The Commonest Arguments For OPV Are Overrated

In the leadup to the Queensland election (which I'll have a roundup post on overnight) there's been some undignified arguing about whether Queensland should use optional or compulsory preferencing, with both major parties accusing each other of seeking to rig or corrupt the system.  Opposition Leader David Crisafulli has repeatedly used misleading language to attack compulsory preferencing, claiming that it forces voters to "vote for" candidates they don't want.  (It actually only forces them to rank them, which lacks the same level of obvious moral repugnance).   

In New South Wales, the debate about optional preferencing is typically a lot more sedate.  Optional preferencing has obvious formality benefits in NSW especially at state elections (though this can flow on to worse informal voting rates at federal).  Perhaps partly because OPV is in the NSW Constitution, and perhaps because any government wanting to do so would have a tough time getting its way in the upper house, there doesn't seem to be any serious push to get rid of it.  It also seems that for whatever reason, non-major-party candidates (who might be thought to be disadvantaged most by OPV since they most often rely on winning from behind) actually do well in NSW state elections, much better than in Victoria.  

However, in Queensland, where there's no upper house and the election winners can write the rules for the next election, the debate is an on and off culture war item between the major parties.   In Queensland, OPV first came in from the 1992 election after it was recommended by the post-Fitzgerald Electoral and Administrative Review Committee, although the arguments considered had very little to do with corruption.  It was repealed for the 2017 election in unusual circumstances.  Katter's Australian Party and the LNP had supported an electoral amendment bill to increase the number of seats from 89 to 93.  The minority Palaszczuk government moved an amendment to add the return of compulsory preferencing, which then passed and the whole package was then passed over the LNP's vehement objections.  

The funny thing is that because of the unusual flows in the 2015 election and the increase in the One Nation vote since, preferences played a far greater net role in the 2015 Queensland election under OPV than the two CPV elections in 2017 and 2020 combined.  In 2015 the LNP lost nine seats to Labor that it had led on primaries.  In 2017 under CPV, the LNP lost four to Labor, Labor lost two to the LNP, and the majors lost two and one respectively to the crossbench.  In 2020 preferences changed the outcome in just two seats: Mirani (Labor overtaken by One Nation) and Nicklin (LNP overtaken by Labor).  

OPV lacks the most repugnant aspect of first past the post (that it imposes strategic dilemmas on some voters but not others, in the process disadvantaging startup parties) and is good at keeping unintentional informal voting down, unlike compulsory preferencing,.  That said, Queensland does not have a big unintentional informal voting problem.  The reintroduction of compulsory preferences in 2017 saw the informal rate rise from 2.11% to 4.34%, and in 2020 this fell to 3.40%.  

What I want to tackle here is the main philosophical arguments for optional preferential voting.  I believe these arguments are much weaker than OPV supporters think they are, and that pushes for OPV based around them should be resisted.  

An argument I'm not dealing with at length is that OPV is quick to count.  It isn't that much quicker - nearly all the delay in counting Australian elections comes from the time taken to wait for postals to be received, and a lot of the delay in on-the-night counting relates to the size of primary prepoll counts, by which time there is usually a good idea of preference flows in a seat.  I'm taking it as a given here that we should have the best voting systems even if they might cost a few percent extra in counting costs or make the results and projections a little slower on the night.

The main arguments I am addressing here are:

Argument 1.  The voter should not be forced - as a condition of a formal vote - to rank candidates who they dislike

Argument 2. The voter should not be forced - as a condition of a formal vote - to choose between pairs of candidates they have no view about or are indifferent to

Argument 3. The voter should not be forced - as a condition of a formal vote - to rank all the candidates no matter what their reason for preferring a partial ranking may be

If argument 3 succeeds then arguments 1 and 2 become unnecessary to consider, but arguments 1 and 2 are more intuitively attractive and argument 3 doesn't tend to have any basis but "freedom!", so I will focus more on the first two.  If a voter morally objects to helping a candidate to win in any way, should they really be forced to possibly help that candidate to win just to have their vote for someone else counted?  And if a voter really has no opinion between two candidates, isn't making them pretend to have one just making their vote a less reliable representation of their views rather than more, and wouldn't letting them exhaust their vote be better?  To make any kind of argument against these, I need to point out some negative consequences of OPV.

