MeatballWiki | RecentChanges | Random Page | Indices | Categories

Voting is one of the best ways to quantify opinions in a large group. Online communities that have a common goal will be continually in need of making decisions. In a small group of similar-minded individuals, a consensus decision can often be found by discussion. In such a group, VotingIsEvil.

However, as the group grows larger, CommunityMayNotScale. For each member of the group, there's a certain likelihood that he disagrees with one of the existing members. The likelihood of conflict between two individuals grows geometrically, according to MetcalfesLaw. The likelihood of conflict between two sub-groups grows exponentially, according to ReedsLaw. If individuals with contrary opinions are CommunityExiled, the group may succumb to GroupThink. If not excluded, they may delay consensus decisions indefinitely. Therefore, VotingIsGood.

In a situation where not everyone agrees, voting allows us to quantify and measure the opinions of the group as a whole. Such measurements must always be imprecise to some degree, although the poll options can be developed in consensus through wikis etc. In a community where voting on issues is accepted policy, votes help curb social problems within the group, as voting is an accepted way of ConflictResolution. "If the vote is against me, I will respect it." The only members that need to be CommunityExiled are those that do not respect the voting process, likely a small minority and probably too extremist and contrarian to be of much help. Quite possibly those who would most like to be GodKing. (Similarly, you would not want people in your project that continually tell you that reality is a construction of their mind and that the other group members don't exist.)

The fact that real world voting systems are broken in a social as well as often a technical sense should not be used to discredit voting as a whole. Democracy is "broken as designed", corrupted by media concentration, direct lobbyism and certain reflected assumptions about humanity and the need of control by a chosen few. Voting, when implemented anew in software, can work around these problems easily.

Voting can be used to gather input from outside the group as well. The open source developers behind KDE, for example, use regular polls on [KDE-Look] to make decisions about the look and feel of the desktops. How else would you determine which default "theme" to use for your program?

Besides using advanced voting mathematics like preferential voting, it is also necessary to devote much time to the actual user interface. If you want to allow votes from people who are not members of the group, and you want to avoid them making uninformed decisions, you have to present them with a summary of all arguments on the matter in question. Such summaries can obviously be worked on easily using a wiki before a poll is SlashDotted.

In conclusion, voting is a great way to overcome stagnation and to avoid dictatorship. As our society becomes more networked, it is increasingly necessary to measure the opinions of groups, or the network will either become strictly hierarchical or consist of small, isolated, dogmatic groups. Neither alternative is desired. VotingIsGood. Not allowing people to vote is evil.

-- ErikMoeller, AnonymousDonor

cf. VotingIsEvil.

This page makes a giant equivocation. Voting is not polling. Rather, polling is the strategy we use to hold a vote. Hence the phrase, "Go to the polls". Vote means will. When you have a set of stakeholders sharing power, time comes when you need a declaration of how each intends to wield his or her power. We use polling in a democracy because that is the only technique to quantify the will of the people, since in a democracy each (enfranchised) citizen is given power over the State. But it's not necessary to use polls. You could just ask people what they intend to do.

The way to avoid dictatorship is to DevolvePower. If you DevolvePower to your CommunityMembers, then you do in fact need to poll them to make decisions. The presumption behind a democracy is that the community members will kill each other (turns out to be true) without a voting system to give them an alternate avenue of gaining and negotiating control over the State. We all know states that hold sham elections. Voting itself is only the last step in the construction of a democratic state.

This page is more about polling, but polling is only one tool to gain enough information to make a decision. -- SunirShah


Mr. Moeller writes, Voting is one of the best ways to quantify opinions in a group. There are three problems with this claim.

1. Voting is one of the simplest ways. It is only one of the best if simplest=best. See VotingIsEvil for more discussion.
2. Polling is the best way to quantify opinions in a group. Polling is voting, with at least a modicum of statistics thrown in to achieve some degree of accuracy of the general public's actual opinion. Voting measures a much more complicated admixture of opinion, personality, opportunity, desire, etc.

