Voting in a wiki can be used for a variety of purposes. It can be used to approve admins and bureaucrats, to make changes to policy pages, to discover editor preferences, and to make those collective choices that benefit from unification but are rather arbitrary, as is, in the physical world, car driving on the left or car driving on the right.
In English Wiktionary, the key policy pages—those for criteria for inclusion and entry layout—are protected, and are meant to be modified only through a vote. This policy-locking has stopped uncontrolled growth of policies. Policy-locking did not prevent seeking consensus in the main namespace. It did not bar discussion in Beer parlour. Instead, it prevented wanna-be regulators from further improsing rules through editing policy documents, subsequently claimining that their edits to the policy documents represent community consensus, and that "this is how we do things in Wiktionary".
Compared to seeking reconciliation of differences by editing in the main namespace and by conducting discussions in Beer parlour, voting has some advantages. It is timely: it ensures a collective decision is made within a given time frame, or it becomes clear that consensus swings in neither way. It enables broader participation: many of those people who do not bother to discuss in Beer parlour come to a vote. It provides transparency or traceability: it prevents editors from falsely claiming something is a community decision. It lessens the power of verbose editors who are eager to discuss and often dominate a Beer parlour discussion. It sometimes discovers editor preferences for the first time, preferences unobvious from the editors' editing in the main namespace.
Vote pages are the most authoritative and reliable documents of an existing community consensus or its lack. They make the presence or absence of consensus conveniently available for anyone to inspect, without the need to perform a costly research into the prevailing practice in the main namespace, and into past discussions in Beer parlour.
One weakness of voting in English Wiktionary is its poor definition of what "consensus" means in relation to votes. The percentage of 70% has been used in some votes as the threshold that a vote has to reach in order to pass. A case in point is Wiktionary:Votes/pl-2009-03/Removing vote requirements for policy changes, although this vote was like a change in constitution, so expected to require a higher threshold; and the vote was left without a clear closing. Another mentioned threshold was 66%, or 2/3. Both thresholds give implied preference to the status quo, with the status quo often having no more legitimacy than what the vote proposed. An advantage of this high threshold is that it prevents further proliferation of new regulations unless they are supported by a strong consensus. The absence of a formal policy on a particular subject is often no fundamental hindrance to the progress of Wiktionary. Various disunities can develop in the main namespace without significantly barring the progress of Wiktionary.
A vote that did not pass is not necessarily a failure on the side of the proposer or the voting community. While creating votes that are rather unlikely to pass is a poor practice, the proposer of a vote does not usually know in advance how much support the proposal is going to generate. Before the vote runs its course, the proposer can only estimate the result of the vote from patterns of editing in the main namespace, and from the discussion that preceded the vote, a discussion in which eloquent opposers sometimes create the impression that there is a strong opposition when in fact the vote can easily pass. Thus, votes are not only a formal means of confirmation of consensus that everyone already knows is there, but rather also a means of discovering the scope of support or editor preferences for the first time.
Discussions on voting: