We make the case for a shift in what students learn in a first economics course, taking as our exemplar Paul Samuelson's paradigm-setting 1948 text. In the shadow of the Great Depression, Samuelson made Keynesian economics an essential component of what every economics student should know. By contrast, leading textbooks today were written in the glow of the Great Moderation and the tamed cyclical fluctuations in the two decades prior to 2007. Here, using topic modeling, we document Samuelson's novelty and the evolution of the content of introductory textbooks since. And we advance three propositions. First, as was the case in the aftermath of the Great Depression, new problems now challenge the content of our introductory courses; these include mounting inequalities, climate change, concerns about the future of work, and financial instability. Second, the tools required to address these problems, including strategic interaction, limited information, principal-agent models, new behavioral foundations, and dynamic processes including instability and path-dependence, are available (indeed widely taught in PhD programs). And third, as we will illustrate by reference to a new open access introductory text, a course integrating these tools into a new benchmark model can be accessible, engaging, coherent and, as a result, successfully taught to first year students. Deployed to address the new problems, following Samuelson's example, the new benchmark provides the basis for integrating not only micro-and macroeconomics but also the analysis of both market failures and the limits of government interventions. JEL codes: A10, A22, B20, C11.Acknowledgments: We are very grateful to the other members of the CORE team with whom we have worked over the past seven years, participants in CORE workshops in Auckland, Bangalore, Bristol, Cape the teachers and students who have provided feedback, all of whose contributions have shaped the ideas presented here. Sahana Subramanyam has provided outstanding research assistance on the topic modeling part of the paper, and we have benefited greatly from the experience of our collaborators, Simon DeDeo and Suresh Naidu, with these methods. Roger Backhouse and David Colander provided extensive and astute criticism of an earlier draft, leading to substantial improvements. We have also benefited from the comments of Yann Algan,