A QCon London 2026 talk by Yinka Omole argues that engineers who invest in foundational, recurring problems build more durable expertise than those chasing new tools and frameworks. Drawing on historical examples—FORTRAN, CASE tools, AI code generation—the talk notes that despite repeated predictions about the end of programming, the developer population keeps growing. Case studies include PostgreSQL's long-term architectural payoff, WhatsApp's use of Erlang for reliability at scale, Netscape's costly browser rewrite, and Amazon Prime Video's shift from serverless to a simpler ECS architecture that cut costs 90%. The talk introduces Dan McKinley's 'innovation tokens' concept to help teams evaluate new technology adoption, and concludes that core skills like system decomposition and reliability design will remain valuable even as AI coding tools evolve.
Table of contents
A Long History of Predictions About the End of ProgrammingWhy the Profession Keeps GrowingFoundations That CompoundMatching Technology to the ProblemThe Risks of Rewriting SystemsWhen Simpler Architectures WinManaging Innovation CarefullyIdentifying Durable Engineering ProblemsAI Tools and the Future of EngineeringAbout the AuthorSort: