A group of respected MIT researchers presented their view on the current state of internet and blockchain voting. We’ll refer to their document as the “MIT paper”. Despite the developed framework for scrutiny of such voting systems appearing reasonable on the surface, its application lacks fairness due to the fact that the primary use was designed to debunk a rival system, and in the process it significantly entangled two vastly different conditions that need to be looked at separately: remote voting and in-person voting.
The analysis below represents an unbiased comparison of voting systems that are publicly available right now.
In short, the MIT paper scrutinizes 5 areas, which represent groups of well defined criteria.
1. Ballot secrecy
a. Eliminates the possibility to buy votes
b. Eliminates possible coercion
2. Software independence
a. An undetected change or error in a system cannot cause an undetectable change in the elections outcome:
- At the stage of ballot issuance;
- At the stage of casting a ballot;
- At the stage of counting the votes.
b. It is possible to check the correct behavior of a voting system at each stage:
- Recorded votes are as intended;
- Collected then as recorded;
- Counted then as collected.
3. Voter-verifiable ballots
a. Before a ballot is cast, a voter must be able to verify that the prepared ballot reflects her intended choices.
a. When an error is detected, can the one who detected it convince others that an error indeed occurred?
a. Detection of any misbehavior doesn’t need to be perfect; it only needs to happen with sufficiently high probability.
The above criteria are clear except one, which was modified with respect to the original MIT paper for good reason. We believe that the primary emphasis of the paper being on the software independence criterion was made so incorrectly. It was presented as though people should have the ability to perform a manual check without using any software, since the system is supposedly otherwise too complex. This we equate to stating that the safety and integrity of an airplane or a skyscraper should undergo a manual inspection without the use of instrumentation since the instruments are too complex. In fact we feel very strongly that the correct approach is the ability to check a voting system by any software that the public would like to use, which it so happens was mentioned as acceptable in footnote 8 in the MIT paper.
Voting systems comparison
For simplicity’s sake, below is a comparison where we gave provisional scores for criteria to assess which system is more secure.
0 if the security is equal,
+/- 1 for more/less secure with respect to blockchain compared to mail-in voting.
A blockchain based solution for remote voting does not show an overly significant advantage compared to mail-in voting (above) as much as it does for in-person voting (below) with a score of only +4 out of a total 11 criteria; nevertheless, it still wins.
As it relates to in-person voting, a blockchain solution clearly wins over paper-only ballots with a score of +1 for 9 of 11 criteria. For in-person voting there are no criteria from MIT paper, which can be used by an honest opponent that can prove paper-only voting is better than blockchain one.