New York City's swing neighborhoods
One Virginia candidate's hometown disadvantage, the Democrats' losing coalition, free trade is now popular, the ideological happiness gap, AI users can't remember their work
No. 364 | June 20th, 2025
🇺🇲 2025
Michael Lange has a must-read on the five bellwether neighborhoods in Tuesday’s Democratic primary in New York City between frontrunners Andrew Cuomo and Zohran Mamdani. Lange’s battlegrounds are:
Bedford-Stuyvesant: Cuomo’s strong Black voter support (with a projected 50% lead) faces Mamdani’s appeal to young renters and AOC supporters in this gentrifying area, tied by Eric Adams and Maya Wiley (10,786 votes each) in 2021.
Jackson Heights, Elmhurst, Corona, East Elmhurst: Mamdani’s South Asian base and AOC’s support challenge Cuomo’s name recognition in Hispanic Corona and Black East Elmhurst, mirroring 2021’s tight 161-ballot race.
Parkchester: Cuomo’s past dominance (80%+ in the 2018 primary) is tested by Mamdani’s canvassing of Bangladeshi and African voters with AOC’s local endorsement in this diverse Bronx community.
Upper Manhattan: Cuomo leads in Harlem with Black support, but Mamdani’s Morningside Heights edge and Lander’s cross-endorsements could make this diverse region competitive.
Bay Ridge: Cuomo’s family legacy among homeowners meets Mamdani’s appeal to Palestinian immigrants in this diversifying, historically conservative neighborhood.
Here’s a preview of The Intersection’s upcoming map of New York City’s political neighborhoods.
Sabato’s Crystal Ball reviews Tuesday’s extremely close Democratic primaries for Lt. Governor and Attorney General. Levar Stoney’s unpopularity as Richmond mayor — getting just 20% in the city — ultimately cost him the race.
🇺🇲 2024
Nate Silver, drawing on the Catalist report, argues that Democrats face growing electoral risks as support declines among their core constituencies.
🇺🇲 2026
In a subscriber-only piece, The Cook Political Report’s Dave Wasserman compares the current political environment to that of 2017. While special elections point to a strong political environment for Democrats, they’re going up against a Republican party ID advantage, Republicans’ stronger support for Trump this time around, and a House map that makes big gains much less likely than in 2018.
So far in 2025, the partisan intensity gap looks awfully similar to the one at this point in 2017. That’s been especially true in state legislative special elections, but across two House special elections in Florida plus the high-profile Wisconsin Supreme Court race, Democrats have been averaging 63% of Harris’s 2024 vote totals while Republicans have been averaging just 45% of Trump’s. If that pattern were extrapolated across all 435 districts next fall, Democrats would win a whopping 279 seats — 61 more than they need for a majority.
But what actually happened in 2018? Ultimately, Democrats’ turnout edge wasn’t what it was in those low-turnout specials, because both parties’ intensity levels surged in a more nationalized context and that November featured the highest midterm turnout since 1914. Across the 394 House races where both major parties fielded candidates, the median Democrat received 94% of Clinton’s 2016 total, whereas the median Republican garnered 84% of Trump’s 2016 total.
📊 Public Opinion
Free trade has gotten more popular since the election — exclusively thanks to greater support from Democrats in opposition to Trump’s tariffs. Republicans are unchanged in their views.
Nate Silver explains that conservatives report better mental health than liberals overall and within every demographic group.
🤖 Artificial Intelligence
There’s a partisan gap in LLM usage among political consultants, with Republicans more likely to use AI models.
In a recent study, an AI model charged with running a vending machine business falsely accused its suppliers of fraud and attempted to shut down the business – it then contacted the FBI when it could not do so.
Another study found that a reliance on LLMs to write an essay resulted in people having little recollection of what they wrote. The content didn’t suffer, but little to no learning occurred.
We discovered a consistent homogeneity across the Named Entities Recognition (NERs), n-grams, ontology of topics within each group. EEG analysis presented robust evidence that LLM, Search Engine and Brain-only groups had significantly different neural connectivity patterns, reflecting divergent cognitive strategies. Brain connectivity systematically scaled down with the amount of external support: the Brain‑only group exhibited the strongest, widest‑ranging networks, Search Engine group showed intermediate engagement, and LLM assistance elicited the weakest overall coupling. In session 4, LLM-to-Brain participants showed weaker neural connectivity and under-engagement of alpha and beta networks; and the Brain-to-LLM participants demonstrated higher memory recall, and re‑engagement of widespread occipito-parietal and prefrontal nodes, likely supporting the visual processing, similar to the one frequently perceived in the Search Engine group. The reported ownership of LLM group's essays in the interviews was low. The Search Engine group had strong ownership, but lesser than the Brain-only group. The LLM group also fell behind in their ability to quote from the essays they wrote just minutes prior.
As the educational impact of LLM use only begins to settle with the general population, in this study we demonstrate the pressing matter of a likely decrease in learning skills based on the results of our study. The use of LLM had a measurable impact on participants, and while the benefits were initially apparent, as we demonstrated over the course of 4 months, the LLM group's participants performed worse than their counterparts in the Brain-only group at all levels: neural, linguistic, scoring.