My Years with General Motors

A failed engine posed the problem of control

In 1923, General Motors tried to put Charles Kettering's copper-cooled, air-cooled engine into production. In Sloan's retrospective, corporate research had championed the design, and the center committed Chevrolet and Oakland to it before the operating divisions had accepted a production-ready engine. Tests and manufacturing exposed serious problems. The Chevrolet version was withdrawn. The Smithsonian preserves a radiator emblem from the short-lived car and describes the engine as a one-year failure. ¹

For Alfred Sloan, the episode was not merely a bad engineering bet. It revealed an unanswered constitutional question inside GM: when central research sees an enterprise opportunity but a division must build, sell, and repair the product, who can commit the corporation? Research autonomy without operating consent had failed; divisional autonomy without a way to develop shared technology would fail differently. My Years with General Motors builds its account of “decentralized operations with coordinated control” around problems of this kind. ²

The engine story keeps the management system attached to consequences. Customers, dealers, mechanics, factory employees, and divisions carried the cost of a corporate experiment. The incident also lets Sloan portray later structure as learning rather than bureaucracy: the center and the divisions needed defined jurisdictions, evidence that could travel between them, and a forum for decisions neither could responsibly make alone. ³

Durant's federation needed a center it could survive

William Durant had assembled GM from automobile and parts companies through acquisition and personal dealmaking. The federation preserved entrepreneurial energy but could not reliably reconcile brands, capital demands, inventories, and product decisions. When demand fell, local freedom and weak financial coordination threatened the whole combination. ⁴

Sloan's answer was not to replace the divisions with one operating hierarchy. Each division received responsibility for a business, while central committees and staff groups handled enterprise policy, capital, appointments, shared research, and questions that crossed divisional boundaries. The distinction between policy and administration was never perfectly clean—the copper-cooled engine had already shown why—but it supplied a vocabulary for deciding where a question should go. ⁵

Comparable financial information made that settlement governable. Donaldson Brown and other leaders brought financial methods through GM's direct ownership and personnel connection with DuPont. Return on investment, budgets, forecasts, inventories, and capital requests gave the center a way to compare unlike operations without issuing every instruction. The link is a documented institutional transfer, not a general resemblance between large companies. ⁶

Product policy added another kind of coordination. GM placed brands along a price ladder—famously, “a car for every purse and purpose”—so divisions could serve distinct markets while the center limited destructive overlap. Local managers retained room to act, but access to capital, executive appointments, performance measures, and the definition of each brand's territory remained powerful central controls. Decentralization changed how authority operated; it did not make the center disappear. ⁷

The settlement remained contested

Sloan's lucid model can make the final arrangement look more settled than it was. In The Struggle for Control of the Modern Corporation, historian Robert Freeland uses archival records to describe a long contest among owners, central staff, and operating executives. His history of GM's organizational order complicates the memoir's cleaner division of policy from operations and its portrayal of departures as exceptions. The multidivisional form was a negotiated governance settlement, not an idea installed intact. ⁸

The people with recognized standing in that settlement were principally owners and managers. Factory workers did not acquire decentralized authority because a division did. In the 1936–1937 Flint sit-down strike, workers occupied GM plants to force recognition of the United Auto Workers. Their leverage came from interrupting production, not from the committee and reporting system Sloan describes. ⁹ The University of Michigan-Flint's oral histories preserve participant accounts of conditions and collective action that an executive-centered institutional history cannot supply. ¹⁰

Dealers and households appear in the memoir mainly through markets; communities, roads, resource use, pollution, and the long dependence of a city on a plant sit even farther outside its measures. A division can meet its return target while shifting costs to workers or the public. Financial visibility made GM more governable from the center, but it did not make every consequence visible. ¹¹

The book was itself a corporate struggle

The apparent voice of a single executive was collaboratively produced. The publication record credits John McDonald and Catharine Stevens as editors. Yale's guide to McDonald's papers describes Stevens as the project's research associate and McDonald, though officially its editor, as responsible for most of the text. ¹² According to McDonald's later account, the book was ready by 1959, but GM tried to prevent publication because it feared the account could be used in an antitrust case. Fortune began serializing it in 1963, and Doubleday issued the first book edition in 1964 after a struggle involving Sloan, McDonald, the corporation, and Time Inc. ¹³ Current catalog records disagree on the page count of the 1990 reissue, so chapter locators below follow the 1964 edition rather than pretending that pagination is stable. ¹⁴

That authorship history changes the evidentiary status of the calm retrospective narrative. My Years with General Motors is neither a neutral transcript nor a solitary recollection. It is a participant model, shaped with professional help and published against institutional resistance after the organizational form had become famous. Peter Drucker's Concept of the Corporation, Chandler's Strategy and Structure, Freeland's archival history, and labor testimony disagree with parts of its field of vision. Together they show why Sloan's central problem endures: distributing operating authority works only when an organization can also decide which consequences require common control and who has standing in that decision. ¹⁵

The documented subject relation is to General Motors under Alfred Sloan, and the transfer from DuPont is described above. The links to delegation, decentralization, and responsibility, organizational intelligence, and benefit for all life, like the comparisons to Drucker and Chandler, are editorial reading paths. They do not assert that Sloan used those later concepts or that the works reach equivalent judgments. ¹⁶