Consecutive Runoff Approval Voting

From electowiki

Perhaps some adjustment of the consecutive runoff approval voting method described below would be appropriate. This writer has been running tests using that method on political websites, and it has be found that many people holding the currently dominant perspectives are not prepared to accept a method that requires three consecutive runoffs. It is also challenging to construct an experimental straw poll that does not instigate arguments (or worse). What was found, however, is that more people are amenable to a method that only requires two consecutive runoffs.

It seems appropriate to make certain points that were brought up in the discussion of the three runoff method below. Consecutive runoff approval voting was designed to comply with three dominant principles: removal of the black hat syndrome (also called the spoiler syndrome), the ability to count a ballots at the voting station level in such a way as to produce simple numerical sums that can then be added into larger tabulations, thus avoiding any need to send any complex information to the larger tabulations, and achieving a method that provides extreme overall simplicity. It does seem to comply with these principles. The third principle that two-consecutive runoff voting is required to comply with is that it must harbor no black hat (spoiler) syndrome. This method always results in an absolute majority winner, so by some definitions, it may not constitute a "winner take all" method. The structure of the two runoff method is as follows.


In the first runoff, the approval method is used to choose exactly two candidates to run in the second runoff. The first runoff is simply an approval election; each voter can give just one vote to as many candidates she or he "approves of," or finds acceptable. There is one practical consideration, however, due to the potential problem of voters casting an inordinate number of votes. So it would seem advantageous to limit the total number of votes that each voter can cast to some arbitrary number. A limit of 20 is suggested, but many people holding currently dominant perspectives seem to prefer a limit of 10. At the end of the polling, all of the votes are simply added up, and only the two candidates with the most votes go on to the second runoff.

The second runoff is quite simple; the two remaining candidates run against each other, and the one who achieves an absolute majority becomes the winner. Obviously, there can be no black hat (spoiler) syndrome, since there can be no third candidate.

The design of this method was undertaken with the express assumption that some interested party will often, if not always, attempt to use the black hat (spoiler) syndrome to circumvent the will of the voters. Therefor, we are not merely attempting to attain a method that precludes "voter strategies," but to gain a method that precludes the black hat (spoiler) syndrome from being exploited by dominant special interests. This is a somewhat subtle consideration, which only becomes more tricky if it is assumed that special interests will actively attempt to use this syndrome to manipulate elections.

It seems reasonable to assume that any ranked voting method will be susceptible to black hat manipulation, since it appears obvious that if special interests vigorously promote black hat candidates, voters will effectively be forced to cast their highest rank vote for whatever, generally minimally, acceptable candidate is perceived as likely to defeat the black hat candidates. In fact, this writer has engaged in countless discussions, in which it appeared that, with only four, at most, in a race, most (if not all) reasonably simple methods of ranked voting could enable a vote for a white hat candidate, from the perspective of some individual voter, to eventually lead to the actual election of a candidate that that individual voter perceives as a black hat. And this is a result that would have been avoided if the voter had given a gray hat candidate their highest rank vote. But at minimum, it seems reasonable to assume that a voter would tend to cast his or her first rank ballot for a likely-to-win gray hat candidate, rather than a "long shot" candidate, if a strong black hat is in the race. This would eventuate in the evolution of a closed two-party system.

Two-consecutive runoff voting cannot cause the situation in which a voter causing the election of a black hat by giving a high-rank vote a white hat, or even a case in which a voter perceives a need to give a high-rank vote to a gray hat in order to avoid the election of a black hat. And the obvious reason is that there are no ranks involved in this method. However a "gray hat" syndrome is present, in that, if a black hat is in the race, voters may feel some pressure to include some gray hats of "darker shades" in the first runoff if a black hat is in the race. However, this gray hat syndrome is vastly more benign than the black hat syndrome; for example, a voter could still vote for as many white hats as he or she desired.

It seems likely that the gray hat syndrome would be further ameliorated if a three runoff method is employed. On the other hand, this method requires three consecutive runoffs, which present-day online straw poll voters seem to dislike.


Observe below that I use hardly any mathematical methods to describe Consecutive Runoff Approval Voting. This is probably because I work largely in natural language study, and have decided that whenever the potential number of criteria appears to exceed the number of methods they might be applied to, mathematical analyses tend to be less than fruitful. I do have one very prominent criterion, which is that any proposed system of voting should be completely free of the Black Hat Syndrome (or "spoiler effect"). I have concluded that every method is susceptible to a Gray Hat Syndrome, in which the presence of a Black Hat "ogre" candidate could cause the election of a Gray Hat, where a White hat would otherwise have been elected.

