The cornerstone of New Classical economics is the production function. The function was largely an ex-post explanation of a phenomenon of labor and capital quantity indices over short periods, commonly referred to as the Cobb-Douglas. Despite the interventions by Joan Robinson, Pierangelo Garegnani, Luigi Pasinetti, and others, the New Classical economists defended the functional form of the Cobb-Douglas on the basis of its empirical consistency. This evidence, such as the 1957 paper by Robert Solow in the Review of Economics Statistics, was held up as proof positive that the Cobb-Douglas form was the iron law of production.
Then Anwar Shaikh wrote his Humbug paper, and all that changed. Okay, well actually nothing changed. Solow was then editor of the Review of Economic Statistics in which his 1957 article appeared. In order to respond to Solow's seminal work in the same venue in which it appear, Shaikh agreed to a railroading. Shaikh had to agreed to allow Solow to reply to the article and not write a followup rejoinder. As with the rest of the Capital Controversy, the Humbug paper was ignored into oblivion.
What the Humbug paper showed was quite damning for the burgeoning New Classical school of economics. It showed that the Cobb-Douglas form equation did not arise as the result of some iron law of how the factors of production convert raw materials into output. Instead, the functional form was the necessary algebraic result of accounting identities themselves. The Cobb-Douglas function, rather than describing production, actually described income distribution.
To drive this point home, Shaikh created per capita income and capital data that spelled out the word "HUMBUG" on a coordinate plane. Using this nonsensical - and frankly impossible - data, Anwar showed that the same empirical results could be derived from ANY data, so long as it conformed to a proper aggregate accounting framework. Thus, Shaikh showed that the very notion that marginal productivity determined the composition of factor inputs was weak at best. So long as all income distribution is accounted for, any index numbers for labor and capital out of wages and profit will yield results in the form of a Cobb-Douglas function. The residual, rather than being total factor productivity as interpreted by New Classical economists and their descendents, was actually the geometric mean of the wage and profit rate each weighted by the wage and profit shares, respectively.