For 1 and 2, I said "one of the best", and in that context, that means that it's a good balance of simplicity, ease of use, speed, and accuracy, all of which are important in project management. Sophisticated polling has advantages in one area (accuracy) that are outweighed in other areas (complexity, speed etc.).

3. Quantifying opinions is rarely what is desired. Quantifying need is what is important.

Quantifying opinions is often a good way to quantify needs. "Would you use feature X?" Without quantifying what people want through voting or polling, you have to rely on mere guesswork.

Not at all--you can observe behavior--or is that what you mean by mere guesswork?

Fortunately, for online communities, we have a better way of quantifying: measure what people actually do (vote-with-your-feet) rather than what they say. If SunirShah wants to know what random users find of interest in his site, he doesn't have to poll them: he can just look at what pages they view, which pages are the most linked to, which pages are the most edited, etc. That's real quantification.

That is a very naive view of communities and hyperlinked information. Many links are not followed not because their targets are uninteresting, but because their placement is ill chosen or their text is not sufficiently descriptive. In a wiki, hopefully, these structures will gradually evolve, so the wiki process is often better at obtaining a good model than doing it by voting -- that is, voting with your hands. But the "feet" alone, that is, the page statistics, are not of very much value because they are much less accurate than discrete options in a poll. That's why many websites nowadays have polls like "Are you aware that you can also do X here?" or "Which of A, B, C, D do you care the most about?" (see the current InfoAnarchy poll).

You assert that polls are more accurate, but that's as naive a view as my assertion. (That is to say, properly implemented, either way has value; improperly, they are without value.)

Voting is of primary interest for the following aspects of wikis: 1) code and static content -- which features do we want, and how do we want to display them? 2) otherwise unresolvable edit conflicts -- people disagree about the placement of a certain word in the article, etc. 3) policy -- which units, formats, patterns etc. should be used? 4) planning -- what is our long term strategy?

As most of these are complex issues, voting is a suboptimal way of dealing with them. And I have yet to see an unresolvable edit conflict.


The flipside argument works as well; as the community grows larger, the size of the consensus will grow. If one defines "consensus" as "absolute, total, and enthusiastic agreement", then yes, the increased likelihood of conflict makes problems inevitable. But if "consensus" is defined as "collaborative decision-making in which all ideas and concerns are incorporated", then problems are not inevitable. In other words, conflict-management mechanisms need to scale at the same rate as conflict increases. You implicitly assert they can't; I assert they can.

It should be worded, "As the group grow larger, however, consensus on the average will take longer." This is a supportable claim. If time is of the essence, that could be a problem. However, at least for non-commercial online communities, time is rarely of the essence.

Excluding the contrarians

If you accept this premise, "The only members that need to be excluded from the group are those that do not respect the voting process, likely a small minority and probably too extremist and contrarian to be of much help," then sure, VotingIsGood.

What more is there to say?

Arrow's paradox

Moved to ArrowsTheorem

On design decisions

This is somewhat of a tangent, but it's a classic example of where majority vote (design-by-committee) does not best serve the interests of the community.

Mr. Moeller writes: "How else would you determine which default "theme" to use for your program?"

Design decisions are generally best made not by mass voting but by interested professionals who understand interface design.

That's your belief or assertion, I do not agree with it. The purpose of interface design is to find the most efficient and beautiful solution to a given problem. To do so, you expose a sample, as you correctly point out, to different designs. There's no problem in using voting to narrow down the initial choices, and there's no problem in using voting to make a decision about the final choice. As I pointed out, however, you need to explain to people the differences accurately so that they can make reasonably informed decisions. If you just throw a few screenshots at them, then, yes, you will likely not get a good decision.

As for extrapolating from small samples, that's one of the biggest problems of design: Getting a "representative" sample is very hard. Cultures as well as individuals vary greatly in their reaction to particular patterns. Even the most powerful design will always have a number of fanatical opponents. The more data you get, the better.