Consecutive Runoff Approval Voting --- The Basics:

The number and viability of subtle, yet pervasive, ways for "election hackers" to create security holes in all forms of "IRV" stunned me. And I began the search for a reasonably simple, stable version of "IRV" that could withstand brute-force assaults. I found none. I also found that "IRV" was advocacy was being generously supported by The Ford Foundation, Rockefeller Brothers Fund, and Carnegie Mellon Foundation. (It is now advocated by New America's Leadership Council, whose Board of Directors seems to include the Executive Vice President, Wal-Mart.) So, I reluctantly concluded that "IRV" would doom all other viable efforts to institute any spoiler effect neutralizing election methods. It was depressing. And probably hundreds of Green Party, or semi-Green advocates had by this time been 100% sold on the amazingly attractive, yet highly problematic, "IRV" methods.

So, below, I repost what I see as one of the best explanations of a method that can actually completely overcome the baleful consequences of the spoiler effect, which I call the Black Hat Syndrome these days. Here it is:

There are so many reasons why, in this age of coordinated electronic propaganda, democracy just doesn't work. First of all, as I have said so often, it is totally foolish to believe that people who are one paycheck away from starvation would be permitted to wield the real power of the vote when their electronic media environment is owned by a few multi-billionaires. Ask yourself "Would that be possible?" Or try this more difficult one: "Can pigs fly?"

Our plurality voting system extracts 99% of the decision making power of the voters (very roughly speaking). If you would like to vote for a White Hat, say Nader, but there is a Gray Hat, say Kerry, and a Black Hat, say Bush, in the contest, you would be suicidal to vote for the White Hat. So you are screwed.

As I have pointed out before, if you include all of the proposed metods of counting, there are potentially many different methods for counting the ranked ballots of IRV, and the best of them require that specific information about every ballot bust be conveyed into one location, one computational funnel. And the the computation of the vote is exceedingly intricate for the best methods of counting. Moreover, you will still need to give your first rank vote to a Gray Hat if a Black hat is present in a contest.

All these woes disappear if you just have three distinct runoffs for each election. The first runoff is an approval election that narrows the field to, say, eight candidates. With the approval method, each voter can cast just one ballot for each potential candidate that she or he approves of. There would need to be some practical limit on the number of people each voter could give one vote to, maybe 20, since we can't deal with lists bearing thousands of names. The second runoff, in which each voter could give just one vote to as many of the remaining eight as she or he wishes, would narrow the field to just the two candidates who get the most votes. The third and final runoff would be simply a race between those two candidates. I would suggest that, for each runoff, the voting could be allowed to run over a span of three days. The ballots would be hand-marked, and hand-counted by randomly drafted citizens, with each day's results announces at each polling station on the morning after that day's voting. I would give the citizens a week to deliberate between each runoff, so there would be three days for the first runoff, plus a week, the same time for the second runoff, then three days to complete the final runoff. Such a procedure of consecutive runoff approval voting would ensure that the citizens would feel that they would be making real and serious decisions.

I have claimed that IRV-style voting methods would not provide any substantial improvement over "our" current plurality method. All questions about voting methods tend to be tricky. One of the most disturbing things about IRV-style voting is that it would "cloak" the security holes of plurality voting; another is that, despite all the claims otherwise, even the most moderately secure forms of IRV require massive computer intervention, and they still retain a spoiler effect (Black Hat Effect) of their own, since in every system that offers a highest-rank choice to the voter, the voter is ultimately forced (out of self preservation) the give that highest-rank vote to a Gray Hat when Black Hats are present. It does absolutely no good to surmise that such scenarios are unlikely, since reality dictates that powerful anti-populist power brokers can always use money power to maneuver to make such scenarios inevitable.

And, when it comes to voting: SIMPLE IS BEAUTIFUL!

