In principle, improvement always seems to be a good idea, but views of what might be an improvement do vary. What one person thinks is an improvement might seem quite the opposite to someone else.
In these articles I have tried to argue that balanced approval voting (BAV) would greatly improve elections, but my notion of "better" has always put great weight on ending the two-party duopoly. In my view, that alone would lead to important improvements in our democracy. Still, there are people who think democracy itself is a bad idea and others who claim there is great virtue in the two-party duopoly. On such basic issues, it would be hard to find a middle ground, but for other objections some accommodation might be found.
No doubt there are people who object to BAV merely because they yearn for more options when they vote. For example, they may feel overly constrained with only three options for each candidate. They might think it very important to allow voters to distinguish between their top choice and the two or three other candidates they consider worse, though still acceptable. No doubt such a yearning explains much of the appeal exhibited for ranked-choice voting.
Provided voters are honest and strictly follow instructions, adopting a different score voting system, say with five or seven scores, the two-party system would remain in jeopardy as much as it is with BAV. This would have to be a balanced score voting systems, so there would have to be an odd number of scores and the default score (to use whenever no score, or more than one score is specified on a ballot) would necessarily be the middle score.
While additional scores would provide voters with the extra flexibility they want, in the past we have observed that, these systems could encourage a strategic-voting assault. And we also observed that collecting such detailed information from individual voters is not particularly useful. Elections are to survey opinions across a wide sweep of the population and statistically, a BAV election would infer such information from the simpler responses provided on BAV ballots.
Still, these are technical arguments that might fail to persuade and there may remain a significant block who nonetheless yearn for something more than BAV provides. Might there be a way to gain more support for the improvements that BAV offers? The purpose of this article to suggest such a compromise. It describes yet another balanced and evaluative voting system. Its advantage over BAV is not that it is apt to improve elections but rather that it might serve as a way gain support for adopting it. In truth, nearly this same system was introduced in an earlier article for dealing with the possibility of too frequently tied BAV elections. But in the present context it deserves a name, so let's call it balanced approval voting plus a runoff (BAV+). BAV+ does not add much to BAV besides complexity, but it does little other harm. For the example we assume BAV+ to have five scores.
As the name suggests, voters submit their ballots only once, but those ballots are tallied twice to simulate a pair of elections; the first, and more significant, tally simulates a BAV election and then the second selects a winner from the runoff candidates.
The candidates in the runoff might, for example, be the two with the highest net votes from the initial BAV election. But we leave such details as an implementation detail; strictly speaking then, BAV+ is not a single voting system but rather a family of voting systems differing in how this selection is done. As well there might be more than just five scores used in the second election.
It seems best to keep the number of candidates in the runoff small, however. And in general, five scores would seem to be sufficient. Five is the number of scores we assume in what follows.
The ballots for BAV+ are in essence what you would expect for score voting with five scores. Voters might be asked to choose one of the scores would be -2, -1, 0, 1, 2. Alternatively, they might be given a ballot similar to the one illustrated below, and asked to check off at most one box for each candidate. When a ballot specifies no score for a candidate (or more than one score) that will be treated as having been assigned 0, the middle of the odd number of scores.
For the sake of clarity, it is convenient to use the scores -2, -1, 0, 1, 2. However, I suspect that for an election, voters would find a ballot more like the one illustrated above to be less confusing.
For the initial tally (of the BAV election) the ballots are counted by treating -1 as -2 (and 1 as 2). In essence, the distinction between opposition and weak opposition is simply ignored, just as the distinction between support and strong support is ignored. With these reassignments, the sum of the scores is twice the net vote for the simulated BAV election when computed as the count of negative votes (whether -1 or -2) subtracted from the number of positive votes (either 1 or 2). Multiplying the net vote by two (or any positive number) changes the tally, but in a manner that has no effect on the election outcome (however, doing so happens to be convenient for what follows below). The BAV tally being now complete, a few of the top contenders move on to the final runoff election.
The final runoff election is then conducted as a score voting election, (for each candidate) simply adding up the score ( -2, -1, 0, 1 or 2) from all of the ballots. To some, his BAV runoff election may seem unconventional because 0 assigned as the default value when either a voter specifies no score or perhaps more than a single score for a candidate. Traditional score voting typically ignores that there is any need for a default score but when an actual election forces a choice to be made, the traditional choice would be to use the smallest score, -2.
Notice that by using BAV+ we introduce the possibility of an effort by some large conspiracy of voters to gain an advantage by steering clear of the scores of 1 and -1. By sticking strictly to specifying only the maximum or minimum scores (in effect, restricting to only -2, 0 and 2) the conspirators would strengthen their voting power and improve the chances that one of their favored candidates to win. But notice that the conspirators are simply altering their votes to exactly match how those votes were tallied in the initial BAV election. Their conspiracy would improve the odds that the winner of the BAV+ election will be the same as the winner of that simulated BAV election. The advantage they stand to gain through their conspiracy is to reduce the opportunity for some candidate that the conspirators oppose to overtake their preferred candidate in the runoff, but that may not be what they would want. While this effort could affect the outcome of the election, it is not at all clear how to predict which candidate that would favor.
Provided many voters participate in a BAV+ election, the likely winner of the runoff would seem likely to be the same as the winner of the preliminary BAV election. Unfortunately, without evidence from several BAV+ elections we cannot be entirely confident of this. But if, in fact, this prediction turns out to be true, support for simplifying elections by adopting BAV would likely grow. So, this could possibly be a way for eventually having BAV adopted.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
