Pie Palace

I can’t get over the technical trouble our friends to the South are having with their midterm elections. For some reason they’ve decided to use machines to accept people’s ballots, instead of the tried and true paper method. I’ve been a scrutineer during federal Canadian elections, and I have to say the process is easy, reliable, and visibly secure. It goes something like this:

1. Elections Canada (EC) prints ballots and election lists. It then hires people to handle each poll.
   
2. EC staff show up at the poll.
   
3. A representative of each candidate shows up at each poll.
   
4. Voters cast their ballots. The EC staff and the representatives of each candidate watch, and verify that the person is a legitimate voter.
   
5. The ballots close. Each voting box is dumped out in front of the EC staff and the representatives of each candidate.
   
6. The EC staff classify the votes, with the agreement of the candidate representatives.
   
7. The EC staff count each vote, with the agreement of the candidate representatives.
   
8. The number of votes are compared with the number of voters. Everything is recounted until the number of votes add up.
   
9. The candidate reps get unofficial results from each poll, which they can then verify against the official results published later.
   
10. Everyone gets to go home about an hour after the poll closes. The results are phoned in to Elections Canada.
    
11. Paper ballots are kept on file for any necessary recounts.

Let’s compare this to the US system:

1. Some unknown manufacturer builds a machine that may or may not physically perform as expected.
   
2. Some unknown manufacturer writes software to run the machine that may or may not perform as expected.
   
3. Authorities use voting machines without verifying that the hardware and software works as intended in all scenarios. If you’re a programmer, you’ll probably cringe at this phrase. The only way to verify that software works in all situations is to run it through every possible situation. I mean every possible sequence of keypresses, input, memory allocation/deallocation sequences, ordering of interrupts, power glitches, and network hiccups. Even then, there is no guarantee that the tester has thought of every possible scenario. Unless you want to do some deep mathematical proofs (that may be incorrect because they are based on incorrect information about the software at hand), it is virtually impossible to verify that software will perform properly in all situations.
   
4. Election staff are expected to understand how to use systems they are likely not familiar with.
   
5. Candidate representatives are expected to monitor and verify systems they probably don’t fully understand.
   
6. Voters are expected to interact with a system that they have likely not used before.
   
7. Voters are expected to trust the system to register their vote as they intend.
   
8. A large sum of money changes hands. (from the state to the manufacturer of the unverified voting system)

The paper process has a number benefits:

• It’s easy to use.
  
• Local variation in how counting is done at each poll means that an attack on any one poll may or may not work on another poll.
  
• The involvement of lots of people who have an interest in the system (voters, scrutineers from each candidate, EC staff) mean that all the voting system is closely monitored in each and every poll.
  
• The system is transparent enough that most folks can grasp it quickly, and then notice if something is amiss.
  
• Because there are so many people involved, anyone who wanted to attack the system would either have to get every observer in on the attack, or would have to trick every observer.

I’m not sure what the benefits of electronic voting are. Sure, it shuffles public money to bloated corporations with questionable ethics, makes the system less transparent, and makes the voting process easier to subvert; but I don’t think those are good things.