DistrictSim™ is an iterative algorithm using parallel processing to search for new plans. Seeded by an initial plan, a first stage seeks to "shrink" districts so they break fewer county lines, in many cases resulting in fewer county splits than actual enacted plans.

Further steps reduce the partisanship bias away from the gerrymandering present in enacted plans and compactifies the most difficult plans - even State legislative districts with hundreds of seats that can take a week or more to draw by hand.

Here's our map for the 203-seat Pennsylvania General Assembly:

This proposal reduces the partisan bias as seen in the high symmetry of the seats/ votes curve with -0.35% democratic vote bias to capture 50% of seats, while achieving 186 county splits. The enacted General Assembly map has 221 county splits.

Our proposal for the 50-seat Pennsylvania State Senate is:

This proposal reduces the partisan bias to near zero, as seen in the high symmetry of the seats/ votes curve with 0.35% democratic vote bias to capture 50% of seats, while achieving 46 county splits. The enacted State Senate map has 51 county splits.

These proposals reduce county splits substantially as compared to the enacted plans, keeping communities of interest together, while eliminating partisanship bias. This is an extremely difficult if not impossible task to perform by hand, requiring at the very least weeks of painstaking work. We are not suggesting these machine-generated maps replace work performed by districting committees and public input but they can provide a starting point that humans can improve upon.