Hacker News Reader

Comment by jtbayly | Hacker News Reader

Not quite what you're asking, but will feel like it to people, here's a quote from the linked article (quoting somebody else):

"One problem of RCV is that especially with lots of candidates, a winner can be chosen with a relatively small number of first-place votes. This was exactly what happened in the city of Oakland, California, in 2010 when Jean Quan, who received only 24 percent of votes in the first round, ended up winning the election because she was the second and third choice of many voters. In Burlington, Vermont, after the leader in the first round did not win an election, RCV was repealed."

It might be a technical majority, but when so few people actually preferred the candidate that won, it makes sense that they repealed it.

The benefit of the simple plurality is that you've got to actually get people to want you.

Not sure what I think about RCV, personally.

View on HN

I'm not sure why this is a flaw, to be honest - it seems like selecting a candidate who is at least acceptable to the greatest number of people is preferable to picking one who is preferred by a slight majority but possibly highly divisive.

The heart of the problem is that what people think they're getting probably isn't going to match up with what they actually get. I don't think RCV is really any worse or better than FPTP (well it's probably a little better), but it behaves in a way that is going to offer weird results for people that they didn't expect. That's why it's bad.

At the end of the day, no one cares how we decide our leaders so long as their candidate wins. And when their candidate loses, that's when there are suddenly massive flaws in the process.

It's not a flaw. In two-party systems there's often a strong pushback on any system that would break the deadlock. Those arguments won't make any sense, but sadly they don't need to in order to convince people.

I think it's not necessarily true that the elected candidate really is acceptable to most people under RCV. My #2 vote might simply be my "least-bad" choice, not my "second-best" choice. I might be pretty disgruntled if that #2 option is chosen.

Imagine a system where each person gets 100 votes to allot as they wish. Say I give 97 votes to A, 2 to B, and 1 to C, while you give 35 votes to A, 33 to B, and 32 to C.

If we instead had an RCV system, we'd perhaps both express those votes as "A over B over C", even though under the vote-allotment system we were expressing pretty divergent opinions. I clearly find candidate B unacceptable, if barely more tolerable than C, where you appear to find all three of them more or less acceptable.

Isn't that the whole point of RCV? People can put their first choice as the 3rd party candidate that they prefer, but if they still don't have enough they can fall back onto a more established party for their 2nd/3rd choice.

In Austriala the above happens all the time it is considered normal.

http://www.abc.net.au/news/federal-election-2016/guide/mpor/

Here the Labor party candidate polled second highest (with 27%) and was elected thanks to Green Preferences. The Liberal candidate lost despite polling 42% of first preference votes.

Heaven forbid a compromise, eh?

I would claim this is true of FPTP too - just in FPTP, people vote for their 2nd, or 3rd favorite candidate. At least with RCV your preference is known, rather than hidden in some corner of your mind.

That's not a flaw. One reason I prefer Approval Voting is that it makes this clear. Approval Voting works the same as RCV but without ranking. Vote for all candidates you approve of.

The "bayesian regret" is almost as good as RCV, but it's much simpler and doesn't suffer from the misperception you highlighted.

https://electology.org/approval-voting

The Bayesian Regret of Approval Voting is radically BETTER than with IRV.

http://scorevoting.net/BayRegsFig.html

Approval Voting is problematic, especially in a multicultural electorate, for situations where there is not a concrete definition of the meaning of the approval threshould, because different people interpret "approve" differently (it's basically asking the to impose a binary threshold on a continuous-but-not-consistently-quantifiable value), so it suffers problems from inconsistency in input data. The impact of any individe all voter is small enough and hard enough to viscerally percieve that there is no strong feedback method to, over time, align understandings of "approval" so the meaning is consistent across ballots.

Approval Voting is not problematic for situations where you don't need ballot secrecy and approving is, e.g., a binding commitment to participate if the approved option is selected (or where non-approval is a binding opt-out), or where some other consistent concrete consequence of approval exists.

A similar problem exists with most numerical rating (as opposed to ordinal ranking) methods, and to other ranking methods with smiller number of rankings than alternatives, and, perhaps, to a much lesser extent, to ranking systems that allow (but don't require) equal rankings.

Because most analyses silently assume a consistent the mapping from an internal utility function to ballot marking, this is frequently overlooked.

Reply on news.ycombinator.com