User:SteveBaker/Consensus Essay

I'm not fond of polls - achieving consensus is obviously better - but, I will argue, consensus has it's limits and there comes a time when binding voting becomes the only fair way to proceed.

What is consensus - exactly?
From WP:Consensus: Consensus does not mean that everyone agrees with the outcome; instead, it means that everyone agrees to abide by the outcome.

So in order to achieve consensus when one does not have unanimity, one has to discuss the concerns of the people on either side of the debate and work to flip their opinion or gain their acceptance of the opposite opinion. This process must continue until (per the policy) there is absolute unanimity of acceptance. The policy has something of a 'get out' clause if the hold-outs are refusing to form consensus through bad faith. But since we can never assume bad faith - that doesn't help much. Flipping the minds of entrenched opposition is hard. It may mean using the force of cold, hard logic, or it may require empassioned oration - or you may have to give ground and make compromises.

We are also told that there are no time limits on consensus. There is never a time when a decision is made. At any point when there is consensus, you have it - as soon as just one person re-opens the debate in good faith (which we are required to assume) - then you no longer have it. This means that a decision may flip-flop about as individuals pop up and make an objection.

That is a reasonably clear statement of how to proceed - but then the article dilutes that clear statement.

We are also introduced to the idea of 'rough consensus' - which appears to mean that some kind of supermajority in favor of the view exists - but (as the article quotes) "some vocal and unreconciled folk, some who operate 'outside the law" are allowed to be ignored when deciding whether a rough consensus has been reached. It is in no way clear how many "some" is in this context. We are also told to consider "the strength and quality of the arguments themselves" - but this is difficult because one tends to find arguments in favor of one's own opinion strong - and those of the opposition weak - or else a rational person would already have flipped their opinion on the basis of these strong arguments. This clause therefore makes no sense. The strength of an argument should be measured only by the number of minds it changes - and that quantity is already accounted for by the number of people who moved from one side of the debate to the other.

What we end up with is a very idealistic concept where in a land of rainbows and happy elves, we meet together (probably wearing togas) and have a rareified debate in which everyone listens carefully to great orators and each, with an open mind then forms an opinion on the basis of rationality. At the end, we link arms and proceed as one into the setting sun.

Meanwhile, in the real world
Yeah - right.

In the real world, we can often achieve complete consensus when there are half a dozen friendly editors. Consensus works really well. It forces a debate - it requires everyone to listen to the empassioned plea of the minority - and in the end, everyone is accepting of the answer - even if they don't entirely agree with it. GREAT!

When we have 50 editors, we can sometimes achieve a rough consensus - it's hard to make 50 people agree on anything non-obvious - but by noting that we can ignore those who operate outside the law - and letting our assumption of good faith slip a little - we can often get a decision.

When there are 100 people, decision making is painful in the extreme - sometimes we can appeal to enough existing policy to force through a decision - but it's rare.

When there are 450 people (as was the case for the recent 'non-Admin rollback' debate, we have absolutely no possibility of ever making an actual decision. A 2:1 supermajority is dismissed as useless.  Many believe that even a 10:1 supermajority would not suffice.  It's acrimonious - 200k bytes of arguments spread over several talk pages is simply too much for anyone to read, so we know that great arguments are being ignored by the majority and the minority alike.  The policy document says that we should give more weight to good arguments - but you can't even FIND the good arguments.  Worse still, the original suggestion rapidly gets fragmented into half a dozen similar proposals - some of which are voted on - many of which we'll never know who approved and who didn't.  We are paralysed into inaction.

When this many people are involved and something like 150 of them need to be 'flipped' you simply cannot discuss the matter with them one-on-one. Rational debate in an open forum becomes impossible when the number of people involved becomes large. Worse still, how many of those people took the HOURS of time it would take to read all of the debate on both sides? How many will add their "vote" and never return in order to be persuaded to change it?

