In House Arrest: How An Automated Algorithm Constrained Congress for a Century, Data & Society Faculty Fellow Dan Bouk reveals how an automated algorithm implemented in the 1930s has led to an imbalance of power in Congress that lasts to this day.

Bouk’s report provides yet another example of how hidden algorithmic systems used by the state to make fundamental decisions about democratic process and practice that entrench inequities in our society.

In 1929, Congress introduced an automated algorithm (meaning a recipe, a plan, or set of instructions) as a tool to apportion the House of Representatives based on the data from the decennial census. This came after the events of the 1920 census, during which time Congress failed to come to a consensus on apportionment. This, Bouk argues, resulted from a decision to cap the number of representatives in the House at 435.

Prior to the automated algorithm, Congress would review different methods and consider different House sizes to balance power among the states. A crucial variable in this equation was the size of the house: Congress could choose to add or remove seats—and it almost always added them.

While the automated algorithm was intended to serve as a backstop for when Congress could not come to a decision themselves, it ended up, over time, becoming the only method used for apportionment. The number of House seats stopped being a variable.

The result: the automated algorithm made/makes it difficult to see alternative models for the distribution of power in Congress, and it keeps Congress from growing along with the US population. Smaller states have outsized power in the electoral college as a result and the gaps between representatives and their districts have widened.

Bouk argues that there are two paths forward: we can either pass a bill to increase the size of the House, while maintaining the automated algorithm, or, we can put the algorithm and the apportionment process back in Congress’s hands every ten years.