If you’re reading this post, you probably know a decent amount about computers, math, and (less certainly) statistics, voting machines, and how the electoral college system works.
Suppose that you were on the team of J. Corrupt Fakeypants, a presidential candidate in 2020, and Mr. Fakeypants has asked you to lead the effort to rig the vote in his favor.
Fakeypants has a number of priorities:
1. He does not want to go to jail. So he wants you to do this in a way that does not point a giant neon sign at him.
2. He wants to be President. So he wants you to do a good job of rigging the election.
How do you do it? I think it’s pretty hard.
First of all, Fakeypants needs to be doing pretty well. Nobody is going to believe that the polls missed a 20% swing in the election, and if the results seem that way, there will be lots of people investigating. Fakeypants will probably go to jail.
Second of all, if Fakeypants actually wins, your efforts run the risk of sending him to jail in a scenario in which, you know, he actually wins.
Third of all, every vote you tamper with increases the risk that your vote tampering is discovered, so you want to tamper with as few votes as possible, but…
Fourth of all, if the margin is ultra-close, you will trigger recounts that increase the possibility that your tampering is discovered.
We will stipulate that installing malware on any given voting machine is fairly easy (though each time you do it, you run a very small risk of being caught). And each type of voting machine does require a custom-created piece of malware (and each hacker you bring into your project increases the risks of leaks to your project).
Bringing a well-resourced state actor into play may help you overcome the scale problems of your project, but well-resourced state actors are also the targets of cyber-intrusion from the US security apparatus, and you have a hard time gauging how likely it is that if you ask your buddy V. Putin to help out, the NSA will know immediately.
The voting machines are for the most part isolated from the internet on the actual voting day and during the counting process. You also have no polling that you trust to be better than the overall aggregate public polling as filtered through good aggregators.
What is the method you use to maximize your chance of winning the election and minimize your chance of being caught, with the constraints above? I think that it’s a very difficult project, probably unfeasibly difficult.