True, but voting reduces your data set much more than studying the behavior of a reasonably representative sample.

Whether you use voting or more accurate polling methods is obviously not a question of ideology, it's a question of cost. For open source projects, basic voting systems will likely be the most affordable.

Lastly, from the designs of both MacOS and Windows it's quite obvious that the design process these companies use is utterly flawed. Their results are despicable, such as graphical wheels for volume control in video players (Apple QuickTime). They have not taken the time to properly test these interfaces and to poll their users not only on their beauty, but also on their efficiency.

Of course. Those are straw men: they are good examples of bad design. I would never assert that Microsoft is a bastion of design principles, nor would I ever assert that everything Apple is perfect. But the truly well-designed elements, such as Engelbart's system (which was not perfect, but so absurdly innovative one can forgive a few flaws), or the original Macintosh (MacPaint?!) are certainly not something that would have come out of voting.

All the advances in computer interface design, from Engelbart through the iMac, were not made based on voting, but based on people who

  1. actually studied the needs of different kinds of people (and you need only a relatively small representative sample to properly extrapolate to the general public)
  2. then used expert knowledge to devise innovative solutions to the needs.

Trust the experts! That's what your opinion boils down to. Of course, the experts themselves start to disagree once the group gets sufficiently large, and will have to resort to voting once again, or to the good old GodKing .. Furthermore, if the experts use polling of small samples to arrive at decisions between different options, they use nothing but a non-binding voting process.

I wasn't talking about polling; I was talking about studying behavior. You're right: I am saying "Trust the experts". But I'm saying "Trust the experts to be experts". Their actions should be open and transparent. But, for example, I'd rather trust the programmers to write code that works than to generate it through a voting process by non-experts.

In terms of design decisions, most people are bad at devising answers. They are, however, good at knowing whether or not a system needs fixing. It requires a certain mindset to translate the stated "I want this" to the actual "My need is this". It is the need that should be solved, not the want that should be served. -- AnonymousDonor

To some extent, that's true, and that's why voting presents a given set of options. These options can much easier be devised in consensus of concerned and interested parties than an actual decision can be made in consensus. The purpose of voting is not usually to devise answers, but to select and validate them. --ErikMoeller

The validation is only necessary if you have a group that already believes VotingIsGood. In a group that believes VotingIsEvil, voting doesn't validate decisions.

Sure, but have you ever considered just talking to people? -- SunirShah

I recently read an interesting [article in The Economist] about why abortion is such a big divisive issue in America and not so much in other places. They say that the main reason is that America legalised abortion through a Supreme Court ruling (a small council of experts) whereas other countries legalised it through legislation or referenda:

"Why does abortion remain so much more controversial in America than in the other countries that have legalised it? The fundamental reason is the way the Americans went about legalisation. European countries did so through legislation and, occasionally, referenda. This allowed abortion opponents to vent their objections and legislators to adjust the rules to local tastes. Above all, it gave legalisation the legitimacy of majority support.

... [as for America:]

It would be hard to design a way of legalising abortion that could be better calculated to stir up controversy. Abortion opponents were furious about being denied their say. Abortion supporters had to rely on the precarious balance of power on the Supreme Court. Legalisation did not have the legitimacy of majority support."

Not so much an argument for voting vs. consensus, but just an interesting example of voting being good. (perhaps this should go on the page FairProcess?)

-- BayleShanks

Actually, that is the perfect example of why VotingIsEvil. Left to themselves, the majority of Americans would vote against abortion. Voting always disadvantages the minority, which is why we have things like the U.S. Constitution. The Supreme Court's responsibility is to uphold the values of the Constitution no matter what the populace currently believes, as the Constitution is meant to represent beliefs more important than merely the current day's opinion. Now, while you may believe the Constitution is broken, that's another story, but for all intents at purposes, the population was protected from going back on their earlier promise to themselves. -- SunirShah

I'm not sure I would count abortion among the American people's earlier promises to themselves -- I bet that at the time that the Constitution was written, legal abortion wouldn't have much of a chance (if they could have known what we did about fetuses at that time, at least). Open/legal/widespead abortion was not a previous "right" that was being threatened by popular will; it was a liberal/secular development that was new (in that way, it was similar to the women's sufferage movement).

