In his book, The Trouble with Computers: Usefulness, Usability, and Productivity, MIT Press 1995, Thomas K. Landauer reveals a "productivity paradox". Using data on 254 firms from Paul Strassmann, Landauer finds that none of Strassmann's case histories showed "any conclusive evidence of positive value of computer use" -- in fact, they demonstrated "a slight tendency for companies that spent more [on computers] to do worse". This suggests the cost of computers exceeds their benefits, and significantly, since the effect is visible even in overall results.

We've all heard the war stories featuring disasters in computerization. Whether downsizing from mainframes to distributed systems or growing from stand-alone to networked systems seems to make no difference. Downtime, inflexibility, data loss/corruption and hardware failures add serious costs that more than wipe out any hoped-for benefits.

Here we are, 40 years and $1 trillion worth of computer hardware into this "computer age" and still no results. We enter the on-ramp of the information superhighway -- only to find that our wheels have fallen off.

Electricity

Magnet makes clear that he doesn't really take issue with Landauer on past performance -- but he sees the tide turning in favor of computer investment: "For years, info tech didn't seem worth the investment. But at last, some smart companies are figuring out how to make computers pay."

These few companies "have learned they must restructure themselves and how they work as they weave computers into their most basic processes. A technological revolution, in other words, is more than a merely technological matter: It entails an organizational transformation too." But the payoff can be huge: After Caterpillar radically redesigned its work around information technology, their truck engine plant produces twice as many engines with the same workforce.

Both Landauer and Magnet are pointing to an historical nexus in our midst. To gain perspective, consider the introduction and assimilation of an earlier technology into American industry. From its introduction by Thomas Edison, it took about 40 years before electricity paid off -- driving the breathtaking 5% per year growth in productivity during the roaring '20s. In retrospect, it's easy to see why.

At first, manufacturers simply unhooked their steam engines or water wheels and replaced them with electric motors. The organization of the plants remained the same. A central shaft above the rafters was turned by the prime power source outside the building. Inside, individual work stations took power off this shaft by simple pulley, belt, and clutch mechanisms. No flexibility was allowed because of the mechanical constraints. Electric motors turning central shafts made no difference over steam or water power -- and cost more.

Slowly, managers learned to take advantage of the flexibility afforded by electric power to arrange plant interiors more efficiently and to locate larger plants on cheaper land away from the water. The way they worked was reconstructed. Finally, in the '20s, this all came together with a roar -- economists today ascribe 75% of the growth in that decade to electrification. The 40 years of wandering through electricity's Sinai is about the time required to flush a new set of managers through the system.

Computers

Unexpectedly, the answer comes from the dreaded military-industrial complex. Cold War pressures drove multi-decade R&D efforts that led to the integration of information technology throughout the Department of Defense. The armed services were restructured around info tech -- and the way they fight was rebuilt to take advantage of information.

There was no other choice. Faced with always being outnumbered, the U.S. & NATO had to fight smarter -- gather, analyze and use information on a massive scale while denying it to the enemy. The Persian Gulf War demonstrated dramatically what happens when info tech meets old style. The former Soviet Union finally gave up the Cold War once they realized they must scrap the Red Army and rebuild it on the US/NATO model. They just couldn't afford to do so -- neither financially nor socially.

So the military pioneered in these two principles of computer exploitation -- restructure the organization and reorganize the way it works. But there is a third principle -- change the paradigm.

For decades, the Defense Department poured money into Artificial Intelligence (AI) research. By far the most important outcome has been a different understanding of what computers are -- general purpose symbolic processors. And what they are good for -- to simulate, to emulate, to imitate -- that is, to model dynamic processes in the real-world.

Software instantiates these models -- it turns general-purpose hardware into a specific system model. In this perspective, a computer running accounting software is modeling the processes of accounting -- transaction journalizing, general ledger entries, and the like. Another running a word processor is simply modeling a typewriter. This one, computing artillery tables, is modeling internal and external ballistics. That one, operating a robot, models the physics of the arm and its load.

Whether these models are numeric underneath is incidental. The AI enterprise is about modeling aspects of human intelligence -- which nobody thinks is numeric. Symbols in computers can be mapped into numbers, but they can represent any entities we choose just as easily. And arbitrary relationships, laws of combination and modes of interaction can also be programmed in.

Non-numeric models are even more valuable than numeric ones -- the world of human endeavor is overwhelmingly non-numeric. Specifically, software can be developed to represent the structures and procedures of businesses, i.e., to model the dynamic processes of an enterprise and its operations. This is the third principle -- tailor software to model your operations.

This third principle is complementary to the second. Together they are analogous to digging the tunnel simultaneously from both England and France, meeting under the English Channel. Close coordination is essential -- they might easily just slide past one another. Traditionally, those who structure/do operations and those who program computers are so far apart organizationally, they might as well be on different planets.

This is where the first principle comes into play -- restructure to allow and encourage such non-traditional cross-fertilizations. This leads away from hierarchies toward flatter, more distributed structures. Away from functional fiefdoms toward heterogeneous team approaches.

For strategic managers, computers, as stereotypical numeric processors, are no good. And computer programs, as stereotypical generic, mass-market packages, are low-yield sidebars. The Fortune article reports on a study that found Return On Investment in info tech capital averaged 81% while ROI on all other capital averaged 6.3% over a recent five year period. Regardless of a company's size, enterprise-wide system models, tailored by operations managers, hands-on workers and information system designers, keynote the investment strategy for global competition.