Talk:Spoiler effect
From Wikipedia, the free encyclopedia
- Alternative voting systems which avoid the spoiler effect include instant runoff voting, also known as single transferable vote.
This is wrong on two accounts:
- IRV and STV are not the same thing. The first is a single-winner system; the second is a multi-winner system. The second reduces to the first in the case of a single district.
- IRV and STV can suffer from the spoiler effect. For example, if the "p-ist" vote was split between 3 p-ists, the most popular and "major party" p-ist could be eliminated early, even though they would handily beat the winning q-ist.
Contents[hide] |
[edit] IRV & Spoiler Effect
IRV is free from the spoiler effect, if implemented correctly, because it meets the Independence of Clones Criterion (ICC). See: [1]
--
I've read many other places that IRV only stops the spoiler effect when the third party is weak. In the scenario where the third party becomes almost an equal, the spoiler effect can come back into play. For example if Nader was looking like he might be almost an equal with Bush and Kerry, and we were under IRV, liberals might be afraid to vote for Nader because if it was a runoff between Bush and Nader it would seem more likely that those who ranked Kerry first instead of Nader would put Bush second.
IRV isn't free from the spoiler effect at all. In order to be "free" from the spoiler effect, it would have to be free from Independence from Irrelevant Alternatives Criterion (IIAC). It's not. It's not even compliant with relaxed versions of IIAC, such as Local IIAC. --- RobLa 03:51, 12 May 2004 (UTC)
12-17-07: I disagree somewhat with RobLa's claim that freedom from spoiling requires independence from irrelevant alternatives. Suppose candidates can withdraw from contention after election day, that the votes are published at the end of election day, and that withdrawn candidates are deleted from the votes before calculating the winner. If Nader by not withdrawing would elect Bush, and by withdrawing would elect Gore, Nader would have an incentive to withdraw and not be a spoiler. (Nader's supporters who prefer Gore over Bush would bang on his door until he withdraws. In 2000, Nader admitted that Gore was the "lesser of evils.") Candidates who choose not to run out of fear of being a spoiler would no longer have that fear.
A huge part of the "spoiler problem" is that some potential candidates choose not to run. For example, in 2000 John McCain chose not to run (as an Independent) in the US presidential election after the Republicans nominated George Bush, because he knew Gore would win if McCain and Bush both ran. (McCain and McCain's supporters preferred Bush over Gore.) One cannot measure the spoiler effect by considering only the votes and the candidates on the ballot. (By the way, McCain was probably a Condorcet winner in 2000. Most of Gore's supporters preferred McCain over Bush, and most of Bush's supporters preferred McCain over Gore, and a significant number of voters thought McCain was better than both Gore and Bush.)
I agree IRV has a serious spoiler problem and that IRV would maintain the "two big party, each party nominating one candidate per office" system, would tend to defeat centrists by squeezing them between a candidate to their left and a candidate to their right, would therefore deter candidates and parties from taking moderate positions on the issues, and would therefore continue the polarization we've seen under plurality rule. For instance, the Democrats would not nominate both Obama and Clinton under IRV because that could easily elect the Republican:
30% 17% 8% 45% Obama Clinton Clinton Romney Clinton Obama Romney x (x = doesn't affect Romney Romney Obama x the example)
IRV begins by eliminating Clinton, who was ranked top by only 25% (17% + 8%) of the voters. Then IRV counts 8% more for Romney, giving him a majority (45% + 8%). If Obama had not run then IRV would give Clinton a majority (30% + 17% + 8%). Since Obama's supporters prefer Clinton over Romney, this is classic spoiling.
The official book on Robert's Rules has a section on IRV. The book's authors call IRV by the generic name "preferential voting," a hint they were unaware other preference order voting methods exist. They wrote IRV is better than plurality rule, which should not be considered a strong endorsement. They point out that IRV can defeat the best compromise, which is the same as saying IRV is susceptible to spoiling. Presumably, if they had been aware of Condorcetian voting methods they would have endorsed one of those, since Condorcetian methods are better at electing the best compromise.
