For Burroughs Corp. and Sperry Corp., high noon approaches.
Armed with powerful computers and backed by sophisticated subcontractors, the two giant data processing companies are a month away from a showdown in a $6 billion duel for one of the biggest computer contracts ever awarded.
One of them will walk away with as much as 20 years of lucrative Air Force business and the other will go home empty handed as the result of a Pentagon procedure never before used to choose a government computer supplier.
It's called a computeoff.
As in playoff. It's a sudden-death match that pits Burroughs and its software and systems engineering subcontractor, Planning Research Corp. of McLean, against Sperry and its associates from Computer Sciences Corp., another big local government contractor.
At stake is the job of replacing the central computer systems at about 100 Air Force bases around the world--nearly 150 computers in all, plus software, systems support, engineering and everything else needed to modernize the bases' electronic data processing systems.
Rather than simply ask for bids or solicit competing proposals, the Air Force is spending $50 million to have each of the teams build and operate a prototype of the system they want to sell the government.
For months, managers, engineers, programmers and technicians from PRC, Computer Sciences, Burroughs and Sperry have been running their gear through tests at an Air Force base in Montgomery, Ala.
The Air Force has been grading the competition on a variety of qualitative and quantitative scales--20,000 tests in all--and will make its decision Feb. 1.
By staging the competition, the Air Force hopes to reduce the risks of cost-overruns and technical snafus. By actually building and operating prototype computers before getting the contract, the companies and the Air Force are trying to find out what will and what won't work, and what it will cost.
Air Force officials contend the concept reduces or eliminates many of the uncertainties in such a large contract.
The computeoff has added $181 million to the cost of the contract, and perhaps delayed installation of the new system for a year or two while the two competitive computers are tested. The Air Force argues that those negative factors will be made up many times over in savings of time and money later, because of the experience before the main contract is actually awarded.
A somewhat similar system has been used to choose aircraft, by awarding contracts to two rival manufacturers to build planes and then putting the models to the test in a "flyoff."
Playoff, flyoff, computeoff--the rules are the same: The winner gets the contract. The loser gets a handshake.
"The objective of competition is exactly that--one's going to win, one's going to lose," says Air Force Col. Robert L. Hedges, the hard-nosed program manager of the project.
"At the end of all this, the Air Force in essence is going to say, 'Team A, you win, and Team B, thanks for coming,' " says an official of one of the companies involved in the competition.
That's one of the most explicit things any of the companies is willing to say about the computeoff at this stage. With decision day approaching, both teams worry that the slightest misstatement will tilt the Air Force in favor of the other side.
Neither wants to talk about details of their proposed system. Nobody wants to voice any criticism of the Air Force or of the computeoff concept--although it becomes clear in interviews that the companies are not totally enamored of the sudden-death format.
The official company spokesmen are outwardly enthusiastic about the computeoff concept itself. "It's good from the standpoint of the customer, because he knows he gets a system that works," said Harry Traylor, Sperry's field director for federal projects.
His counterpart at Burroughs, Robert Johnson, general manager of the company's Air Force systems division, said, "The competitive situation that the flyoff produces is a pretty good way to keep the pressure on until the end . . . I think it's going to produce a better system for the Air Force, no doubt about it."
But some officials of the companies indicated they are not altogether pleased with the competitive format, in part because of the high risks it places on the contractors. The companies have been forced to assemble teams of personnel and commit equipment that would not be necessary in a traditional contract proposal.
Burroughs has formed a new Air Force systems division dedicated to the computeoff. The 100-member group evolved from the team that originally drew up the proposal for Burroughs' entry. As Burroughs "second" in the duel, Planning Research has its own team that's trying to outperform the Computer Sciences crew supporting Sperry. None of the companies will speculate on what will happen to its team members if they lose the competition.
