AB-EXIT Civic Simulation
Can a third-place candidate win only with AB-EXIT?
Run Alaska's 2022 governor race twice: first without AB-EXIT, then with the payment already selected in Referendum Lab. In the central scenario, Bill Walker loses without AB-EXIT and wins only after AB-EXIT is introduced.
Everything else stays equal: the same candidates, electorate size, pre-existing political preferences, campaign spending, and official election data. The only institutional change is the AB-EXIT option; every modeled behavioral response to that option remains an explicit, inspectable hypothesis.
Projected first-round winner
Bill Walker wins
Loading the versioned Alaska profile…
Official result vs AB-EXIT counterfactual
First-choice votes; projected bars recalculate with the payment.
Where Walker's votes come from
Projected decomposition, not observed voter declarations.
What changed—and what did not
Same people. Same candidates. Different participation mechanism.
- Weakly attached voters can choose a payment instead of a low-conviction ballot.
- Every Exit increases the relative weight of every remaining vote.
- Capable non-voters regain a reason to participate.
- A viable independent attracts pragmatic support previously trapped by party strategy.
Payment robustness
Testing the full displayed range…
Audit the assumptions and falsification boundary
Observed
- Candidate first-choice totals
- 266,943 ballots cast
- 601,161 registered voters
- Alaska's majority and ranked-choice rule
Hypotheses
- Committed versus weak-voter segmentation
- Money-versus-influence response
- Activation of historical non-voters
- Independent affinity and RCV transfers
How this result fails
Calculating the nearest joint assumption stress…