 | This page in a nutshell: WP:CONSENSUS usually works well, and sometimes works very well, but sometimes does not work well |
One of the key policies of Wikipedia is that decisions are (normally) made by consensus. Sometimes this policy works very well. Sometimes it works reasonably well. Sometimes it does not work very well. This essay contains thoughts on when consensus works and when it does not work as well. One of the limitations of consensus, that I will refer to in this analysis, is that Wikipedia has multiple classes of stakeholders, some of whom are represented better than other classes. In particular, readers (as opposed to editors) are not represented except to the extent that editors have the same interests as editors. The limitations of consensus in general have to do with what classes of stakeholders are represented in determining consensus, and when there is a difference between "local consensus" and "global consensus".
First, consensus works well for determining the contents of articles, with a few exceptions. Readers expect articles to be accurate and unbiased, in accordance with other policies of Wikipedia such as neutral point of view. Normally, the editors of an article are collectively seeking accuracy, completeness, and neutrality, so that a consensus of editors will serve the interests of readers. The most obvious exception is when there is a "consensus disconnect" between a consensus of Wikipedia editors and a larger scholarly consensus. A "consensus disconnect" sometimes occurs with regard to fringe science, where there is a scientific consensus that a viewpoint is fringe, but a group of Wikipedia editors support the fringe position, so that consensus cannot be reached in Wikipedia to deprecate the fringe viewpoint. Fortunately, in these areas, the global consensus of Wikipedia is that fringe science should be labeled as fringe science, and the ArbCom has authorized the enforcement of that policy.
Second, consensus has a mixed track record with regard to user conduct. It works well at blocking or banning trolls, flamers, vandals, and other editors who are not here to build an encyclopedia. It does not work as well, and sometimes does not work at all, in dealing with editors who dislike each other and have long-standing enmities, or with editors who are uncivil but have reputations as excellent content creators. One reason is that "community consensus", established at the noticeboards, is really only the consensus of users who frequent the noticeboards, who tend to be a fractious and contentious subset of editors. My own opinion is that "community consensus" at the noticeboards is something of a will-o-the-wisp, except for users who are clearly NOT HERE, but that is only my opinion. Some means of dealing with user conduct issues is needed that is less extended than the ArbCom, but can be appealed to the ArbCom. I would prefer to do away with community enforcement with respect to conduct issues, but will be satisfied to retain it as an alternate path if there is a route to the ArbCom other than Go Directly to ArbCom.
Third, consensus is usually the method by which English Wikipedia policy is made, and consensus usually works reasonably well. However, consensus has not worked and is not working with respect to paid advocacy editing, where all attempts to reach consensus have failed. The underlying reason, in my opinion, is that the editors and the readers do not have the same interests. Many editors are interested in the process as well as the content, and have a variety of strongly-held views about what the process should be. (Some editors think that Wikipedia is an experiment in electronic democracy. Some editors think that Wikipedia is an experiment in electronic anarchy. Some editors think that Wikipedia is a a democracy.) Because there is no single point of view, consensus is not reached. Readers are only interested in the content, and do expect that Wikipedia will be accurate and present a neutral point of view. (Also, it is clear to me that some editors do not understand how paid advocacy editing is an existential threat to the neutrality of Wikipedia.)