THE GREAT SOVIET COMPUTER SCREW-UP $ Soviet industry is in big trouble with computers. Its hardware isn't modern. Breakdowns occur endlessly. The telecommunications are terrible. And Soviet managers have lots of sneaky reasons for not wanting effective information systems.
By Daniel Seligman RESEARCH ASSOCIATE Michael McFadden

(FORTUNE Magazine) – WITH SOME KICKING and screaming along the way, the business managers of the Western world have long since adapted to computers. No sizable capitalist enterprise could be competitive nowadays without computer-driven management information systems -- what the data-processing department calls MIS. A problem that Mikhail Gorbachev must find maddening these days, as he looks around for ways to cure the sick Soviet economy, is his country's broad failure in MIS. With every year that goes by, it is increasingly apparent that Soviet industry has fundamental problems with computers. Russian managers aren't exactly kicking and screaming, but they also aren't adapting very well. Computers present several challenges to the Soviet leadership. Presumably it worries about the military implications of the U.S.S.R.'s inferiority in computer hardware. Gorbachev must also be concerned about the computer's implicit threat to the official Communist monopoly on ideas: a few Apple II computers hooked up to printers could make instant best-sellers out of dissident literature. So just about all computers in the Soviet Union are closely guarded in state-run institutions. Americans worry about hackers electronically breaking into institutions and gaining access to various business and military secrets. In the Soviet Union, things are reversed: the state's problem is to prevent any computer users from breaking out of the institutions. Thus far, at least, the regime seems to have dealt effectively with these widely publicized problems. It has been reasonably successful in stealing and otherwise acquiring Western technologies with military applications. And the KGB seems to have had little difficulty in preventing dissidents from getting their hands on printers and other duplicating devices. But the problem posed by computers for Soviet enterprises -- the state-owned counterparts of Western corporations -- is another matter. At the enterprise level, the Russians have generally failed to exploit the fantastic efficiencies made possible by the new electronic technology, and this failure is a major reason for expecting the gap between the Soviet and Western economies to keep widening. Among those who have been most closely tracking the record at the enterprise level is a team at the University of Arizona's College of Business and Public Administration, under the direction of Seymour E. Goodman. The team's members have worked hard at interviewing the occasional Soviet emigre who is knowledgeable about information systems; most of their data, however, are based on exhaustive analysis of Soviet technical journals. Professor Goodman, who was recently in Brussels briefing NATO officials on Soviet computers, believes the Russians are deeply distressed by their shortcomings. ''They sit there and watch this competition between Japan and the West over development of a fifth generation of computers,'' he observed recently. ''They know they can't just sit out the contest. They know they have to at least look as though they're doing the same kinds of things, but that's not easy.'' Their problems begin with the fact that the U.S.S.R. has trouble producing modern computers: its leading entries are copies of mainframes that IBM stopped making six to ten years ago. And even these copies are inferior to those produced in East Germany. The best computer a Soviet manager could ordinarily hope to get his hands on would be an East German ES-1055 model, roughly similar to a third-generation IBM System 370. Most Soviet enterprises still lack mainframes of any kind. William K. McHenry, a colleague of Goodman's and author of a recent prodigious Ph.D. dissertation on management information systems in the Soviet Union, estimates that only 7.5% of the country's industrial enterprises -- 3,300 out of 44,000 -- had mainframes in 1984. To be sure, this minority includes a disproportionate number of large, high-priority enterprises. Even so, the evidence suggests that most sizable plants in the Soviet Union are still getting by without mainframes. McHenry, who will be an assistant professor at Georgetown's School of Business Administration beginning this fall, has assembled data showing that only one-third of large plants (''large'' meaning at least 500 employees) had mainframes in 1984. With or maybe even without rounding, the comparable figure for the U.S. would be 100%. The systems in use in the U.S. typically enable managers to instantly call up an infinity of data about payrolls, inventories, production, sales, and anything else useful. American managers take it for granted that they can function more effectively if these data are accurate and timely. But Soviet managers are frequently under pressures to falsify data, and so they are understandably nervous about computerized systems for keeping track of what's happening. A SOVIET MANAGER who had a mainframe and was seriously trying to develop a management information system would confront some discouragements almost unimaginable in most Western countries. One is the quality of Soviet