A convention that runs like a ranked-choice primary
The party's Call to Convention specifies an exhaustive ballot:
the first round determines the top five candidates, then the last-place finisher is
dropped each round until someone holds a majority of the 601 delegates. Since real
people can change their mind at any moment, we can't say with certainty what will happen
at the convention, but using ranked choice results from the June primary, we can
make a model to simulate the most likely outcomes.
Since delegates are chosen from the registered Democrats in each county,
let's assume that every delegate votes like a random
primary voter from their own county. Call this the "representative"
scenario. To simulate this, we randomly sample ballots from the actual, published
Cast Vote Record (CVR), which includes the exact ranking for every voter for every
ballot cast.
First choice only gets you so far
As we saw in the gubernatorial primary, first choices only tell a part of the story. Nirav Shah
argued
that as the primary's first-choice leader he has the strongest support to go against Susan Collins in November.
"What the data show is that we had broad and intense support... we won folks in every
single demographic. And again by win, I mean secured the highest number of first-choice votes." Just like in governor's race,
in this
representative room he does lead the first round — 238 votes to Bellows's 198 and Jackson's 166.
But an exhaustive ballot doesn't crown the plurality
leader. After Jackson is dropped, his support breaks roughly two-to-one for Bellows. (Indeed, the candidates co-endorsed each other in the leadup to the election.)
In a majority of random samples (70 out of 100),Bellows wins on the final ballot.
A slate changes the dynamics
Delegates aren't randomly sampled from the party at large. Each county held a
delegate nominating meeting, open to any registered
Democrat. This provided a mechanism for campaigns to lobby attendees
to select delegates that are pledged (though not legally bound) to
vote for them. Even a modest slate of delegates can easily swing the
margin of error between competitive candidates.
A small slate knocks out the consensus pick
Let's suppose that the Jackson campaign managed to convince a modest slate — 45 delegates,
under 8% of the floor, nowhere near a majority. Watch the first round.
Jackson climbs to 199 and Bellows slips to 183. Bellows
is now last, so she is the one eliminated — the very candidate
who beat Shah head to head a moment ago.
Her support splits almost evenly between the survivors, and Shah's head
start carries the final 298 to 270 — he wins
72 conventions in 100.
This is the classic center squeeze: the broadest
candidate is dropped early precisely because they are nobody's passionate
first choice. A slate a tenth the size of the one Jackson actually built
could have handed the nomination to Shah.
But that isn't what happened
Over July 18–19, Troy Jackson's organization — the Platner network and Our Revolution —
out-organized the field,
running published slates that swept the county delegate meetings, taking
456 of the 500 county seats
(77 of 84 in York alone).
Seat those 456 and the exhaustion mechanism never engages: Jackson
clears a majority on the first ballot, no eliminations
necessary. The candidate who finished last of the three measured in the representative
room now wins in a landslide.