Voters Who Exhaust Who Shouldn't

The main negative consequence of OPV is that it causes some voters to exhaust their votes who do in fact have a preference between candidates or who are at least capable of forming a preference if they need to do so for their vote to count.  The two most important types of such voters are:

1. Satisficers: The satisficer is a voter who will do the bare minimum for a formal vote in order to tick their democratic duty box without expending unnecessary effort or time.  We don't know how many OPV voters who just vote 1 always do it, or how many of these are satisficers as opposed to genuinely only having a view about one candidate/party, but there is a lot of evidence that large numbers of Australian voters satisfice, whatever the system.  The starkest example comes from the NSW Legislative Council, where 83% of voters voted 1 only for a party above the line in 2015, creating a vote that would exhaust once that party had had all its candidates elected or excluded.  But when the same voters started being required to number multiple boxes in the Senate in 2016, over 90% suddenly became able to find six parties they were able to support - and most of these voters were choosing for themselves and not copying how to vote cards.  Also, while some voters responded by satisficing the 1-6 requirement (eg picking a party and numbering the next five boxes) the number doing so is relatively small.  (The 1 only rate in the Legislative Council has since declined, perhaps through contagion effect, but is still far higher than in the Senate.)

2. Confused Voters: Some voters have no problem with ranking candidates in order but still don't understand how preferential voting works, and for this reason think that it is a good idea to stop once you've numbered all the candidates you like.  There is a lot of resistance to understanding the concept of helping a lesser evil beat a greater one.  There is even a fairly common myth that putting a candidate completely last might help them beat your preferred candidates and therefore that it is better to stop numbering when you can.  (Savings provisions under CPV accommodate this and allow voters to leave the very last box blank and still have their vote counted provided there is no marking in that box at all).  

One might say here that if a satisficer wastes a preference they could have had, that is their choice; a fool and the value of their vote are soon parted; where's the victim?  However, this has a similar problem to the argument that "optional voting" (actually optional booth attendance) is fine.  Under voluntary voting, the satisficer doesn't even get off the couch because there's nothing to satisfice.  If a certain demographic of voters is inclined to not vote at all then it is not only the individual non-voters who are affected by that, but also other members of their cohort get under-represented even if they do vote, and this undermines the mandate of a government as representative.  This is the main reason why I support what we call "compulsory voting", and the same thing is a strong argument for compulsory preferences. 

I add also that a special problem with OPV is that parties deliberately seek to increase the number of confused voters by putting out "Important: Just Vote 1" type signs, implying that this is in any way an intelligent way to vote rather than lazy and potentially harmful to the voter's interest.  Often these signs loosely resemble official electoral signage, and there have been cases where the same party will put out "Just vote 1" in some seats and "Number every box" in others - usually with an aim of trying to influence the preferencing behaviour of voters from rival outfits depending on the needs of the seat.  I might take OPV more seriously if its supporters took this problem seriously, but they don't.  I could never support OPV without it being accompanied by laws that absolutely smashed this behaviour by providing for career-ending and election-forfeiting consequences for people who do this stuff.  

OPV Does Not Respect All Partial Preferences

OPV supporters like to pride themselves on how they respect the voter's right to lodge an incomplete list of preferences and still have their vote counted.  But in fact this respect is selective.  Consider the following voters in a contest between six candidates, A through F:

Voter 1: "I like Candidate C, I think candidate D is OK, I've never heard of candidate B and I hate candidates A, E and F.  I find it morally objectionable to preference any of A, E or F above any of the others, even though I hate all three.   I'd like to vote 1 for C, 2 for D, 3 for B and stop."

Voter 2: "I really don't like any of the candidates, but I especially detest candidate D.  I strongly dislike C and I'm not that keen on F.  I've thought carefully about A, B and E and they're really all the same. I object to putting any of them outright first as none of them deserve so much approval.  I'd like to vote against F, C and D by putting them 4th, 5th and 6th in that order, but make no choice between the top three, or if I have to by putting the three equal first".

Under OPV, Voter 1's partial ranking (effectively 1,2,3,4=,4=,4=) is formal, but Voter 2's partial ranking (1=,1=,1=,4,5,6) isn't.  Yet Voter 2's partial ranking of the candidates is just as valid an opinion about the candidates as Voter 1's.  Indeed it's quite possible Voter 1 has an actual order they could honestly rank the hated candidates in if they had to.

I've never seen pro-OPV arguments open the door to partial rankings that are not top-down, even though it would not be that difficult to write a counting system to accommodate such votes, especially in NSW which has data entry.  For instance a 1,1,1,4,5,6 vote could be treated as a third of a primary vote for the first three candidates, becoming half a preference vote each for two if one of these was eliminated, and so on.  A challenge would be interpreting different methods of assigning equal votes (eg one voter might write 1,1,1,4,5,6 while another might express the same preference as 1,1,1,2,3,4) but if we are serious about the idea that an incomplete list of preferences should be respected and reflected, why only respect and reflect those that are top down?  