The Air Force itself is also trying to keep the contest fair and avoid slip-ups: Contracts with the companies stipulate that they not speak to the press without Air Force permission, and the Air Force wants to know in advance what questions reporters will ask the contractors. In a competition where subjective decisions may outweigh such traditional measures as cost, nobody wants to make any mistakes.
About all that officials of Sperry, Burroughs, PRC or Computer Sciences will say is that they think things were going well. Each team is confident of victory.
The Air Force's objective is to buy a powerful computer that can be used for all the mundane tasks a computer can do around an Air Force base--such as payroll, inventory and basic record-keeping--and some of the not-so-mundane things, like communications. The new computer system has been tagged "Phase Four," for its place among the progression of Air Force computers and proposed computers.
The Air Force bases now have a cobbled combination of Burroughs and Sperry Univac equipment, some of it 20 years old. The equipment has been modified through the years by a series of technical updates and, more significantly, software evolution.
The Air Force wants a new computer system from a single manufacturer that can be upgraded and expanded over the next 20 years, that uses microprocessor-based remote terminals and that represents the state of the art without wandering too far from standard computer technology and to meet a variety of technical objectives.
A key requirement is that the new computer operate pretty much like the old system, with similar software. That's to save the Air Force the job of retraining its computer personnel.
That seemingly simple requirement has been a considerable challenge to the computer companies, which had to translate 1.3 million lines of computer programming language from the old system to the two proposed new systems.
The two systems' ability to handle that transition is the subject of many of the key tests in the computeoff, which reached its most intense phase this fall, after two years of preparation.
The head-to-head competition concept--as applied to anything besides warplanes and missiles--has its origins in an Office of Management and Budget recommendation that contracts be solicited on the basis of missions rather than solutions. In other words, the government would tell contractors what job it wanted done, not how to do it.
Around the Pentagon, they use the parable of the trucks and the zeppelins as an example: Instead of giving bidders precise specifications of 50 large trucks to haul military supplies, a service might just tell manufacturers exactly what has to be carried and where. The bidders' can then decide whether that need is best met by 100 small trucks--or one very large zeppelin.
The idea, says Col. Donald W. Sawyer, chief of the policy and procedures group in the Air Force's directorate of computer resources, is to "go out with a functional problem and let industry tell us what the solution is. . . . They're smarter than we are, really." In that sense the computeoff is not a perfect example of the procedure, because of the stringent requirements for adapting the existing data-handling systems.
Tapping additional expertise isn't the only reason for using the competitive format. The government also hopes it will lead to more realistic estimates of what will be needed to maintain and modernize equipment over the life of a contract. The idea is to prevent recurrence of some unfortunate past experiences in which equipment was purchased with little consideration of what would be needed to keep it operational. A disastrous Air Force computer procurement a few years ago is cited as an example.
"The cost of having both contractors complete the first phase of the contract will be returned many times over to the government in the increased competition and reduced prices bid by the two contractors," Hedges says. "Also, having the contractors deliver a product to the government which can be thoroughly tested in an Air Force environment significantly reduces the risk of schedule delays following the production-buy decision."
The procedure ultimately could also lower the overall price of the contract by forcing a bidder to pare a proposal to the bone to stay competitive. "You don't know how much he would charge you if he didn't have somebody to beat," Sawyer says.
"To say nothing of the risk you're avoiding by having two guys prove to you they can do what they say they'll do before you give them the contract," he adds.
What future contracts will be awarded through competitive tryouts has not been decided, but spokesmen predict it will become fairly common in the acquisition of data-processing equipment.
"It kind of depends what you're trying to acquire," said Sawyer. "There are straightforward kinds of buys that competition wouldn't help much."
Contracts as big as the Air Force computer job are rare even at the Pentagon and most people involved agree that future duels are likely to be much less grandiose. But the concept is definitely proving its worth, they say.
"Our experience with Phase Four has caused us to reevaluate the kinds of acquisitions we're going to make in the future and how we should structure them," Sawyer says. "We've got a good mechanism to make this an institutionalized kind of approach."