# The Efficiency Gap The efficiency gap is a single number intended to summarize how much partisan advantage a district map confers. Nicholas Stephanopoulos and Eric McGhee proposed it in a 2015 *University of Chicago Law Review* article with an explicit legal purpose: to give courts the judicially manageable standard that partisan-gerrymandering doctrine had always lacked. ## The construction Start with [[Wasted Votes]]. A vote is wasted if it did not contribute to winning a seat — every vote for a losing candidate, plus every vote for a winning candidate beyond the threshold needed to win. The efficiency gap is the difference between the two parties' wasted votes, divided by total votes cast: > **EG = (wasted votes for Party A − wasted votes for Party B) / total votes** A map where both parties waste votes at equal rates has an efficiency gap of zero. A large positive or negative value means one party's voters are being systematically wasted at a higher rate — the arithmetic signature of [[Packing and Cracking]]. Stephanopoulos and McGhee proposed thresholds for presumptive unconstitutionality: two seats for congressional plans, or 7% for state legislative plans, sustained across a redistricting cycle. ## Why it was attractive **It is one number.** Courts handle single thresholds well. The whole history of gerrymandering doctrine is judges saying they can see the problem but cannot articulate a line. **It requires no counterfactual map.** Unlike ensemble methods, it is computable directly from election returns. **It has an intuitive story.** "One side's votes are being thrown away at twice the rate of the other's" is a sentence a jury understands. ## Why it did not survive contact **It bakes in a proportionality assumption.** The efficiency gap implies a specific seats-votes relationship — roughly, a party winning 50% + *n* of the vote should win 50% + 2*n* of the seats. That relationship is a normative choice, not a neutral measurement, and defenders of maps correctly pointed out that no one had voted to adopt it. **Political geography confounds it.** Democratic voters cluster densely in cities. That clustering wastes Democratic votes through the ordinary operation of single-member districts, with no mapmaker intent required. A nonzero efficiency gap therefore does not by itself distinguish deliberate manipulation from residential sorting — the central technical objection, and the one that motivated the shift to [[Markov Chain Ensemble Methods]], which handle geography by construction. **It is unstable in landslides and near-ties.** The measure behaves badly when one party wins overwhelmingly, and small vote swings can move it substantially. **The Court never ruled on it.** *Gill v. Whitford* (2018) arrived with the efficiency gap as its centerpiece and was resolved on standing without reaching the merits. [[Rucho v. Common Cause]] (2019) then held partisan-gerrymandering claims nonjusticiable in federal court altogether — mooting the metric's federal constitutional application regardless of its technical merits. ## What it is still good for The efficiency gap remains useful as a **descriptive and journalistic instrument**. It is easy to compute, easy to explain, and tracks over time — which makes it a reasonable first-pass indicator that a map deserves scrutiny, even though it cannot carry the evidentiary weight its designers hoped for. Its failure is instructive: it is a clean case of a metric that was mathematically well-defined, politically legible, and still insufficient, because the quantity it measured embedded a contested normative premise the measurement itself concealed. ## Related [[Wasted Votes]] · [[Packing and Cracking]] · [[Markov Chain Ensemble Methods]] · [[The Mathematics of Redistricting]] · [[District Compactness Measures]] · [[Gerrymandering]] · [[Mathematics and Power Hub]]