flight.
Thirty days after their chance encounter, IBM sent American Airlines a formal research proposal.
The Decade of Development
A formal development agreement was signed in 1957. The first experimental system flickered to life in 1960. But getting from experiment to reliable production system would take four more years and cost far more than anyone had budgeted.
IBM had actually been beaten to market. In 1958, Sperry Rand launched the UNIVAC Air Lines Reservations System for Eastern Air Lines. But UNIVAC could only track seat inventory in real time. Agents still had to prepare separate handwritten tickets and manually keypunched cards. It was a half-measure.
Meanwhile, IBM had gotten ambitious—perhaps too ambitious. While developing SABRE for American, the company signed contracts with Pan Am and Delta Air Lines for similar systems. IBM's internal code name for the American project was actually SABER, which prompted American to come up with its own name: the Semi-Automated Business Research Environment.
Then IBM made a costly mistake. The company implemented DELTAMATIC on the IBM 7070, PANAMAC on the IBM 7080, and SABRE on the IBM 7090. Three different mainframe computers. Three incompatible systems. In retrospect, building all three on the 7090 would have made everything simpler.
The programmers working on SABRE at IBM's facility in Briarcliff Manor, New York, could see Sing Sing prison from their windows. By 1962, as delays mounted and costs ballooned, they had developed a dark joke: the difference between us and the prisoners is that the prisoners know when they'll be getting out.
SABRE finally launched in 1964. It was the world's first fully operational computer reservations system. More than that, it was the world's first online transaction processing system—the ancestor of every digital service you use today that responds to your inputs in real time. At the time, observers called it "the world's largest private real time commercial data processing system."
Seven Thousand Reservations Per Hour
The original SABRE ran on two IBM 7090 computers in a "state-of-the-art data center" in Briarcliff Manor. One handled real-time processing. The other served as backup and ran batch jobs. American Airlines agents interacted with SABRE through computer terminals manufactured by Raytheon.
The system could process 7,500 reservations per hour.
That number is worth pausing on. Remember the rotating file cabinet that could support eight operators maximum? Remember the ninety-minute average to complete a single booking? SABRE could handle over a hundred reservations per minute. The bottleneck had shifted from the system to the speed at which humans could type.
IBM packaged what it had learned into a product called PARS, the Programmed Airline Reservation System. PARS ran on the IBM System/360 mainframe family—Models 40 through 75, though Model 65 was the airline industry's favorite. The operating system beneath PARS evolved into something called ACP (Airlines Control Program), and eventually, in 1979, became TPF (Transaction Processing Facility).
By 1971, PARS was the industry standard. Nine of the top ten major U.S. airlines were using it. Even United Airlines, which had experimented with UNIVAC, gave up and deployed Apollo, a PARS-based system. American migrated SABRE itself to PARS between 1971 and 1973.
During this migration, the SABRE data center moved from New York to an underground facility in Tulsa, Oklahoma—presumably harder to bomb than a suburban New York office building, should the Cold War turn hot.
The Weapon Disguised as a Tool
In 1972, a man named Max Hopper became director of SABRE. He had a vision that would transform the system from American's internal tool into something far more powerful and, eventually, far more controversial.
Hopper pioneered SABRE's deployment beyond American Airlines itself. Starting in 1976, travel agents could access SABRE directly. By the mid-1980s, SABRE processed 45% of all airline reservations in the United States. Fifty thousand travel agents used it daily.
Here's where things got interesting—and ugly.
A 1982 study by American Airlines revealed something remarkable about human behavior. Travel agents selected the flight appearing on the first line of search results more than half the time. Ninety-two percent of the time, they booked something from the first screen.
American Airlines owned the system. American Airlines could control what appeared first.
The temptation was irresistible.
Corruption by Algorithm
At first, American's manipulation was subtle. The airline tweaked the relative importance of factors in its ranking formula—how much weight to give flight duration versus departure time versus number of connections. Each tweak just happened to favor American flights.
Then they got bolder.
In late 1981, New York Air launched a new route from LaGuardia to Detroit, challenging American in an important market. Shortly after the new flights loaded into SABRE, they mysteriously started appearing at the bottom of search results. New York Air's reservations dried up. The airline was forced to cut back from eight Detroit flights a day to zero.
On another occasion, SABRE deliberately hid Continental's discount fares on 49 routes where American competed. A Sabre employee had been assigned to write a program that would automatically suppress any discount fares loaded into the system by competitors.
