The documentary
2000 mules was released recently which is a documentary about ballot harvesting, the core evidence was very tech related so i wanted to go through it.
Location Data
most of the data is GPS phone tracking data that shows the weekly movements of a set of devices. They claim that it has to match a pattern of:
- 10 ballot boxes
- 5 non profits.
To determine whether it's "at" a location, they use a geofences of 30 metres and check if the trajectory seems intentional. For example going straight through a fence will not count, they have to stop inside the fence or turn into the area specifically.
With this data they imply that these people are collecting ballots from the non profit organizations and dropping them off at each ballot box.
Are they going to ballot boxes
it is stated in a document to Arizona officials (that i found on twitter) that they used 100 feet radius for the geofence. Much like in Australia, these ballot boxes and voting location are in easy to reach and popular areas such as:
- Libraries
- Parks
- Shopping centers
- Sausage sizzles
An example box I look at with a radius of 30 metres included a 5 star coffee shop. Other included Popular fast food places so this data could be mapping the unhealthy eating habits of non profit works.
is its accurate?
there are lot of legal restriction around personal data but one of the ways they make it more safe is by making it Coarse. Apple clearly states what is Precise vs Coarse location to them:
Information that describes the location of a user or device with the same or greater resolution as a latitude and longitude with three or more decimal places [1]
If the data is Coarse (only three decimal places) this under there terms is not classified as personal data and they can do whatever they want with it, like sell that data.
if the data bought is Coarse then the mule would have a radius of about 1 block. This means its not a single dot walking into a circle, its 2 circles overlapping, that would also make the trajectory calculation dubious because all the movement inside that radius wouldn't be tracked.
The videos
the videos are interesting but they aren't successfully linked to the device data. No person is shown twice. If they had at least shown the car around the ballot boxes, it would be a clear way to put a data point to a person but it seems they only requested CCTV of the ballot boxes and not any other cameras along the route of the mules.
I would imagine showing 5 mules in one city would be very easy if you followed all the CCTV in the routes they took. So while it seems odd watching people shove 10 ballots into the ballot box at 2am at night, its not proof of them being a mule or that its even a crime.
Whats next
I'm interested to get my hands on the data and see what I can do with it and I'm interested to see more videos of the same mules