My Finger is on the Button

Push the Button

Inputs
8,000,000,000
0.050
Std. dev. of final blue share around its expected value.
0.550
Red share = 0.450. Blue voters die if μ falls below 50% on election day.
red bloc: — · blue bloc: —
Per-vote
Red Votevote against blue
Pivotality framework
Expected harm
Expected personal benefit
Net expected value
Share-of-outcome framework
Expected harm
Expected personal benefit
Net expected value
Blue Votevote for blue
Pivotality framework
Expected benefit
Expected personal cost
Net expected value
Share-of-outcome framework
Expected benefit
Expected personal cost
Net expected value
Distribution of final blue vote share

P(blue wins) =

Pivotality:

The Thought Experiment

Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?

The model

The final blue share X is modelled as Normal(μ, σ²). In direct mode you set μ (expected blue share); in detailed mode you set P(blue wins) and the conditional loss share bloss independently. The expected death toll if blue loses is D = bloss · N.

Pivotality framework (marginal causal contribution)

Asks: what is the chance your one vote is the one that tips the outcome? With X ~ Normal(μ, σ²), a single vote shifts μ by 1/N, so pivotality = φ((0.5−μ)/σ) / (σ · N). Expected harm (red) or benefit (blue) = pivotality × D.

Under this view your vote only matters if it changes the outcome. With billions of voters that probability is astronomically small.

Aggregate note. Per-vote pivotality is a derivative; naively multiplying it by the size of a bloc would linearise the response and overshoot the death toll. For a bloc as a whole the natural counterfactual is "everyone in the bloc voted the other way." In both directions expected deaths collapse to ~0 (a unanimous-blue world has no losing scenarios; a unanimous-red world has no blue voters to die). So the bloc-level harm or benefit equals the current expected death toll D · P(lose). The two frameworks therefore agree at the bloc level and only diverge in how they slice that total among individuals.

Share-of-outcome framework (collective responsibility)

Asks: what share of the collective outcome do you bear responsibility for? Total expected deaths in the world = D · P(lose). Red voters share the harm equally across the red bloc; blue voters share the prevention equally across the blue bloc.

Per red voter: D · P(lose) / ((1−μ) · N) = bloss · P(lose) / (1−μ). Per blue voter: bloss · P(lose) / μ.

Personal stake (red vs. blue, by counterfactual)

Each row compares casting that vote against the opposite. If you vote blue and blue loses, you die: expected personal cost = P(lose) × 1 life. If you vote red instead of blue, you avoid that fate: expected personal benefit = +P(lose) × 1 life.

The choice between these frameworks is a contested moral question. Pivotality says “only votes that change outcomes matter.” Share-of-outcome says “everyone bears a share of the collective result.” Neither is provably correct.