Virtual Voting Machine Demonstration

Minister Cullen's argues that the voter can be sure that his/her vote is recorded correctly because he/she sees it displayed on the machine.

Here are two simple programs which demonstrate that a direct recording electronic machine can display one vote while recording another.

"Implementation A" records votes correctly and "Implementation B" records the opposite of each vote (all 'Yes' votes are recorded as 'No' and vice versa).

The difference between the two programs is one word in the source code, demonstrating that a small change in the source code can radically change the result.

You can download an uncompressed version of Implementation B here.

The problem of lack of transparency can be demonstrated as follows:

• Have a few people vote using Implementation B, with the Results window closed.
  Each person should not see the others votes.
• Can each voter tell whether his/her vote was recorded correctly? How would they know?
  You could test this by running one implementation, not telling the voter(s) which one you are running. Ask them to vote and ask them whether they think their vote was recorded correctly.
  Repeat this a few times, since they have a 50% chance of guessing.
  If they cannot tell which version is running (without looking at the results or seeing which one you ran), then they could not tell whether a real direct recording electronic (DRE) voting machine was recording votes correctly.
• After everyone has voted, show the results.
  Unless all voters have voted the same way, each voter will not be able to tell from the totals whether their vote was correctly counted. They might be suspicious, but could not confirm their suspicion without the ordered list of votes shown in the bottom section of the Results window (which would not be available in a real election, for reasons of privacy).

This is available for Windows 95 and later only.

The source code is provided for any programmers who are interested. It's written in Borland Delphi (Object Pascal), simply because that's what I usually use for Win32 development. The difference between the source code of the two programs is the omission of the keyword "not" in one place in "Implementation B".

I (John Lambe) have made two separate copies of the whole source, rather than using a conditional define since this is probably easier to understand for non-programmers.

You can distribute this freely.

This idea was proposed by Colm MacCarthaigh.