Plain English with Derek Thompson
Why the Midterms Suddenly Look So Bad for Republicans
Nate Silver tells Derek Thompson that the Senate, once a lock for Republicans, has flipped to favoring Democrats as Trump's second-term agenda collides with rising gas and diesel prices, an unpopular war in Iran, and the historical pattern of midterm backlash against the party in power.
Key takeaways
- Prediction markets moved fast and hard: Kalshi gave Republicans an 80% chance to hold the Senate through most of 2025, but that flipped to over 60% odds for Democrats by late September 2026. Silver's own model, Flipper, shows the generic congressional ballot moving from a six-point Democratic edge to eight-to-nine points among likely voters, with some polls showing D+11 to D+13.
- Rising energy costs are driving the shift more than any single scandal. Gas has gone from $4 to $4.50 a gallon, but diesel (used heavily in trucking, agriculture, and home heating oil) is around $6.80. Silver says this explains why "tornado alley" farm states and cold states like Maine and Alaska are hurting most, since only 2% of consumer cars run diesel but entire regional economies depend on it.
- Trump's own missteps, not just structural midterm backlash, accelerated the collapse. He ran on no new Middle East wars and lower gas prices, then launched a war with Iran (which Silver calls "one of the biggest blunders in the history of American politics") and let gas prices climb—both directly undercutting his 2024 promises.
- Democrats' messy, decentralized identity is working as an advantage this cycle. Candidates like Abdul El-Sayed in Michigan, Mary Peltola in Alaska, and James Talarico in Texas were shaped by local primaries to fit their states, while Trump's demand for loyalty pushed Republicans toward weaker, more sycophantic nominees like Ken Paxton in Texas.
- Silver's model projects Democrats winning the House popular vote by roughly eight points, a shift in line with historical second-midterm patterns like 2006 (Bush) and 2018 (Trump's first term), not an outlier prediction; he argues rival forecasters like Cook Political Report lag because they wrongly start from an assumption of a neutral environment instead of pricing in automatic backlash against the incumbent party.
- AI politics remain unsettled and don't clearly favor either party despite Trump's enthusiastic "superintelligence" posts; voters across the ideological spectrum oppose data centers for different reasons (job loss, pollution, distrust of tech billionaires), and Silver sees Trump's full-throated AI boosterism as a rare miscalculation of populist instincts.
Notable moments
- "Calci now says Democrats have over a 60% chance to win back the Senate"; Thompson opens by framing just how fast the odds moved.
- Silver on the political identity of Trump's party: "if you're not loyal to Trump, you're cast out of the party... that creates sycophantic candidates who are maybe not ideal for their state."
- On diesel-driven pain in red states: "the tornado alley states that have agriculture or oil and a lot of trucking... are being hit hardest by the combined increase in energy prices."
- Silver's blunt assessment of the Iran war's political cost: "fuck around and find out has been the motto of recent presidencies."
- On why Republicans haven't broken from Trump yet, Silver notes J.D. Vance and Mike Rogers quietly hedging on the Iran war, calling it the start of a possible "preference cascade."
- Thompson's "jeans" metaphor for the two parties (always too tight or too loose, prompting constant electoral course correction) is used to explain the midterm backlash cycle.
- Silver on 2028 Republican alternatives: he doubts prediction markets giving J.D. Vance 50% odds, comparing Vance's position to Dick Cheney trying to succeed an unpopular Bush; "it would be insane."
- On the fused Democratic message, Silver credits John Ossoff's "affordability and corruption" pitch as unusually effective because he can satisfy both centrist and progressive wings, unlike Kamala Harris who satisfied neither.
Time saved: 1 hour, 1 minute.