Fragments: June 2

This article from Martin Fowler's 'Fragments' series discusses three key tech topics: Greg Wilson's critique of common but flawed metrics for measuring AI tool productivity (e.g., lines of code, tickets closed, developer surveys); Benedict Evans' historical perspective showing automation of accounting didn't reduce jobs but changed them, highlighting the difficulty of forecasting AI's impact on work; and Stephen O'Grady's analysis of closed vs open AI model benchmarks, noting closed models lead innovation but open models catch up faster over time. The piece offers insights into measuring developer productivity, the Jevons paradox in tech, and the evolving landscape of AI capabilities.