We were able to create an algorithm that produces eight voting districts of equal populations. Our first algorithm produced compact districts that were contiguous. These districts looked very natural. Due to the random nature of the algorithm, the districts are completely unbiased. Our second algorithm produced districts that aren’t completely contiguous, but are still more compact than the existing districts in Maryland. They preserve communities of interest based on our value function. Because we used a value function, there is a chance that we inadvertently created biased districts.
With more time, we would like to find data from previous elections to simulate the results using the districts our algorithm created. Simulating elections would give insight to any bias associated with the districts. This bias could be due to the natural population distributions in Maryland, or the algorithm itself.
Finally, we would like to see how our algorithm would hold up creating districts for other states. Other states could be more naturally biased than Maryland. We would have to simulate the elections of many states to determine if the algorithm can be used for more than just Maryland.