Anyhow, tyranny of the majority isn't a very strong argument here, because both pro- and anti- abortionists' arguments are fundamentally concerns about a tyranny of the majority. The pro-abortion people think that abortion means that the minority of fetuses are being murdered (because they are too young to fight for themselves), and the anti-abortion people think that no abortion means that the minority of women who want an abortion are being unjustly interfered with by others. This is not the sort of case that can be solved by a straightforward application of principals like "tyranny of the majority", because the debate is fundamentally about one of the axioms of such a system (the "axiom" which specifies which beings are members of society).

If things had turned out differently, the Supreme Court could have ruled that fetuses are humans and therefore abortion must be illegal. And then you could still try to argue here that it's a good thing they decided it rather than a popular vote, because they protected the disadvantaged minority of fetuses from others. So if your argument works equally as well no matter what the evidence was, I don't think that the evidence can really be said to support the argument. -- BayleShanks

I don't want to get into a constitutional argument over abortion, especially since my understanding of these issues is shallow, but let me point out one thing. The U.S. Supreme Court only has jurisdiction over matters pertaining to the U.S. Constitution. In this case, I presume the problem more or less comes down to the lack of a definition for person in the U.S. Constitution, and the commensurate lack of discussion in the Founding Papers. Thus, I suppose in this case, they ruled in favour of the rights of the only person, which is the mother. Like I said, whether you believe the Constitution is broken is something altogether different from the purpose of the U.S. Supreme Court, which is to maintain the philosophical principles of the nation.

Don't forget that the United States has had to ammend the Constitution to clarify what constitutes a person, and has gone to war over the issue. And we shouldn't forget that the U.S. Supreme Court had also failed to protect the Constitution during WWII, allowing for the construction of race-based concentration camps, only recanting their actions as myopic after the war ended. But these failures and patches are tiny compared to the greater idea of a constitutional democracy.

If I was slightly more cynical about the membership of this site, I'd remind you that many here spend an inordinate amount of time tracking the Constitutional challenge to the DMCA based on your right to FreeSpeech, even though the DMCA was duly voted in by the majority of representatives. Since I am cynical, I can say that as I was apparently born a terrorist, I am glad that I have Charter protection in these uncertain times after the World Trade Center attacks, even though this country has also had difficulty in the past with its own constitution. -- SunirShah

Hey, I'm totally in favor of a Constitution; I don't really consider that an argument against voting, I think of as an add-on feature to a system of voting; in order to introduce extra stability, you make a subset of the laws "really important", including all meta-laws, and you can't change those without a super-majority.

It's true that the U.S. Supreme Court is something other than voting; I would prefer that individual court cases ultimately be decided by some sort of system other than direct vote by all Americans (although I think that in a LiquidDemocracy voting system, I would prefer to put most cases to the vote; maybe there could still be an appointed Supreme Court at the highest level, though, for the purpose of ChecksAndBalances?).

But, I do think the current Supreme Court system is "broken" in that the Court often ends up making substantive policy changes which go beyond a direct interpretation of the law. I think that when a fundamental underspecified "hole" in the law is uncovered, such as "Is a fetus a person?", the court should decide the case at hand, but ultimately the legislature should be encouraged to decide the issue for the long term. I.e. perhaps the Court should do something like declare that the law is underspecified and that therefore their decision in that case has the force of normal law but not the force of the Constitution; then the legislature could make a policy on the issue with a normal majority vote rather than a super-majority Consitutional amendment. This way any new addition to the Constitution would represent "the will of the people" (or at least, something that "the people" don't care enough about to change), rather than some additions being the decision of whoever happens to be in the Supreme Court at the time.

