The Great American Gerrymander Machine

Let's automate the drawing of legislative districts and reclaim our democracy

Share with others:

Print Email Read Later

You may be surprised by the state of our democracy when one party's congressional candidates win 1.1 million more votes than the other but end up being greatly outnumbered in the House of Representatives.

You may be alarmed when one party's congressional candidates win 83,000 more votes in the state of Pennsylvania but end up with five House seats to the other party's 13.

You may be astonished that schemes are afoot in Pennsylvania and other states to change the Electoral College such that, if they had been in effect last year, the presidential candidate who won by nearly 5 million votes would have been declared the loser.

But I have expected for more than two decades that we would experience the out-of-control gerrymandering of legislative districts that is now causing such assaults on our political system.

I first realized that our country was headed toward unbridled gerrymandering in the late 1980s when I spent several years of my career as a history professor developing (with much assistance from colleagues who had the necessary technical skills) a computer-based historical atlas called The Great American History Machine.

This software application would superimpose county-level census data since 1790 on a map of the United States, as well as similar data of federal elections. So, for example, a student in an American history course could instantly produce a map showing the percentage of Democratic presidential votes in every county in 1880 and compare it with a parallel map of the percentage of people enumerated in the 1880 census who were born in, say, Ireland.

I displayed this software at a number of conferences, and people frequently suggested that contemporary politicians would be delighted to have it to draw legislative districts in favor of their respective parties.

For a number of reasons The Great Machine itself would not have been the right application for that undertaking. But I gradually began to realize that the developments in computing that were enabling me to produce my harmless educational application eventually would enable others to ravage democracy through gerrymandering.

The 2010 census turned out to be the moment when the devastation began.

It just so happened that the GOP had the upper hand after the 2010 elections. Republicans controlled redistricting in states with 40 percent of congressional seats vs. the Democrats' 10 percent. Courts, divided governments or commissions drew up the rest. And the Republicans took full advantage of state-of-the-art gerrymandering technology -- as the Democrats would have, given the chance.

The usual antidote to gerrymandering is to have nonpartisan commissions draw up congressional boundaries. Such commissions are used in some states and in other democratic countries, but I would suggest an alternative.

The problem at hand is a result of the computer revolution, and that revolution could also provide the solution.

Given the development of geographic information systems over the past generation, it is quite possible to develop a process by which a computer could divide a state into the required number of compact, contiguous districts. An algorithm could be ordered to produce, say, 10 different maps, with each one being constructed by having the algorithm start at one of 10 equidistant locations around the perimeter of the state.

The legislature then would the choose one of the 10 maps, each of which would be fair to all of the identities in the state (age, income, race, educational level, etc.), One of the wonderful features of a computer is that it can be kept ignorant of information that is supposed to be ignored -- such as political affiliation. In this respect, a computer is better than even the most impartial commission.

What we need is a movement to enact such a process, and Pittsburgh is the ideal place in which to organize it -- for two reasons.

First, Pennsylvania is one of the states most affected by gerrymandering.

Second, Pittsburgh now serves as an eastern partner of Silicon Valley.

The movement should begin by identifying and, if necessary, producing programs to execute its proposed redistricting process. Then it should establish a well-advertised website that shows the boundaries of current congressional districts and enables users to see how different they would look if the program had been applied to the 2010 census returns.

Finally, the movement should be ready when the 2020 census is released to demonstrate how its process would work and to see that the media compare legislature redistricting proposals with those produced by the anti-gerrymandering program.

At that point, with the public clearly understanding the issue, we could look for redress from our loss of democracy, whether it be from Congress, state legislatures, the courts or the last resort: a constitutional convention.


David W. Miller is professor emeritus of history at Carnegie Mellon University ( His Great American History Machine will soon be revived and made available on the Internet thanks to CMU's Olive program, the first archiving system for the preservation for executable content.


Create a free PG account.
Already have an account?