Ranking Disliked Candidates Is Not Objectionable

The claim of OPV supporters in argument 1 is that it is morally objectionable to require a voter to choose between two candidates they do not like in order to cast a formal vote.  But why is it so?  If it was forcing a formal voter to express positive approval of one of the candidates, that would be  objectionable, but it isn't - the voter is simply ranking the two relative to each other.  If it was forcing a formal voter to help a disliked candidate beat a candidate they liked, that would be objectionable, but it isn't - if a voter ranks the candidates in order of preference, their vote can only reach a disliked candidate when nobody else is left but candidates the voters dislikes as much or more.  At this point, someone at least as bad as the disliked candidate is going to win anyway, so the voter is not doing their own preferences or interests any harm by making a choice.

It might be that the voter is not covered by arguments 1 or 2 but wants to withhold their preference strategically.   A strategic withholder is usually a left or right flank voter (say Greens or One Nation) who thinks one of the majors is really bad and the other one is mediocre/disappointing, but under OPV chooses to exhaust their vote.  The strategic withholder might want to exhaust their vote as a tactical way of sending a message to the major party closest to them that they want that party to do more to appeal to them and earn their preference.  But there is no reason for the voting system to accommodate these people.  If there are any significant number of them then they can have much the same net effect under compulsory preferencing by each deciding that since they're not allowed to exhaust their preference between the majors, they will each flip a coin.  'Sorry Albo, weak on Gaza, heads you get my preference undeservedly, tails it goes to Dutton even though he's worse.'  Yes, this introduces a random element to the result but it's insignificant compared to the random element introduced by ballot draws.  The withholder might say that they object to even a chance of their preference going to the party they want to withhold from, but since they've already decided to withhold a ranking that they actually hold for strategic reasons, I don't have any sympathy for that.  

I believe that the number of voters who actually do end up randomising (or donkeying part of their ballot) because they actually can't make up their minds under CPV is low (though I've partially randomised my own ordering on longer ballots at times) and the main reason for that is nearly all voters can actually make a preferencing decision if they have to, or else are content to copy their party's recommendations.   

OPV support is often self-centred

When I listen to the arguments of OPV supporters I see a lot of "me" and not a lot of "us".  Some voters would like to not make choices between lesser and greater evils or between parties that they hate.  Some voters would like to not have to spend any time thinking about obscure candidates.  It's all about the feelings of these petals whose psychological comfort levels are apparently the most important thing, and not so much about picking a system that works best in representing the community overall and in protecting less informed voters from having their opinions under-represented.    We know regarding the latter that preferencing rates under OPV differ wildly by party in a way that looks very much connected to the education levels of party supporters, and that is bound to also break along English literacy lines.  

I also think that parties that appeal to voters with arguments based on the self-centred aspect of making voting easier, are themselves making that appeal based on their own self-interest when it comes to ranking other parties, not the voter's.  In particular, the LNP has run so strongly on OPV firstly because it doesn't like getting flak for recommending preferences to One Nation, and secondly because it doesn't like having to choose between Labor and the Greens in inner city seats.  And, of course, OPV reduces the flow of preferences from Greens to Labor in seats where the Green vote is high and the right-wing minor vote is weak, so the LNP may see some advantage there despite the last two elections not really showing CPV causing them much of a problem.  

Halfway houses are better than OPV (or CPV)

Having noted that I don't like fully-fledged OPV, I've also noted often enough that I don't like how compulsory preferential voting is executed - in particular the lack of savings provisions for voters who make genuine and often irrelevant errors.  But that's not reason enough to embrace full OPV with all its problems, especially not in Queensland where voter error is a less significant issue.  There are many intermediate systems, many of which I covered in this 2020 article.  Without going into the world of "just vote 1" with its ludicrously high exhaust rates and severe disinformation problems, we could increase the number of votes that are at least partly counted while still instructing voters to number multiple boxes or all boxes, and hence keep the preference flows up.    In the Senate, it proved impossible to maintain full preferential voting while allowing for practical voter control over preferences and also high formality, and hence we do have a semi-optional preferencing system there, one which is working well.   Many of the hybrid solutions allow for those purists who insist on deliberately exhausting their vote and who actually know what they're doing to do so without the unintended consequences of pure OPV.