The best counter-example I saw was to take a page from James T. Kirk: You change the rules of the game. The word "fair" here is typically used to in the contex

The best counter-example I saw was to take a page from James T. Kirk: You change the rules of the game. The word "fair" here is typically used to in the context of there being an unambiguous "winner". i.e. someone that everyone can agree won the most votes, or is most representative of the votes as cast. You can change that definition slightly, to "a fair vote is where everyone can agree that the process used to choose the winner is unambiguous, and that there is a clear winner." The method for this kind of "fair" vote is to to count all votes as cast. Then, pick one voter at random. That vote determines the winner. This is a "probability" method of picking winners, not a "counting" method. This is arguably "fair" in that each candidate has a probability of winning in direct relation to the votes they received. Each voter has their vote counted identically to all other voters. It has other benefits. There are no "safe" areas which can be ignored, because the candidate is assured of getting sufficient votes to win. Instead, every incremental vote is an incremental increase in the probability of the candidate getting elected. The "counting votes" problem has similar issues to share splits in companies. If a company has 3 share holders (49%, 49%, and 2%), and requires 50%+1 votes for consensus, each share holder has an equal amount of power. This is arguably an imbalance with respect to the shares they purchased. In the probability method of picking winners, each voter has exactly the same power, which is arguable the most "fair".