Core Ideas We evaluate an old and widely accepted yield‐based N fertilizer management algorithm. The algorithm, in itself, was a “ballpark” recommendation that served an important public function. The overconfidence in this algorithm may have harmed agriculture in a number of ways. The algorithm’s empirical derivation was seriously flawed. We examine the origins, implications, and consequences of yield‐based N fertilizer management. Yield‐based algorithms have dominated N fertilizer management of corn (Zea mays) in the United States for almost 50 yr, and similar algorithms have been used all over the world to make fertilizer recommendations for other crops. Beginning in the mid‐1990s, empirical research started to show that yield‐based rules‐of‐thumb in general are not a useful guide to fertilizer management. Yet yield‐based methods continue to be widely used, and are part of the principal algorithms of nearly all current “decision tool” software being sold to farmers for N management. We present details of the theoretical and empirical origins of yield‐based management algorithms, which were introduced by George Stanford (1966, 1973) as a way to make N fertilizer management less reliant on data. We show that Stanford’s derivation of his “1.2 Rule” was based on very little data, questionable data omissions, and negligible and faulty statistical analysis. We argue that, nonetheless, researchers, outreach personnel, and private‐sector crop management consultants were obliged to give some kind of N management guidance to farmers. Since data generation is costly, it is understandable that a broad, “ball park” rule‐of‐thumb was developed, loosely based on agronomic principles. We conclude by suggesting that technology changes now allow for exciting new possibilities in data‐intensive fertilizer management research, which may lead to more efficient N management possibilities in the near future.