Each major economic era mistook a dominant output for the source of value: yield in agriculture, throughput in industry, information processing in the knowledge economy. Each automation wave then commoditized that output without abolishing value creation. If the named resource were the engine, automating it should eliminate value creation. It never does. This paper argues that the underlying engine was elsewhere: in the encounter architectures through which organized systems meet, generate new possibility, and render some of it convertible. This pattern exposes a gap in economic measurement. Productivity tracks conversion efficiency on a fixed possibility set. It does not tell us whether the possibility set itself expanded—whether new reachable configurations appeared. This paper names that missing dimension generativity and distinguishes it from productivity. The distinction helps reframe several persistent puzzles in contemporary economics, including the productivity slowdown, the weak diffusion of frontier gains, and the repeated failure of reskilling programs in encounter-starved regions. These patterns suggest that economic dynamism depends not only on processing known inputs more efficiently, but on sustaining the conditions under which new possibilities emerge. AI sharpens the issue. The same technologies that increase information-processing efficiency can also degrade the encounter conditions that support breakthroughs, spillovers, and cross-domain recombination. Evidence from remote work and proximity research suggests that when encounter architecture degrades, incremental output can persist even as generative capacity weakens. The paper proposes falsification criteria and asks a narrower question: whether the historical pattern of output-source misattribution, together with current evidence on encounter architecture, warrants developing a science of generativity alongside the mature science of productivity.
Paul Campillo (Thu,) studied this question.