Other problems with Consensus
The idea of consensus is that you argue until every one either agrees or admits that they should probably shut up and let the majority have their way. But if you ask a yes/no question, that leaves THREE possible outcomes:


 * 1) We reach consensus that the answer is "YES".
 * 2) We reach consensus that the answer is "NO".
 * 3) We fail to reach consensus at all.

But all of the policies and guidelines on the subject fail to answer the question: When making a binary decision - what happens if consensus is never reached?

If you are going to employ "consensus politics" something always has to be the default position in the event that a decision is not made.

Usually, the default position is taken to be the status quo. If you want to change the status quo, you have to have a consensus to change that. But how did the status quo come about? It came about from some earlier consensus - and our policy says that there is no definite moment when the debate is over and consensus takes hold. We are told that at any time someone may come along and reopen the debate - and at that moment you no longer have consensus. So does it really take a new consensus to reverse a previous conensus decision - or is it enough for one lone objector to wreck the original consensus and thereby remove the previous decision? We have a complete ambiguity in the policy here. Read it one way and once 100% of people agree to do something, it takes 100% of them to agree to change it back...read it the other way and once 100% of people agree, it only takes one person to reverse that position!

This is the cause of a lot of major argument.

Another problem is that what happens when consensus fails depends entirely on how you ask the question. In the case of the "Non-admin rollback" debate, the question was "Should we have this feature added to Wikipedia?" - and the lack of consensus is being taken to be the same thing as a consensus to say "No". But why is "No" the default outcome? Why isn't it that a failure to agree not to do this means that we should to it anyway?

Suppose that instead of asking, some developer had adopted the "Be Bold" approach and simply added the feature - then, when people complained about that change, he goes to the appropriate Talk page and asks the question "Should we remove the non-Admin rollback feature?" then the lack of consensus to remove the feature would imply that it should be kept! Merely asking the reverse question changes the answer.

Imagine a scenario where we have some bad situation where 100% of people agree that a change is required. Suppose are two possible changes on the table 'A' and 'B'. Since 100 people want to debate - we find that there are 20% of people are passionately against 'A' and 19% who are passionately against 'B'. If we try to achieve consensus by asking "Should we do 'A' or should we do 'B'" then we'll fail to achieve that consensus. If the default position is the status quo then we're stuck with what? Doing nothing. That's something that nobody wants. We could ask the two questions separately - "Should we do 'A'?" - and the answer will be "no consensus - so no action" - "Should we do 'B'?" - and again, no consensus. All that's really required to break this deadlock is a simple, straight line, binding vote - but the Wikipedia guidelines don't allow that - so we're screwed.

Conclusion
For things like the form of words used in an article - consensus is fine and has a wonderful track-record of producing well-considered results. But that's because it's possible to express more than one view in a single article. This always allows for a compromise "We'll do both" or perhaps "We'll say neither". But for matters of policy, "we'll do both" isn't a possibility - and "we'll do nothing" is paralyzing - that is not a good way to proceed since policy cannot turn on a dime every time another voice is heard.

So how do we fix the system?
In my opinion changes to the policy is required to allow voting in certain restricted circumstances:


 * In the case where the contents of specific article-space content is involved, the present rules apply.
 * For Wikipedia/WikiProject policies and guidelines:
 * In a situation where less than 30 individuals have contributed to a debate, a consensus is required to deviate from the status quo. Once this consensus decision has been maintained without significant dissent for 1 month, it becomes the status quo and a new consensus is required to undo the change.
 * In a situation where 30 or more individuals are involved in the debate and no consensus has been achieved within 1 month - we ask the proponents of each position to produce their own position papers summarising their arguments - then we take a straight-line vote amongst people who are required to confirm that they have read the position papers:
 * In the case where the decision is to change the status-quo, a 2:1 supermajority is required to carry the decision.
 * In the case where we are choosing between a number of options none of which is the status quo, a simple majority is all that's required.

SteveBaker (talk) 21:33, 12 January 2008 (UTC)