Everything comes at a price, and voting method selection is the very last place we should expect to get something for nothing. One difficulty with Approval voting is reflected in the fact that it really should be called "Acceptability voting"; You would pretty well thwart the elite (ish) anti-populist power brokers, but would still occasionally struggle with an irksome conundrum. For example, suppose three fourths want, say, Ralph Nader very, very, much but all four-fifths of us still find John Kerry slightly acceptable. Even though no one is very excited at all about Kerry's campaign, while three fourths totally love Nader's program, Kerry will win. This is a bad, though not ruinous, outcome. But it will not encourage participation. So, I advocate consecutive runoff approval voting. That is:

A first Approval runoff that narrows the field of candidates to eight:

A week later, another Approval runoff that narrows the field of candidates to two.

A week later, another final Approval runoff that selects the office holder-elect. (Technically, that would be nearly equivalent to the present Plurality method, except that no "third candidate" could be on the ballot to "spoil" the voting).

Of course, these advantages will not come "for free." Three consecutive runoffs would be required. However, as any old organizer knows perfectly well, the more you get people to participate or struggle, the more engaged they will always become, and the more they will see themselves as holding a stake in that process and its outcome. As I have reiterated again and again, voting is like military service, and people are willing to vote for the same reasons they are willing to fight. If you can get people to spend years in hard military battle, then you can get them to go out and vote three times.

»»» Note: The "Winner Take All Effect" has nothing to do with the Black Hat Syndrome (or "spoiler effect"). The example below demonstrates this reality:

The following sentence typifies a notion that tends to be constantly reiterated in the political blogs:

The winner take all system naturally reduces to 2 dominant parties and the Republicans were one of the two parties that survived.

This is really at the heart of the central dilemma of US politics. If you hope to have any influence on the ideological structure of US politics, you must capture one of two dominant parties, because the voting structure imposes a two party system. What distresses me more than anything is that virtually every political blogger around insists on courting a totally incorrect theory about why this is the case! This contention is that: "The winner take all system naturally reduces to 2 dominant parties and the Republicans were one of the two parties that survived" -- and this is just flat-out WRONG!!!

Careless thinking might lead us to see a winner take all syndrome as the cause of our inflexible two-party pseudo-democracy. BUT THAT IS TOTALLY WRONG. The real cause of the two-party pseudo-democracy is really the direct consequence of THE BLACK HAT SYNDROME! This Black Hat Syndrome (or "spoiler effect") is the outcome that results when we have a White Hat (relative to each individual voter -- say, for example, Ralph Nader), a Gray Hat (say, for example, John Kerry), and a Black Hat (say, for example, George W. Bush) in a political contest. You cannot vote for the White Hat without "sacrificing" the vote you would have otherwise used to fend off an election of the Black Hat. So you will just never get to vote for a White Hat, and thus anything like a third party is out of the question.

I will now provide an example of an implementation of a realistic solution (which definitely is not IRV -- unlike IRV it does not demand that all information from every ballot must be gathered into one central counting location, and it only requires simple addition -- it is intrinsically extremely simple). It requires three consecutive runoffs, but if we can ask people to fight and die in Iraq for years, we can surely ask them to vote three times. Besides, it provides a deliberative process, and an opportunity for participation, instead of a mere ritual. (I indulge in a liberal perspective, but these arguments would also hold if viewed from a conservative perspective.)

Consecutive Approval Voting ---

Round one of a Consecutive Approval Voting election is an approval, not a plurality method of election. Therefor, each voter gets to give just one vote to each candidate that she or he "approves of" (finds acceptable) up to twenty choices (so wacky people don't list thousands of candidates out of a phone book). From start to finish, parties are only advocacy networks; this voting system is "blind" to parties. So there can be no negotiating. Now, for example, Intelligent Greens will vote for some Democrats, as well as some greens. And intelligent Democrats will vote for some Greens, as well as some Democrats. So some Democrats and/or some greens will undoubtedly get a very high percentage of the maximum possible vote. Given a modicum of intelligence on the part of the voters, some Republicans would possibly get up to 40% of the maximum possible vote, supposing that Democrats could muster, say, 35%, and Greens held, say, 25% of the maximum possible vote. Now, the eight candidates who garner the most votes get to go to a second round.

The second round is, again, an approval contest between the eight remaining contenders. No negotiation between parties is allowed. Each voter can give exactly one vote to each of the eight remaining contenders that she or he "approves of" (finds acceptable). Once again the votes are added up, and the two candidates who have received the most votes go to the final round.

The final round is between only the two remaining contenders, and there will be no third candidate to act as a "spoiler," So, the Black Hat, or "spoiler" syndrome is entirely eliminated.