Voting should be a quick and painless process, but often isn’t. Every election cycle there are reports that people are waiting in line for hours to make their choice. Michelle Obama even warned during her speech at August’s Democratic National Convention that would-be voters in the 2020 US presidential election should make a plan and “pack a brown bag dinner, and maybe breakfast too”.
By election day, postal ballots will already have been sent in from across the country and early voting conducted in many states. Yet the US has a labyrinthine election infrastructure that changes from county to county, with huge differences between areas.
Here, the New Statesman takes a visual trip around voting inefficiencies in America – inefficiencies that end up with many citizens simply not having their political voice heard.
This graphic simulates the impacts of resource allocation and voter density on queuing time at the polls and people exiting the queue without voting. Every simulated 30 minutes – around eight seconds on screen – there is a 1 per cent chance the voter leaves without voting.
There is also variance from state to state. In Georgia and Wisconsin, some polling locations were closed in the spring primaries due to the pandemic. This means more people had to funnel through fewer voting booths. Despite a reduced in-person turnout, with mail voting increasing from 10 per cent to nearly three quarters between the 2016 and 2020 primary, the number of people per in-person polling site in Milwaukee was seven times higher in 2020, according to data from the Wisconsin electoral commission, with over 10,000 people per polling site.
Speed up or restart the graphic, which shows the average number of voters per hour turning up in Milwaukee to vote in person, with the buttons below.
There are other resource shortages, too. When a voter arrives at a polling station in America, in two thirds of precincts, poll workers sign in voters using a separate piece of voting hardware called a poll book. This allows voters to go to any polling location to vote, with the downside of requiring an internet connection to a database that verifies and updates who has already voted.
This, in theory, should make it easier to vote. But in California, poll books were found to be the cause of delays. In LA county’s March primary, 30 per cent of precincts had at least one poll book that did not register a single voter, and poll workers had issues bringing poll books online when demand increased on election day. When poll books were offline, people had to cast provisional ballots – a paper-ballot alternative that slows the counting process down as it requires signature validation, which is another hurdle for counting the vote.
Once checked in, the voter is often presented with an electronic voting machine rather than pen and paper. In Louisiana, and in some parts of Texas and Tennessee, this could be an electronic voting machine with no paper record, meaning there is no way to verify your vote was correctly tabulated. A recount with this system therefore involves a computer re-counting electronic records from the location.
In states where there is a paper trail, votes can be tabulated by machine rather than counted by hand. This means the printed paper is scanned by a computer and pixels counted to determine where the voter marked their ballot. In Georgia, the hardware used is configured to capture a lower resolution image than it is capable of producing. This was described by election security consultant Harri Hursti in court testimony as “like driving your sports car locked on the first gear”.
Georgia also categorises whether a ballot is valid or not using “arbitrary” default settings. These settings consider any ballot oval filled in less than 35 per cent as an “ambiguous mark”, referred for manual review. Ovals judged to be less than 12 per cent filled are discarded – without being flagged for review. You can see what this means in practical terms by using our interactive below.
Voting inefficiencies are not evenly distributed – and it is often disadvantaged groups who suffer the most. Analysis of the 2012 wait times by race (PDF) showed that while white voters waited on average 12 minutes to vote, black voters waited 25 minutes. One reason that a voter might find their location has queues is the type of hardware used. The average wait time of polling sites using direct-recording electronic machines was 17 minutes, compared to 11 minutes for areas using optical scan machines.
The wait time difference could also be due to the imbalance of resource allocation – voting hardware and poll workers – sent to different precincts. The analysis calculated that a mostly white precinct had 90 fewer people per voting machine compared to minority precincts. Resource deficits continue, with Georgia voters in the June primary finding a bottleneck at their precincts where a single scanner was allocated to a location.
Small differences in election infrastructure add up over the course of a 12-hour polling day. A seemingly insignificant problem can snowball until it has a significant impact on turnout – as people leave queues, give up on opaque processes, or don’t bother to try to vote due to negative past experiences.
You can get a sense of this by using our interactive below. Change the failure rate of the poll books, the number of voting machines, the likelihood of a voter leaving the queue and the number of voters arriving – and see the impact it has.