IRV can be patched fairly well by allowing the candidates to withdraw after the votes are cast, as described above. In the example, Obama would choose to withdraw in order to defeat Romney and elect Clinton. The Democrats and Republicans would presumably decide to stop spending money on primary elections and would nominate more than one candidate per office. Candidates who want to win would recognize that the way to win is to adopt moderate positions on the issues to try to appear to be the best compromise. With many candidates competing to be the best compromise, voters would be free to rank the less corrupt of them over the more corrupt.
I'm not a fan of plain IRV. I much prefer Condorcetian voting methods. However, I cheerfully acknowledge that if IRV is patched to allow withdrawal then it would be nearly as good as Condorcetian methods.
Condorcetian methods that allow withdrawal would be even better. It would then be a waste of effort to organize voting strategies hoping to manufacture a majority cycle that elects the strategizing voters' preferred candidate, since the "patsy" candidate--the candidate strategically raised over the sincere winner--would have a strong incentive to withdraw. I believe strategic voting would be rare, given the option to withdraw and assuming a good underlying voting method. I believe candidate withdrawals would be rare if the underlying voting method elects within the top cycle and is independent of clones. (For an example of such a voting method, google Maximize Affirmed Majorities.) --Steve Eppley
—Preceding unsigned comment added by 171.159.64.10 (talk) 02:56, 18 December 2007 (UTC)
[edit] Stupid question
I always knew this as vote splitting. How is it different? (add an explanation if it is, add to the merger if not) 142.177.168.90 14:15, 18 Jun 2004 (UTC)
Yes, this should be merged with vote splitting. However, it should not be merged with Independence of Irrelevant Alternatives - the two are related but different subjects.
- I would disagree that they should be merged.
- First, I believe it would be accurate to say that Vote Splitting is related to the Independence of Clones Criterion (http://condorcet.org/emr/criteria.shtml).
- The spoiler effect can refer to either independence of irrelevant alternatives and Vote Splitting and I have seen the terminology used in both contexts.
- In the case of independence of irrelevant alternatives, one can think of candidate B as a spoiler for candidate A if, when candidate B was removed, candidate A wins the election.
- The situation is different when it comes to clones/vote splitting. If candidate B & A split their votes, it does not imply that either candidate would win the election.
- Ericgorr 23:09, 10 Nov 2004 (UTC)
-
- In that case, should the spoiler effect article be a disambiguation page? Tim Ivorson 13:35, 11 Nov 2004 (UTC)
[edit] Discussion of Mathematical Definition
I think that the "spoiler effect" is better understood as sensitivity to irrelevant alternatives rather than sensitivity to clones in the form of vote splitting. Given that, Condorcet methods reduce the spoiler effect by only being sensitive to IA's when there is no CW, and approval arguably eliminates the spoiler effect under certain assumptions. --Hermitage 06:33, 14 July 2005 (UTC)
- Hmm, it seems to me that violations of IIA can be broken into two distinct scenarios: the "spoiler effect" and the "center squeeze" effect. As the article states, the "spoiler effect" is typically reserved for irrelevant alternatives with weak core support (few number of first choices), as in the Nader scenario. The "center squeeze" occurs when the irrelevant alternative has strong core support (large number of first choices). That's why IRV is typically said to eliminate the "spoiler effect" but not the "center squeeze". Would it be better to say that the "spoiler effect" corresponds to violations of the mutual majority criterion? That is, a "spoiler" is an irrelevant alternative who is partitioned into the "mutual minority" by a majority of voters? The difficulty here is that "spoiler effect" is quite a colloquial term, a possible "neologism" as suggested below, and tough to assign a mathematical definition to. --Progressnerd (talk) 19:15, 11 April 2008 (UTC)
