Primary nights rarely come one at a time. On the biggest ones, a dozen states vote at once and the returns land in a blur. The interesting question isn’t just who won where — it’s why the same national moment reads so differently across five separate electorates.
So we did the obvious thing: synthesize all five at once, and interview them in parallel.
The five
We built a synthetic likely-primary electorate for each of five states voting the same night, raked to each state’s own active-voter file:
| State | Synthetic electorate | Modeled turnout |
|---|---|---|
| California | 40.7M pop | Low-mid |
| Texas | 29.4M pop | Low |
| North Carolina | 10.0M pop | Mid |
| Virginia | 8.2M pop | Mid-high |
| Massachusetts | 6.6M pop | Mid |
Each electorate is its own population — its own demographics, its own priors, its own open-ended “why”. Nothing is shared across state lines except the national backdrop.
What travels, and what doesn’t
Running them side by side, the pattern that jumps out is how little the national story explains on its own.
- The same endorsement lands as an asset in Virginia and a liability in Texas.
- “Electability” is the dominant word in California’s open-ends and barely registers in Massachusetts, where synthetic voters talk about record instead.
- Turnout assumptions swing the North Carolina result more than any candidate attribute — it is the state where our five runs disagree most with each other.
“I’m not voting on vibes this time. Show me what they actually did.” — synthetic voter, North Carolina, age 52
That sentence shows up, almost verbatim, in three of the five states — and never in the other two. Finding where a sentiment travels is the whole point of running them together.
How to read the map
The globe locks onto all five states as it acquires them — each reticle fires as its electorate finishes synthesizing. It’s a literal picture of the method: five independent populations, resolved in parallel, held on one screen so the cross-state comparison is right there.
Every electorate here is synthetic. State populations describe simulated electorates, not polls of real voters, and none of this is a forecast.