Wikipedia:Decision Making Process

For Wikipedia to be able to scale, it is agreed upon by many of its users that the project needs a decision making process to formalize those policies that should be consistently enforced. Such a process could also be used to interpret the policies in individual cases.

The following different types of process have been suggested:

Voting See also: wikipedia:vote

Voting can be implemented in various different ways. Issues and persons can be voted on. The following is a proposal to vote on specific issues, both policies and their individual application. It was originally proposed on wikipedia-l by Erik Möller.

Users can create two types of polls, inquiry polls (non binding) and policy polls (binding, with enforcement). Only a smaller group of users (still larger than the current admin group) can create policy polls, but the same group of users (e.g. everyone with >=n contributions) can vote in both types of polls. Policy polls can contain only specific types of options: ban user X, delete page Y, etc., but still follow the same discussion/voting principle.

Polls get their own namespace, and on the page where the poll is, users can also provide arguments for or against the different options. So I would go to Poll:Ban Trollxy[?] and could see the different opinions and vote on them. The poll page might look like this:

Inquiry polls would allow the options to be defined freely and primarily be used to gather opinions in less extreme conflicts among reasonable persons. As voting styles, both first-past-the-post (winner takes all) and preferential voting are reasonably simple and should be supported, policy polls work better with fpp voting (clearly distinct options).

Recently added polls would be listed on a separate page like Recent_changes. The poll would be closed after a given timespan, defined by the person who creates it. For policy polls, depending on the type of action, we could set different threshold for whether we want to take it, e.g. banning an anon user should be easier than banning a signed in user. Minimum number of votes may be necessary, but not too high.

Possible problems

Voting is criticized on a number of levels. For example, Kenneth Arrow has suggested several criteria that a democratic system should seek to provide, and concluded that they are mutually exclusive (Arrow's Theorem (http://c2.com/cgi/wiki?ArrowsTheorem)). Especially the common first past the post voting style is frequently criticized for producing undesired outcomes. (It is worth pointing out, at this point, that most criticisms of voting systems focus on systems where persons/candidates are voted on, not issues.)

It seems to be the case that first-past-the-post voting is problematic when dealing with many similar options, such as "NPOV should be a standard policy", "NPOV should be a standard policy, but with amendment X", "NPOV should be a standard policy, but open to modification" and so on. In this case, a fairly equal distribution of votes is possible, with no clear majority for one option. Here, especially, preferential voting could tell us which option is generally most agreeable. This also avoids strategic voting, because the voter can still give his personal preference the highest ranking.

In other cases with fewer, more distinct options, first past the post voting might be more appropriate and less problematic. When a poll basically boils down to "Yes", "No", and "No opinion", there does not seem to be much potential for strategic voting.

The power to be obtained from framing the question cannot be ignored.

Voting relies on having a clear electorate, and a means for individuals to reliably identify themselves. Wikipedia has neither.

Presumably the poll text will be frozen once voting begins?

Consensus

It has been suggested that decisions should be made strictly through a discussion process aimed at reaching a broad consensus that a certain decision should be implemented. For example, on the question whether users should be banned for consistently violating the NPOV policy, a discussion could be held where every interested party would present their arguments for and against a particular solution. If a consensus can be found, the respective solution is implemented. Otherwise, nothing is changed.

Possible problems

Several criticisms can be brought against such a process. The success of a particiular decision making process does not seem to depend on how good the arguments or how clear the opinions are, but merely on how motivated the individual participants are. Even when the majority clearly agrees that a certain action is necessary, a single individual motivated to post long justifications for their positions may justify the process ad infinitum. While it can be argued that this is desirable in order to hear all arguments, arguments often depend on a particular point of view, such as "humans cannot be trusted to act nicely", or the exact opposite. Thusly, arguments might quickly become deeply philosophical, about questions that have divided mankind for ages.

In other cases, a small minority might manage to get a particular change implemented for the simple reason that most people do not care enough about it to write detailed arguments against it. The majority may then, eventually, be surprised by a change that nobody really wanted.

Conclusions

What, then, can be learned from the above? The different options should be weighed against each other, and a decision should be made.