- Another thought. IIA says that the addition of a candidate that does not win should not be able to change the winner, but that's not exactly what "spoiler" means colloquially. "Spoiler" refers to someone who does not have a chance of winning. Clearly, in the "center squeeze" scenario of IRV, the two candidates with the most first choices are better positioned than anyone to win. The addition of one of those candidates may change the outcome even though that candidate loses, but I don't think that candidate can be considered a "spoiler", because s/he had a legitimate chance at winning. Moreover, it would be silly not to consider Condorcet as eliminating the "spoiler effect" even though it technically does not satisfy IIA. Given that, which irrelevant alternatives are "spoilers" and which are not? --Progressnerd (talk) 15:50, 12 April 2008 (UTC)
[edit] Removed redirect from "Split Vote"
Just letting y'all know: I removed the redirect from "Split Vote", since that and "vote splitting" aren't the same thing, though they might sound similar. --DewiMorgan 04:29, 18 July 2007 (UTC)
[edit] Spoiler (sports)
Spoiler (sports) redirects here. There is nothing said about sports. 141.151.79.241 22:56, 30 September 2007 (UTC)
[edit] Worldwide View
Somebody added the note "does not represent a worldwide view". I disagree. The Bush/Gore example is from the U.S., but there's also a South Korean example. I would vote to remove the note. Jrvz (talk) 01:52, 30 November 2007 (UTC)
[edit] Is the article title appropriate?
Where does the term "Spoiler effect" originate from? Is it verifiable? It seems like it might be a "neologism"
Also, as I understand it, "vote splitting" is the mechanism that causes the "Spoiler effect". So "spoiler effect" is a subset of "vote splitting". So shouldn't "Spoiler effect" redirect to "vote splitting"? The way it is now, "vote splitting" redirects to "Spoiler effect"; however, a minor amount of vote splitting will not be sufficient to cause a "Spoiler effect". So this is confusing for the redirected readers.
- The term "Spoiler" almost universally refers to a weak or minor candidate in a plurality election who gets relatively few votes but none the less has a "detrimental" or "spoiling" effect on the election outcome. "vote-splitting" is a broader concept that can also involve viable candidates who are similar, and may each receive a substantial percentage of the vote, but allow a dissimilar candidate to win with a mere plurality. The term "spoiler effect" is inappropriate and is not commonly used when there is not an obvious weak "spoiler" candidate. I think the connection between the two concepts should be discussed, but also it should be clear that vote-splitting can happen where there is no "spoiler effect."
- Tbouricius (talk) 18:28, 11 April 2008 (UTC)
Mikiemike (talk) 09:09, 1 February 2008 (UTC)--
I agree. Have spoiler effect redirect to vote splitting, rather than the other way around. --Helenalex (talk) 03:41, 13 May 2008 (UTC)
- I too agree there should be an article titled "vote splitting". Yet I am in favor of keeping this "spoiler effect" article too.
- Already there is a redirect for the hyphenated words "vote-splitting", and that should remain, and point to "vote splitting". For grammatical correctness the article title of "vote splitting" should not have a hyphen. (Clarification: If the title were "vote-splitting effect", then a hyphen would be needed to indicate that the two nouns are being used together as an adjective.)
- As I recall the information about vote splitting originally was added to an article named "strategic nomination" which is yet another related concept that also deserves a separate article. Instead of being bounced around as if it were subordinate to voting concepts that would not happen without vote splitting, "vote splitting" needs an article of its own.
- When it is added, be sure to change links that now point to the "vote splitting" section of this article, and links that point to the hyphenated "vote-splitting" stub. VoteFair (talk) 22:08, 14 May 2008 (UTC)
-
-
- Although I can't think of a case where vote splitting is not one of the contributing causes of the spoiler effect, there can be additional contributing causes to the spoiler effect. For instance strategic nomination may also be involved (especially in U.S. primary elections).
-
I've created a new page for Vote splitting, but it needs work from someone who knows more about political theory than I do. --Helenalex (talk) 08:19, 3 June 2008 (UTC)