OK what could you do about it? Well one thing you can do is make it a rule. Ohio was perhaps the first state to do that, and it only happened recently. In 2018, Ohio voters passed one [state] Constitutional amendment to create a commission – a not very independent commission, it turns out – and written in these rules was the aim that the [partisan] Seat percentage should reflect the percentage of [statewide] Voices. And as far as I know, this is the first example of setting proportionality as a goal.
For a mathematician like me, this really makes sense: state your goals and we can try to achieve them. But when your goals are really vague, it is very difficult to talk about why one might be better or fairer than the other.
When did you discover that your math background had an application to gerrymandering?
I started working on it in 2016. My background in math is in geometry. And I thought what if we thought about what it would mean to be “fair” on the district drawings side?
I started with an intuition that the story is in the shapes and that if we can only find the right shape metric, we will do it [solve it]. I searched for the authoritative literature on all of these “compactness metrics” that would tell the right story, and to my surprise there really was classical math and old, pre-classical math, but there didn’t seem to be any post-1900 math in the mix.
The geometry of discrete spaces has really exploded in richness and depth over the past 100 years, but I haven’t seen many of these ideas in the mix. And it struck me – as I am sure many others have recognized earlier – that districting is really a discrete problem: there are a limited number of people, and we have these geographic parts that tell us where they are. Basically, I got the idea, “Oh, I bet there is something that could be made useful here.” And it blossomed a full-time research program.
One thing we’re going to be dealing with this cycle of restructuring that we haven’t seen in the past is this new approach to the census that actually changes the underlying data. Can you walk me through this and how will that affect redistribution?
Yes. The Census Bureau is committed to doing something modern – which always makes people nervous. In this case, they have “microdata” – the answers to all forms of census in a huge table that includes all of the answers for each individual in the list. The office does not publish all of this information publicly. Instead, it is summarized: census pads or block groups may contain hundreds or a few thousand people, and you get Aggregate Statistics instead of individual responses – so there will be a small part of a map and you will know how many people live there and what their overall responses were.
Now the threat is this: when you have enough of these aggregated statistics, you can toss them into a computer and actually reproduce them Entrance Table that produced the aggregated statistics. The first risk is that you can recover the data at the person level. And risk number two, which is really interesting, is that when you pair it with readily available commercial data – like from Facebook – you can find out what their names, addresses, and phone numbers are for some of these people.
All of that computing power that comes into play in elections is generally pretty healthy and pro-democratic – people come up with ideas for better systems and results. Bad actors are also empowered by computers. That’s the risk – and it’s interesting. Under Title 13 of the US Code, the Census Bureau has an obligation to protect privacy. Does this include protecting people from these “reconstruction and re-identification” attacks that may require you to use third-party data to do so? The Bureau decided that the answer is yes.
So they took up this idea called “Differential Privacy” that was developed by Cynthia Dwork and her colleagues in computer science. And the idea is: what if you could, in a really controlled way, add random noise to all of your numbers to get them a little off here and there, but when you add it up all these differences would break out and you get numbers that are very accurate at high levels, even if they are very loud at low levels?
It’s a great idea. And the beauty of it is that you can do it in a really controlled way. The Census Bureau announced that they would, and chaos ensued. You have already been sued in an Alabama-led lawsuit.