In 1821 David Ricardo added a chapter to the third edition of his Principles of Political Economy and Taxation. He called it On Machinery, and near the start he tells the reader he has changed his mind. For years he had taught that a new machine left every class better off, the labourer included. Now he wrote that replacing human labour with machinery is often very bad for the labourers.
The manner of it stays with me. One of the most influential economists in England put the reversal in the same book as the original view, under his own name, while he sat in Parliament. He kept the earlier chapters. Readers could compare.
A habit worth copying
Forecasts about AI and work tend to get published once. When the numbers drift, the authors move on to the next forecast and the old one stays online with nothing added.
His argument holds up better than its reputation. He worried about the years of transition. Capital moves into machines, gross output can fall for a while, and the workers displaced in that window carry the cost. He still favoured machinery, and he warned that a country which discouraged it would watch its capital go abroad.
My own research on this site keeps a corrections log for the same reason. When one of my numbers turns out wrong, the correction sits beside it with a date, as it does in The Measured Enterprise 2026.