Preliminary considerations about a voting system tends to be abstract, focusing on a simple description without getting bogged down with practical details. It is understood that some adjustments may be needed when faced with practical application but in early stages it is best that such complications are ignored.
Unfortunately though, later there is a tendency to sidestep the required attention to these minor alterations and just assume the consequences will be insignificant. But that is not always so. For example, in an effort to argue that approval voting is balanced, some discussions made the mistake of treating as unimportant the fact that in a real-world election, voters would surely abstain with regard to one or perhaps several candidates; but we show a quite plausible example of an election where such abstentions were in fact a critical factor. Likewise, with ranked voting a simplified theoretical model might well assume that voters will provide a complete ordered list of all candidates. This is a very convenient and simplifying assumption, but in actual practice, some voters will surely fail to list all of the candidates. Does that difference matter? Perhaps not so much, but even so, an examination of that issue is enlightening.
Promoters of IRV or some other ranked voting system are quite happy to concede that a voter does not need to list all candidates; that freedom does seem to make the system more appealing. But let us now give that issue some of the attention it actually deserves.
Just as with plurality voting, IRV provides no way to express opposition to a candidate; at most, one can only deny support for that candidate. But in the case of IRV, care is needed simply to deny support. When you, as an IRV voter, put a candidate on your list, you are showing a level of support to that candidate; it is strong support when the candidate is high on the list but even at the very bottom of your list it still is an expression of support. There is an exception to this, but only for the very last candidate on the list when you list every single candidate; that is because the last candidate on a complete list will never be tallied.
With IRV, when you leave a candidate off of your list, you are in fact giving that candidate no support and that is a lower level of support than you give to any of the candidates you do explicitly include in your list. If there is a candidate that you oppose more than any other then you can feel free to leave that one off of your list. Moreover, if there are several that you oppose equally then you can leave all of them off your list. But aside from this one special case of multiple candidates that you oppose equally, you actually do need to list, in order, all of the other candidates. For the diligent voter then, the option of submitting an abbreviated list does not really simplify much. But more than likely, some voters will fail to understand the impact of omitting candidates. Rather, they may think that by placing a candidate as the last entry on the ballot they are emphasizing a special dislike of that last candidate, whereas the effect is to give the candidate some support.
To be fair, the enthusiasm shown for IRV is likely to stem from its avoidance of the spoiler effect. IRV is indeed an improvement over plurality voting in this respect, but that seems a fairly low bar to clear. The great enthusiasm there has been for IRV is curious because IRV exhibits some serious fundamental flaws.
It had long been recognized that plurality voting is a faulty system except when there were just two candidates. As a fix, IRV was invented so that plurality voting was used not once but many times, removing one candidate at a time until there was just one remaining as the winner. That way, at least that final decisive application of plurality voting with only two remaining candidates avoided using a clearly faulty method. But in all of the preceding steps there were more than two candidates and the choice was doubtfully the right one. The rightful winner of the IRV election could have been eliminated in a faulty early round of counting. Before IRV was promoted to gain such widespread support, did anyone bother to carefully think about this?
But that was not the only questionable decision in the thinking that motivated the design of IRV. The fundamental idea behind IRV was that one by one, candidates would, through a democratic process, be removed until only the ultimate winner remained. So you might naturally think that voters in these single-elimination votes should be asked which candidate should be removed next. But instead, IRV does exactly the reverse of this; it asks each voter to specify which candidate they now want most to win. And that is a quite different question.
A candidate who who is supported by more voters, though with somewhat less enthusiasm will be eliminated instead of another candidate who has a smaller number of very enthusiastic supporters (even perhaps with a greater number of enthusiastic enemies). Again, this raises the question of why no one bothered to think this through more carefully before promoting the adoption of IRV.
In the article on ranked voting referenced earlier, we considered IRRV, the voting system which this very observation seems to suggest. The difference between IRV and IRRV is that IRRV successively asks voters which candidate they would most like to next be removed, not which candidate they would most like to retain. Unfortunately, discussion in that article becomes a bit muddled due in significant part by the complication of voters submitting incomplete lists of their preferences. The article would have been more clear if it had simply stipulated that voters would always submit complete lists, ranking all of the candidates. With that (dubious) assumption we could have simply argued that an IRRV voter would submit exactly the same list as an IRV voter, but in the reverse order. A simplified theoretical view tends to be more clear than when the real world intrudes and in this instance it would have provided more clarity without much actual loss of reality.