Another problem with the current system (and here I'm diverging a bit from the topic at hand) is that it is often necessary to modernize the Constitution, and the way this ends up happening in the U.S. is that the Constitution itself is not amended, but the Supreme Court reinterprets it in some ridiculously unliteral way. I think that's a fundamental problem because it introduces vast complexity into government at the most basic level (a normal unaided citizen can't understand what their Constitution really is today by reading the Constitution because there are these secret effective amendments in the Supreme Court cases).

I'm not sure how to solve that; my current proposal is to give the Court the power to force the legislature to vote on a given Constitutional amendment, so that when the Court determines that the Constitution needs some specific modernization, they have a mechanism of bringing that to the forefront of attention (and, they would have an another option for introducing Constitutional change besides reinterpretation) (the legislature would still need a super-majority to change the Constitution, but at least the issue would get the needed attention).

So, both of my proposed modifications would allow the Court more freedom to propose changes to the Constitution without changing things unilaterally.

Btw, one small gripe with wording: I don't think the purpose of the Court is to maintain the principals of the nation; I think that responsibility cannot be delegated, i.e. the Public ultimately has to do this themselves (although things like the Constitution can introduce a 'lag' or a 'low-pass filter' to screen out short-lived temporary shifts in opinion). I think their purpose is (1) to decide actual cases, (2) provide a BalancingForce.

-- BayleShanks

The legislature has the ability to overrule the Supreme Court on any issue just by passing an Amendment. Consider the Court's decision that slaves were property (as they were not persons). Fortunately, the legislature still considers amending the Constitution to be a difficult and serious decision. I'd rather not see a system where the legislature has to continuously decide how to reinterpret the Constitution because they will eventually become lax.

The argument that the Court creates laws has been made many times before, but that's because the judicial and legislative branches have entered into agreement that this is the proper course of action. The current process for creating a law is seriously and truly to "throw things at the Court and see what will stick." This is hideously expensive and painful for the victims of illegal legislation, but it makes the political choices much easier. I know this sounds cynical, but it's true. If you want an explicit example, abortion in the States is one, and homosexual rights in Canada is another. -- SunirShah

Yes, I agree with you, if the legislature got its act together it could rationalize the system by doing what it's supposed to (in this case, amending the Constitution when necessary). In fact, I tend to think that if the U.S. Supreme Court has its druthers, the legislature would take a much more active role in the Constitutional modernization. That is, I think the legislature is dropping the ball, and the Court is picking up the slack. I was just trying to think of some structural modifications that would encourage the legislature to deal with Constitutional issues.

I agree that I want the legislature to consider amending the Constitution a serious thing, and in fact I think there's a lot of value to everyone in America considering the Constitution as this ancient, holy relic which almost cannot be questioned (is the Canadian one like that? It's quite a bit longer than ours, so I would think it would be less worshiped). But, when the Constitution must be altered, it's important to do it simply, through law, rather than complicatedly, through legal precedent. So I'll put my faith in the super-majority as being a sufficient conservative force, and I would like to see the legislature openly encouraged to change the Constitution more, even semi-regularly (a few times a decade).

-- BayleShanks

The writers of the Constitution didn't assume the legislature would be responsible enough to "consider amending the Constitution a serious thing". That's why Congress can't pass amendments to the Constitution. The states have that power. All the Congress can do is to basically propose amendments.

"The Congress, whenever two thirds of both houses shall deem it necessary, shall propose amendments to this Constitution, or, on the application of the legislatures of two thirds of the several states, shall call a convention for proposing amendments, which, in either case, shall be valid to all intents and purposes, as part of this Constitution, when ratified by the legislatures of three fourths of the several states, or by conventions in three fourths thereof...."



MeatballWiki | RecentChanges | Random Page | Indices | Categories
Edit text of this page | View other revisions