“Gerrymandering is generally defined as the intentional alteration of established political boundaries or the creation of artificial communities by the grouping of political units to form temporary election districts for the purpose of effecting an election outcome.  [1]” There are two methods to Gerrymander.  “Packing” is the process of grouping all of the minority voters as possible into concentrated districts to dilute their votes in other districts.  “Cracking” is the process of splitting the minorities up apart into as many districts as possible to drain their voting power[2].  Below is a graphic of how populations can be manipulated to achieve desired results.



Maryland has a history of having voting districts that are perceived to be gerrymandered.  In 1990, the new census data showed that the population in Congressmen Steny Hoyer’s Fifth District had a large majority of African-American voters.  Hoyer was a very important Congressmen and the Democratic Party was nervous that he was going to lose his seat.  There is no hard evidence, but it is perceived that Hoyer met with other incumbents to resolve this potential threat.

In 1991, House Bill 10 was adopted to create new voting districts.  A district was created with African-American majority so they were free to choose a representative of their choice.  Congressmen Hoyer was given a white district so he could retain his position.  The political boundaries of Maryland in 1991 were not representative of natural political or geographical boundaries.

1991 Maryland Voting Districts


Image from:


Maryland’s districts have remained this way ever since.  In 2011, Maryland created new voting districts after the 2010 census came out (see below).  Steve Shapiro, a current law student filed a lawsuit against the voting districts of Maryland for violating the voter’s first amendment.  Shapiro alleges that the districts of Maryland were gerrymandered by the Democratic Party [4].  The Supreme Court ruled that a three-judge district court shall hear Shapiro’s case.  It is clear that there is a need for a fair way to create voting districts in the state of Maryland.


Image from:

We have encountered two algorithms created to produce unbiased voting districts.  The first is the Shortest-Splitline method.  The Splitline method recursively divides the state into two boundaries using an optimal line.  The optimal line is the shortest geographical line that divides one area into two equally-populated districts.  The algorithm is called recursively until N districts are created.  The algorithm can add constraints so it doesn’t divide political or geographical boundaries such as a neighborhood, census tract, country, etc.

Ivan Ryan’s Shortest-Splitline (Maryland 2009)


The second algorithm is the BDistricting tool created by Brian Olson [3].  He optimizes the compactness of the voting districts.  This means that he tries to minimize the ratio of the perimeter to the area.  These districts are extremely more compact than the voting districts in actuality.  They look more natural and are unbiased as well.


To test the fairness of these two algorithms, Chris Fecor, a programmer in San Francisco, simulated the outcome of the 2008 senate votes using each of the algorithms discussed above.  In 2008, North Carolina voted for 8 Democrats and 5 Republicans.  Popular vote was 60% for Democrat and 40% Republican.  Had the districts been created by the Splitline method, 10 Democrats would have been voted into office.  The BDistricting tool would have resulted in 11 Democratic seats.  By randomly grouping contiguous voting blocks, Fecor created Districts that elected 9 Democrats.  Even though these algorithms were unbiased, they still produced results that didn’t reflect views of the popular vote.  Fecor claims to have obtained similar results from the other states he has tested.  He came to the following conclusion: “Unbiased redistricting isn’t necessarily fair districting [3].”