The Root Causes of the Two Dan Sullivans Problem

Yesterday I was on SiriusXM with Dan Abrams, who has a law and politics podcast, talking about the two Dan Sullivans issue in Alaska and related issues. He was more focused than I would have chosen to be on the possibility of criminal prosecution but it was an otherwise useful conversation. It's now available via YouTube either by clicking this link or watching below. 

Meanwhile, I created a little graphic to explain how the problem of same-name candidates emerges, given that it would be simple enough to prevent this sort of thing with a high threshold for ballot access in a two party system. (I "created" the graphic in the sense that I generated all of the steps and the text, but I used Claude to turn it into a graphic.) Here it is:

Why Alaska's system creates an incentive for same-name candidates

A chain of fixes, each solving one problem while opening the door to the next, ending with a tactic like running two candidates named "Dan Sullivan."

Polarized nominees Jungle primary & its risk Alaska's fix The loophole Historical note

flowchart TD
  n1["1. FPTP elections + Duverger's Law drive a two-party system"]
  n2["2. Primary turnout is low and skews toward the most engaged voters"]
  n3["3. Rising polarization pulls each party's primary electorate further left or right"]
  n4["4. Nominees land left of center (Dem) or right of center (GOP)"]
  n5["5. States adopt jungle primaries to pull nominees back toward the center"]
  n6["6. Top-two jungle primaries risk a vote-split lockout, e.g. early in CA's 2026 governor's race"]
  n7["7. Alaska's fix: a top-four primary"]
  n8["8. Top-four creates a general-election splitting risk: incentive to run '1 of ours + 3 of theirs'"]
  n9["9. Alaska's fix: ranked-choice voting"]
  note79["Adopted together in one 2020 ballot measure, not as two sequential fixes"]
  n10["10. Low-information voters often don't rank; they just vote for one candidate"]
  n11["11. The incentive survives: engineer a ballot where those voters waste their vote"]
  n12["12. Tactic: run a same-name candidate, e.g. two 'Dan Sullivan's in the 2026 AK Senate race"]
  n13["13. Partial fix: ballot labels showing party + 'Incumbent' — weakest on the lowest-information voters"]

  n1 --> n2 --> n3 --> n4 --> n5 --> n6 --> n7 --> n8 --> n9 --> n10 --> n11 --> n12 --> n13
  n7 -.- note79
  note79 -.- n9

  classDef phase1 fill:#bfdbfe,stroke:#2563eb,color:#1e3a8a,font-weight:600
  classDef phase2 fill:#ddd6fe,stroke:#7c3aed,color:#4c1d95,font-weight:600
  classDef phase3 fill:#bbf7d0,stroke:#16a34a,color:#14532d,font-weight:600
  classDef phase4 fill:#fed7aa,stroke:#ea580c,color:#7c2d12,font-weight:600
  classDef note fill:#f3f4f6,stroke:#9ca3af,color:#374151,stroke-dasharray: 4 4,font-style:italic

  class n1,n2,n3,n4 phase1
  class n5,n6 phase2
  class n7,n8,n9 phase3
  class n10,n11,n12,n13 phase4
  class note79 note
    

Phase 1: the problem polarization creates

1

Duverger's Law

First-past-the-post elections mechanically favor two dominant parties.

2

Primaries attract the most engaged

Primary turnout is a fraction of general turnout, and skews toward voters who care most.

3

Polarization sharpens the skew

As the parties pull apart, those highly engaged primary voters skew further left (Dem) or right (GOP).

4

Off-center nominees

Each party nominates someone to the left or right of the general-election median voter.

Phase 2: a fix, and a new risk

5

Jungle primaries

States let all voters choose among all candidates regardless of party, aiming for more centrist nominees.

6

Top-two lockout risk

If one party fields more candidates than the other, it can split its own vote and get locked out of the general, a risk that was present early on in California's 2026 's race.

Phase 3: Alaska's combined fix

7

Top-four primary

Alaska advances the top four vote-getters instead of two, reducing the lockout risk.

8

...but now the general can split too

With four candidates in the general, each party has an incentive to engineer a field of "1 of ours, 3 of theirs" so its own vote doesn't split.

9

Ranked-choice voting

RCV is meant to neutralize that splitting risk, since voters can rank rather than pick just one. Important nuance: Alaska adopted top-four and RCV together, in a single 2020 ballot measure; RCV wasn't a later patch for a problem top-four created on its own; it was designed in from the start to handle exactly this.

Phase 4: the loophole that remains

10

Low-information voters don't rank

Many voters mark only a first choice, effectively voting as if it were plurality, not RCV.

11

The incentive survives

Because bullet-voting reintroduces vote-splitting risk, parties still have reason to engineer a ballot that causes the other side's low-info voters to waste their vote.

12

Same-name candidates

One way to do it: recruit a candidate who shares a name with a better-known opponent, as with the two candidates named "Dan Sullivan" in Alaska's 2026 U.S. Senate race.

13

Ballot design, a partial fix

Labeling a candidate's party and "Incumbent" status helps, but least for the very low-information voters the tactic targets, since they're also the ones least likely to know what "incumbent" means or that they're supposed to rank.

Note on timing: steps 7 and 9 are drawn as sequential in this chart because that's the logical relationship: a top-four primary needs something like RCV to manage the general-election splitting risk it introduces. Historically, though, Alaska adopted both pieces at once, via a single 2020 ballot measure (Ballot Measure 2), not as two separate reforms where one was passed to fix a problem the other had already created.

-- Michael C. Dorf