The value of these opportunities, however, is difficult to measure. We have had more success at measuring how much business information systems cost. In the United States, general-purpose software represents about 10 percent of the cost of business information systems to the corporate sector. For custom software the cost of invention is about 20 percent. The remaining 70 percent is the “dark area” that Dr. Brynjolfsson mentioned earlier. He said that this unknown portion could be measured in various ways—by its value to successful companies, perhaps, or by examining the demand curve for large changes in hardware and software. He said that it ranged in size from 5 to 10 times the other expenditures in business systems cost.
A new information system might be installed, for example, to permit customers to check their bank balances at home. In addition to the hardware, however, people would be needed to answer the phone and enter information into a computer. Jobs and management tasks would have to be changed through extensive planning, retraining, and testing. These functions would take up roughly 70 percent of the invention costs of the system.
A major concern is to reduce the impact of bottlenecks that may jeopardize the business operation. Dr. Raduchel suggested earlier that 90 to 95 percent of the costs of a major re-engineering project were spent on training, testing, data conversion, and other expenses outside the categories of hardware and software. The business invention more than the technical invention is typically the bottleneck for creating new value. These processes are difficult to measure in cost or time, partly because they do not appear in the accounts of the inventing firm. Dr. Brynjolfsson estimated that a firm might need 5 to 10 years after initial adoption of the new technology to create value. This general rule applies to all upgrades in technology. It applied to new technical capability that came with the invention of the platform mainframe computer in 1964 as much as it applies to the capabilities that came with commercialization of the Internet in 1995.
The first reason it is expensive to change business systems is that it is conceptually more difficult. Typically it is more difficult to understand how to change people’s jobs, how organizations work, and to understand what customers might like than to understand hardware and systems software. The latter is cumulative and can be understood by young people in a deep way as they gain scientific or engineering knowledge. The business side of business information system invention draws on things we do not understand in an organized way—the psychology,