telecommunications. The basic telephone system in many areas can barely support 200- or 300-baud links (the baud being a measure of transmission speed), and even at that low level disconnections occur endlessly. To attain the 1,200-baud standard taken for granted by an ordinary personal-computer user in the U.S., the Soviet manager would probably need a dedicated phone line. Getting one would require him to have clout in the Ministry of Communications. Another large discouragement would be the level of service, training, and repair offered by the computer's provider -- that is, by the state. The level is dauntingly low. Furthermore, the manager acquiring a mainframe would have to deal with a dizzying variety of state organizations, some concerned with design work and software, some with telecommunications, some with peripherals, some with preparation of computer rooms. The computer rooms have been a particularly annoying problem. McHenry reports that nobody in the Soviet Union makes complete computer rooms, and managers are frequently encouraged to follow the designs specified by an organization ponderously labeled the Central Scientific Research and Experimental Design Institute of Industrial Buildings and Structures. For many years the institute recommended a false floor made of aluminum, said to be produced at a plant in Latvia. ''Actually,'' McHenry notes in the dissertation, ''this plant would only deliver raw sheets, which then had to be assembled by the customer himself; the customer was obligated to provide the raw aluminum in the first place, and the wait was five years.'' Nowhere in the picture is there any equivalent of a vendor willing to take total responsibility for getting the equipment up and running. So a lot of expensive hardware in the Soviet Union is in the ''arrived but not yet working'' mode: McHenry's dissertation cites ''cases . . . where the computer has been left 'under the stars,' gnawed by animals, or even left in boxes in the snow for longer than a year.'' Even after they're first up and running, the computers in use tend to break down more than would be acceptable in the West. Published data on mainframes + used by the Ministry of Power and Electrification -- presumably a preferred customer -- show that in 1983 most had a ''mean time to failure'' of only 130 or 140 hours. In other words, they were breaking down once a week or so. Among the more interesting black markets in the Soviet Union these days are various covert enterprise-level deals to get around these problems. For example, the desperate manager can opt for off-the-books repair and servicing of computers by freelance technicians. He can also go out and buy stolen computer components. Most managers who do have information systems run them with their own teams of specialists operating out of the enterprises' in-house computer centers. One alternative is to turn to a local service center -- an institution somewhat similar to the time-sharing centers that flourished in the U.S. some 15 years ago. The service centers are strenuously encouraged by the state, and McHenry notes that the Soviet Communist party is a prominent customer. Some party officials originally viewed them as the ideal way for enterprises to use computers and also as a convenient way to ensure that information systems would be standardized throughout the Soviet Union. But years of squabbling over which ministry would control the service centers undermined this goal. The squabbling left the enterprises free to set up their own arrangements and left the state looking glumly at thousands of different and incompatible information systems. THE SERVICE CENTERS have grown relatively rapidly in the past few years, but they have some large built-in disadvantages. One, which was also a problem for their predecessors in the U.S., is that it's much more prestigious to have your own computer. A second problem, one that Soviet managers presumably think deeply about, is that you can never really be sure what will happen when people outside your own organization get their hands on your fishy data. Finally, the service centers are badly handicapped by the phone-line problem, which makes it hard for them to rent computer time to remote-access users. Many Soviet managers using the service centers are reduced to sending their tapes or punch cards across town by messenger. Soviet managers have many perverse incentives to misuse computers. One that nullifies productivity gains in many sectors of the economy is the regime's guarantee of employment to all workers. This means that the manager who acquires a mainframe can't easily use it to replace all the workers with % abacuses who are now on his payroll -- and his plant might literally have hundreds of them. Soviet enterprises typically consist of numerous small fiefdoms, each with its own statistical department, and the evidence thus far suggests that management information systems have not really dented these fiefdoms. Their survival is additionally ensured by the organizational rigidity of Soviet enterprises: an enterprise of a given size in a given industry is ordinarily required to have certain well-defined line and staff relationships. These requirements are intended to make it easier for the ministries to supervise the many enterprises under their jurisdiction, but, of course, the effect is to prevent managers from devising new organizational