Think for yourself. Subscribe to The Free Press today: https://thefp.pub/3DmLpLi It hasn’t been a good decade for polling. In 2016, the national polls were humiliated by confident predictions of a Hillary Clinton win. In 2020, polls didn’t foresee how close the swing state margins would be. And in 2024, pollsters still underestimated Donald Trump’s support by nearly three percentage points. But as most media organizations continue to struggle to accurately cover the American majority, polling might remain the best way to understand the motivations of our electorate, and to see what is coming next for the country. I wanted to understand why polling has gotten so much wrong, what it can reliably tell us about America, and what public opinion about our politics looks like today. So I invited Galen Druke to sit down with me on Conversations with Coleman. Galen is the founder and host of the GD Politics podcast. He previously hosted and produced the FiveThirtyEight Politics podcast, where he covered American elections for nearly a decade. Galen and I talked about the science of political polling, and about how polls are conducted, when they signal real change, and when they simply reflect media noise. We talk about what pollsters got wrong about Donald Trump time and time again: that, perhaps counterintuitively, Trump was seen by voters as a more moderate candidate than any other in his party, and even more moderate than many major Democrats. We spent much of our conversation discussing the Democratic Party’s myriad problems and crisis of leadership. In a recent op-ed in The New York Times, Galen argues that Democrats need their own Trump. We discuss what that really means, and it’s probably not what you think. Galen and I also discuss how Democrats break back into the national conversation, by framing messages that resonate with working‑class voters, swing communities, and those who feel politically adrift. I learned a lot about American opinion—and just who should run in 2028—from Galen’s data-driven background and considered analysis.