[Image via vimeo]
I’ve written in the past about what I call “52 pickup” – the challenges that arise for election officials in the aftermath of legislative redistricting, but a new analysis using Virginia data suggests that the problem may be thornier and more widespread than imagined. The Washington Post has more:
When the legislature redraws its districts, the boundaries are built, block by block, with the help of modern computer mapping. The map is translated into a legal text description of the boundaries, which goes to legislators for approval. But the local registrars, who match voters to those districts, must first manually assign the districts to lists of local street blocks, sometimes specifying down to the level of one side of a city block, or even a single apartment building. For instance, one list item could represent the odd side of 4th Street, address numbers 601 to 649.
That process is also used to update for new buildings, new streets or other changes that affect districts. “It can be very tedious,” said Linda Lindberg, elections director and registrar in Arlington County. [York County’s Walter] Latham said, “It’s about as manual a process as you can use and still be using a computer.”
Mistakes can also arise because of simple error or odd street geography:
Many of the registration errors found by the state in the 28th House District originated when districts were redrawn following the last Census in 2010, and so had been in place undetected for at least three legislative elections. A lawyer for Democrats protesting the 28th House District election said complaints about registration errors there started two years ago. Still, in the November election, some voters got the wrong ballot because no one had noticed that in the street block list “Charles” and “Charlotte” streets had been mixed up during redistricting.
Cul-de-sacs, like the one found … in District 27, can also add confusion since they don’t always have “sides” the way other streets do.
New home construction can also complicate the issue:
Another occasion for errors is new construction, according to Lindberg, the Arlington registrar. Imagine a street split by a district boundary, with homes on the left side of the boundary, but an empty field on the right. Registrars could properly consider this block to be in the left district, since all the homes on the block are in that district. But that means when homes are built on the right side of the boundary, and should be in the right district, they are actually included in the left district.
Finally, the physical location of a home on a lot, as opposed to its digital “location” on a map – especially in rural areas – can also affect its election assignment:
Michael McDonald and Brian Amos of the University of Florida also found the underlying geography of a district, especially in rural areas, can lead to errors. If a house is set far back from a road but the state geolocates it to be along the street, it could be coded into the wrong district.
These problems can happen anywhere, and while they don’t affect huge numbers of voters, they can (as has occurred in Virginia) become a point of contention in close races:
Work by McDonald and Amos shows this is not a problem confined to Virginia. “These errors are happening everywhere,” McDonald said. “If we’re detecting them now in four state legislative districts, my assumption would be that they’re happening for other local district boundaries as well.”
The mistakes may not be common — the thousands of misplaced voters in Virginia and elsewhere pale in comparison to the millions on the voter roll — but as Virginia districts 28 and 94 showed, those misplacements can carry heavy consequences.
The problem is that these “edge cases” are exceptionally hard to detect using existing tools available to election officials. As the Post notes,
The [State’s] investigation of the 28th House District race was a special project, using a recently acquired and exceptionally limited computer mapping capability to overlay a district map and registration address points to find geographic mismatches. This technology isn’t readily available to many registrars … some use it on an ad hoc basis.
This story is yet another reminder of how complicated – and tedious – the “52 pickup” of voter assignment can be. The silver lining, I suppose, is that these challenges will now be closer to the front of mind for state and local officials in the wake of the next great wave of redistricting … when all those cards will end up on the floor again. I’ll be interested to see how they respond … stay tuned!