We revisit the Easterlin paradox about the flatness of the happiness trend over the long run, in spite of sustained economic development. With a bounded scale that explicitly refers to "the best possible life for you" and "the worst possible life for you", is it even possible to observe a rising trend in self-declared life satisfaction? We consider the possibility of rescaling, i.e. that the interpretation of the scale changes with the context in which respondents are placed. We propose a simple model of rescaling and reconstruct an index of latent happiness on the basis of retrospective reports included in unexploited archival data from the USA. We show that national well-being has substantially increased from the 1950s to the early 2000s, on par with GDP, health, education, and liberal democracy. We validate our new index on several datasets, and find that it captures important changes in personal life circumstances over and above nominal life satisfaction. Our model sheds light on several well-documented happiness puzzles, including why life satisfaction did not drop during the COVID-19 pandemic, why Ukrainians report similar levels of life satisfaction today as before the war, and why people take life-changing decisions -like having kids -that seem to make them less happy.