Congress investigated. In 1983, Bob Crandall, American's president, testified and offered a defense that was remarkable for its brazenness: "The preferential display of our flights, and the corresponding increase in our market share, is the competitive raison d'être for having created the system in the first place."
We built the system so we could cheat with it, Crandall was essentially saying. Why are you surprised?
The government disagreed with Crandall's logic. In 1984, regulators outlawed biased search results in SABRE and in similar systems owned by United and TWA.
The Bias That Kept Coming Back
But bias proved difficult to eliminate entirely.
In 1985, the Department of Justice accused American and United of implementing "second screen bias"—their systems displayed competitors' flights, but only after users had already seen a first screen dominated by the system owner's offerings. Both airlines agreed to stop.
In 1986, Delta complained to the Department of Transportation that American was feeding SABRE inaccurate flight schedules. American's flights were listed with shorter flight times than reality, causing the algorithm to rank them higher. If your flight from Chicago to Dallas actually took two hours but you told the computer it took ninety minutes, you'd beat a competitor whose honest two-hour flight time was calculated correctly.
By 1987, Flight International assessed that U.S. reservation systems were "about as close to neutrality as a system could reasonably be expected to get." The regulations had worked—more or less.
Unable to weaponize their systems against competitors, airlines pivoted. They started selling access to reservation systems as a profit center rather than a competitive weapon. The fairness rules were eventually eliminated or allowed to expire in 2010. By then, none of the major distribution systems was majority-owned by airlines anyway.
The European Front
While American regulators were clamping down on bias, European systems remained tilted toward their owners well into the late 1980s.
SABRE tried to expand into Europe and ran into a wall of protectionism. Big European carriers, led by British Airways, refused to grant SABRE ticketing authority for their flights—even after SABRE had obtained clearance from the International Air Transport Association's Billing and Settlement Plan for the United Kingdom in 1986.
American Airlines took British Airways to court. The lawsuit alleged that after SABRE arrived on British doorsteps, British Airways immediately offered financial incentives to travel agents who stuck with Travicom, a rival system. British Airways would tie its override commissions—bonus payments to agents who sold lots of BA tickets—to Travicom usage.
Travicom had its own interesting history. It was created in 1976 by Videcom, British Airways, and British Caledonian as the world's first multi-access reservations system—meaning multiple airlines could participate. The technology eventually became part of Galileo UK. At its peak, Travicom connected 49 international carriers, including giants like British Airways, TWA, Pan Am, Qantas, Singapore Airlines, Air France, Lufthansa, and SAS.
From Monopoly to Commodity
American and SABRE eventually parted ways. On March 15, 2000, they became separate companies. Sabre Holdings traded on the New York Stock Exchange under the symbol TSG until going private in 2007. The company dropped the all-caps styling—SABRE became Sabre—signaling its evolution from American Airlines' secret weapon into an independent technology company.
Sabre Holdings launched Travelocity in 1996, one of the first websites that let consumers book travel directly, bypassing travel agents entirely. Expedia acquired Travelocity in 2015.
Today, Sabre operates three business units: Sabre Travel Network (the descendant of the original global distribution system), Sabre Airline Solutions (technology for airlines themselves), and Sabre Hospitality Solutions (technology for hotels). The company is headquartered in Southlake, Texas.
The Larger Lesson
The SABRE story illustrates something profound about technology platforms. When you control the infrastructure that an entire industry depends on, you face a constant temptation to tilt the playing field. American Airlines built SABRE to solve a genuine problem—the chaos of manual airline reservations in an era of explosive growth. But having built it, the company couldn't resist using it to crush competitors.
The same dynamic plays out today with search engines, app stores, and social media algorithms. The platform owner says they're providing a neutral service. The results somehow favor the platform owner's interests. Regulators investigate. The platform owner defends the bias as the natural reward for having built the platform in the first place.
Bob Crandall's 1983 defense—we built the system so we could cheat with it—was at least honest. Most modern platform operators dress up the same impulse in more sophisticated language.
But there's another lesson here too. SABRE genuinely transformed how the world books travel. The shift from ninety-minute manual reservations to instant computerized booking wasn't a minor improvement—it was the foundation for the entire modern travel industry. Package tours, frequent flyer programs, discount airlines, online travel agencies—none of it would work without systems like SABRE humming away in the background, processing thousands of transactions per minute.
Two men named Smith sat next to each other on a plane and changed the world. One was trying to sell computers. The other was drowning in file cabinets and phone calls. They figured out that a Cold War air defense network and an airline reservation system had the same basic architecture: inputs from many locations, central processing, outputs back to many locations.
That insight created an industry.
The cheating came later.