arrangements suitable to the computer age. Another obstacle to productivity gains is the tidal wave of bureaucratic forms that flows over Soviet managers. ''What we see over and over again,'' McHenry remarked recently, ''is the enterprise getting the data out of the computer and then turning it over to clerks who laboriously hand-copy it onto all these forms.'' As you might assume, the overriding obstacle to rational use of computers in the enterprises is the managers' recurring need to falsify much of their data. The garbage-in, garbage-out phenomenon is pervasive, and there is little the ministries can do about it. Soviet managers frequently have to make purchases in various black markets to get needed supplies. Knowing that such illegal purchases are often necessary, they may set up a slush fund by diverting money from, say, the payroll account; to make this possible they might overstate the number of workers at the enterprise, which would enable them to get extra wages from the ministry in charge. Meanwhile, their computers would inevitably be spitting out garbage. In attempting to gauge the benefits of management information systems, the state has a series of formulas that concentrate on measuring output and unit production costs before and after installation of the systems. Soviet data published several years ago proudly state that in 1965-70 some 680 million rubles were saved by these new computer systems; the savings are said to have risen to 2 billion rubles in 1971-75 and were expected to be 3.8 billion rubles in 1976-80. Goodman and McHenry regard the formulas as something of a joke and the savings as largely fictitious, and they have a fair number of details to indicate that the Soviet authorities repeatedly measure the wrong things. They note, for example, that computer executives' bonuses in many sectors depend on the number of ''tasks'' being performed by their systems and also on the number of ''subsystems'' put in place. So instead of developing a comprehensive accounting system, say, the managers affected opt for lots of subsystems, which may be unrelated and cumulatively inefficient. THE PRESSURES that drive managers to cook the books are clear enough. The economic system in which they operate features unrealistic planning targets, causing chronic shortages of resources -- so they hoard resources, doing their best to conceal from the ministries the true state of inventories and the true number of production workers. The system provides bonuses for fulfillment of the plan but also tends to raise required norms for anyone who overfulfills it -- so the managers suppress any data that might suggest they were capable of increased output. On the other hand, they tend to overstate performance when norms are not being met. With all these fishy data swimming around, the managers tend to shy away from highly integrated information systems, which pull together many different kinds of data and would enable an overseer to examine a lot of different relationships in an enterprise. The more integrated the system, the more likely it is that falsified data will entangle the managers in obvious contradictions. In an integrated system, the manager who overstates production in order to meet his goals will find himself also obliged to overstate sales, leaving him ultimately responsible for revenues that he doesn't actually have. Soviet managers are especially leery of database management systems -- software that combines information from many different company operations. Some such software, including the American program called Total, has been copied with minor modifications. (McHenry notes evidence that some Western programs are copied illegally in Bulgaria, where the accompanying manuals are also translated into Russian.) Total is a product of Cincom Systems, a privately owned company based in Cincinnati, and is probably the most popular database management program ever offered for mainframes. However, the prospects of the Soviet copy are plainly limited. What's ironic about the wide-ranging Soviet problems with computers is that they were once widely viewed as a natural and perhaps even indispensable ally of central planning. In 1971 the Communist party essentially endorsed this proposition: the Ninth Five-Year Plan established the goal of creating a vast computer network that would link organizations at every level of the economy. Timely data would flow up from the enterprises to the ministries, which could incorporate the data into the planning goals flowing back down to the enterprises. Subsequent party congresses have re-endorsed this concept, while not specifying the date at which it should be implemented; the Twelfth Five- Year Plan, to be adopted next year, will undoubtedly repeat this exercise. It is still official doctrine that socialism is the ideal system for introducing computers.Yet it is hard to believe that any rational Soviet official still expects to see the day when computers make central planning work. The Soviet economy's failure to assimilate computers is not only an economic disaster. The failure also has a symbolic dimension. Communism is, after all, an idea claiming to stand for progress, and computers are the quintessential symbol of progress. The performance of Soviet managers in the face of the new technology can be taken as another threat to the legitimacy of the regime.

CHART: Text not available.