Optimal partisan gerrymandering

We look ae the problem of devising an optimal gerrymander from the standpoint of the political party in control of the redistricting process in a singlemember district system involving two-party competition. In an electoral universe with uncertainty, we show that the optimal partisan gerrymander is different from the classic recipe for partisan gerrymandering if parties are concerned with the long-run risk of electoral defeat caused by shifting electoral tides. We look at two different plausible objective functions: (1) maximizing expected seat share: and (2) maximizing the probability of a (working) legislative majority. In general. the optimal districting schemes which will be generated under these two objective functions will be different, although both will make similar use of the two basic gerrymandering techniques, the concentration gerrymander and the dispersal gerrymander. Also, because of the need to minimize risk, both will resemble a bipartisan gerrymander considerably more than has previously been suggested in the literature. We also lcok at difficulties in achieving either objective (1) or objective (2) because of conflicts between majority party legislators’ own selfinterest and the districting that will maximize party advantage, and briefly review evidence that the 1982 California congressional plan was a risk-minimizing partisan gerrymander.