---

My Redistricting Proposals

Friday afternoon, I introduced my redistricting proposals that I crafted in collaboration with Brian Cannon from OneVirginia2021 and leading anti-gerrymandering experts around the country.

The two vital questions they resolve are:

1) who decides who draws the maps

2) how are maps drawn.

1) HB1645 mirrors OneVirginia2021's proposed independent commission in 2019 (SJ274).

My bill creates a tri-partisan independent commission to draw the maps, comprised of three Democrats, three Republicans, and four unaffiliated citizens.  Proposed maps would need support from members of all three groups in order to be sent to the General Assembly.

Their maps would go to the General Assembly for an up or down vote. Only after two such proposals were voted down could the General Assembly amend the maps proposed by this truly nonpartisan, independent citizens' commission. 

I also introduced HJ143.

2) HJ143 is a Constitutional Amendment which, in addition to creating the citizens' commission outlined above, contains criteria to which the Commission must adhere when drawing the legislative and congressional districts and imposes certain requirements on the Commission's activities to ensure accessibility by the public.

The most important part of HJ143 is its criterion of fairness, the mathematical requirement that districts be fairly drawn. This cutting-edge idea, which I developed with the nation's top political-science professors who've studied ways to fight gerrymandering, is a bit wonky, but its basic principle is not difficult to understand.

Whatever maps the commission draws would be run through a hypothetical election based on the partisan leanings of every single precinct in Virginia.

The criterion of fairness necessitates that if Democrats and Republicans have a tie vote statewide, the resulting number of statewide districts would favor neither party.

So if exactly 1,425,252 Virginia voters choose Democrats and 1,425,252 Virginia voters choose Republicans (and, say, 54,225 choose a third party), the model is designed to neutrally create a 50/50 split in the Virginia House, a 20/20 split in the Virginia Senate, and a 6/5 split in the US House of Representatives. If more voters choose Democrats, Democrats should get the majority of seats. And if more voters choose Republicans, Republicans should get the majority of seats.

Now it is just a model. No model is perfect. Unlike my gerrymandering game, votes are not a fixed number. Voters do change their minds. Candidates still matter. And that's a good thing. Indeed, that's what democracy is about. If voters were always completely predictable, we would not have upset victories in elections. Your vote still does count.

So if the vote statewide is tied -- but one party has an inordinate number of good candidates running, while the other party has a number of bad candidates running -- a statewide partisan tie can still under the model give a party with the best candidates a majority of seats. As I said, that's a good thing.

But lines should be drawn so as not favor either party. 

My model would require maps to favor neither political party.

Not intentionally. Not accidentally.

A statistically derived "criterion of fairness" would prohibit unfair maps.

Governor Gerry designed his salamander-shaped map to favor his Democratic-Republican Party.

How does it work? How can we possibly determine how a precinct will vote before it does? And how do we weed out unusual elections where, say, people deviated from their usual party lines?

Virginia has only five statewide elections:  President, US Senator, Governor, Lieutenant Governor, and Attorney General. I purposely chose Virginia's Lieutenant Governor and Attorney General races to make the model on the theory that fewer Virginians know who the LG or AG are than know who their Governor, US Senate, or President is. People may have strong feelings about Trump or Clinton or McAuliffe, but my model presumes that most votes for LG or AG are generic party-line votes.

The model also requires we use the average of six elections -- the last three elections for LG and AG -- to smooth out particular election anomalies. For 2021, we would look at 2009 (Republican landslide), 2013 (close election, with AG race practically tied), and 2017 (Democratic landslide) to make the model. As this is a Constitutional Amendment, it would work 10 years from now too. In 2031, we would look at the six LG and AG elections of 2021, 2025, and 2029 to calculate the criterion of fairness.

How do we determine the results from a tie election when only one of those six elections was actually very close to a tie?  We uniformly shift the votes up or down in each election to create the exact tie for each election in our model. So if one party wins by 53-47, a six percent margin, we move 3% of the statewide winning candidate's vote total to the losing candidate's vote total in every precinct to create a 50-50 statewide tie. (It also works with a small third party. If the victory is 49-45-6, it becomes 47-47.) That way, although each precinct would retain its Democratic or Republican bias in the model, we mathematically force a statewide tie for the model.

Then we have a generic model for how that precinct would vote in a tied election. We apply the new maps to it, and if they meet the criterion of fairness, plus or minus a seat, we know we have prevented gerrymandering.

A bit wonky, I know. But provably fair based on an objective standard.

Want to help me pass these bills? Please contact your other elected state representatives and urge them to support them!

If you're interested in sharing a personal story related to a bill, or want to come to Richmond to testify in committee in support of a bill, please contact my Chief of Staff, Jacob Weinberg by clicking the link and he will work with you.