On genuine neutrality versus enforced consensus

This essay was first posted on Wikipedia.

There is a very serious problem about what goes under the title “consensus” in Wikipedia. Does not the very fact that a supposed consensus can represent a single, controversial position, and that it needs enforcement, suggest that it is not really consensus at all—and that the enforced position is not, in fact, neutral?

Consider Wikipedia’s neutrality policy. It states:

Articles must not take sides.

This policy is non-negotiable, and the principles upon which it is based cannot be superseded by other policies or guidelines, nor by editor consensus.

Wikipedia aims to describe disputes, but not engage in them.

Avoid stating seriously contested assertions as facts.

The tone of Wikipedia articles should be impartial, neither endorsing nor rejecting a particular point of view.

There are many other aspects of the “Neutral point of view” policy page, circa 2026, that are very problematic. But the above statements, at least, continue to express the idea that neutrality involves not taking sides; describing disputes, not engaging in them; acknowledging contested assertions as such; and maintaining an impartial tone. And note again a key point: Neutrality cannot be superseded by “editor consensus,” says the text. Very well.

Compare this with the original ideal of consensus. The procedure to solve editorial conflicts amicably was—or rather, used to be—to seek “consensus,” and that involved allowing everyone (generally speaking) to have a say in the final, settled text. The consensus was not a particular point of view. “The consensus position” existed no more than “the neutral point of view.” This cannot be stressed enough: Originally, when we had to resort to a consensus, what was generally agreed-to was not one position, but a statement about how to fairly represent multiple, competing positions.

But now look at how the policy about “Consensus” describes the process of “ascertaining consensus”:

Consensus is ascertained by the quality of the arguments given on the various sides of an issue, as viewed through the lens of Wikipedia policy.

Wait—what? Are we really judging the quality of arguments about the various sides of an issue? Yes, that is what the text says now. Consensus is no longer primarily about characterizing the controversy. It is primarily, in practice and as we get into the details of the process, about picking winners and losers. Once determined, the winning position is blessed as “the consensus.” Losing positions are “outside the consensus.”

It is striking that, in Wikipedia:Consensus, limited attention is given to the idea that we might achieve consensus through the fair accommodation of multiple positions. There is some initial lip service given to compromise through characterizing the controversy, but this is presented as one not-very-likely tool in the toolbox; it is neither prioritized nor operationalized in the settlement of especially stubborn disputes. As a result, an essential part of the “consensus-building” process now involves closing discussions: Once a consensus has been “reached,” then—within certain limits—no more discussion is permitted. “After a while, it is time to close the discussion so that the community can move on.” Thus, an editor (often an administrator or distinguished editor) takes on the role of declaring a consensus and closing discussion. Such a person is called the closer.

In a semi-official essay offering “Advice on closing discussions,” we see clearer evidence that what occurs is not compromise, not an earnest attempt to represent competing views, but the determination of a winning view among competitors:

The influence of each argument on the final outcome should be weighted by the strength of the argument, with reference to existing consensus as represented by e.g. the Wikipedia policies and guidelines. Numbers play a role, but they do not determine the outcome on their own.

Arguments based on misconceptions of policy/guidelines, or which do not contain policy/guideline-based reasoning at all, should not be considered in your analysis.

No numbers can make up for an invalid argument, and an argument is not made stronger by another editor repeating it.

SPAs, socks, and canvassed editors will normally get little to no weight.

This is the language of determining best and worst arguments, of selecting winning and losing editors. I admit that, due to its vagueness, it is consistent with the idea that consensus-building involves compromise in a process of explicitly documenting disagreements. But, while the word “compromise” does make a weak showing on the policy page, Wikipedia:Consensus, it is not mentioned at all on the more practical guideline page, Wikipedia:Closing discussions, or the more descriptive essay, Wikipedia:Advice on closing discussions. The fact of the matter is that Wikipedia’s “closers,” the declarers-of-the-consensus, do often take sides. Prima facie, that would appear to contradict the neutrality policy. They routinely portray contested assertions as fact. Thus, in practice, the determination of a “consensus” and “closing” discussion are inconsistent with the requirements of neutrality.

There is no shortage of examples. Open up almost any long-standing Wikipedia article on a Culture War topic. Find any seriously contested assertion made in wikivoice. Then hunt through the talk page for discussion about that assertion, and you will often find confident claims about the “consensus,” wielded sternly against people who—well—disagree with the consensus.

A particularly eloquent example can be found on the page titled Gaza genocide. Despite the fact that many people disagree—including both founders—the article’s current verbiage asserts in wikivoice that the Israelis committed a genocide, providing very little coverage indeed of any other views. If you will look at the talk page, there are dozens of references to an alleged “consensus,” both in the international community and on Wikipedia itself. This “consensus” is used as the basis on which to silence debate. This illustrates very well how “consensus” actually can operate in Wikipedia: as a tool to ram through one point of view as “the neutral point of view,” or as the sort of “consensus” that silences dissent.

For a second example, see COVID-19 lab leak theory. The article asserts in wikivoice that the lab leak theory is a minor view lacking evidence. For example, Wikipedia editors wrote the following in the article without attributing these assertions to anyone: “There is no evidence supporting laboratory involvement, no indication that the virus existed in any lab prior to the pandemic, and no record of suspicious biosecurity incidents.” Of course, this is now very widely disputed, including by the FBI and Department of Energy. Naturally, if you look at the talk page for the article, you will find much wielding of “consensus“-talk to silence and suppress disfavored views. For example, in 2025 discussions and RfCs, prior “consensus” was brandished to reject including updated intelligence findings in the lede; it seems including them would constitute “undue weight,” despite new reports.

If we do not simply dismiss the institutional fiction of consensus altogether, then I have a more modest, intermediate proposal. On all three pages mentioned above, there should be much more explicit verbiage to the effect that the immediate goal of “consensus” is not to declare winners and losers on questions of substance, but rather to seek compromise among Wikipedia editors on those questions; similarly, it is not to anoint one controversial position as “the consensus,” when there is manifestly no consensus, but rather to depict the various positions in a controversy fairly and charitably.

So, the rules pages about consensus must make it absolutely, pristinely clear what sort of question consensus-building settles. Does consensus-building typically settle

1. a pre-existing, substantive question that the article raises, or
2. a talk page controversy over how to characterize the positions on such a pre-existing, substantive question?

For example, which of these two questions does consensus-building settle?—

1. Did Israel commit an act of genocide against Gaza in 2023–25?
2. How should the Wikipedia article on the alleged Gaza genocide characterize different views on whether to describe the events of 2023–25 as a genocide?

This is an important distinction that the Wikipedia rules do not yet draw. Not only should we draw this distinction, we must insist that Wikipedia’s consensus-building process, as such, can only seek to settle the latter type of question.

Question Type	Definition	Example
1. Substantive.	A pre-existing question raised by the article.	Did Israel commit genocide in 2023-25?
2. Editorial.	A question about how to characterize positions on the substantive question.	How should the article depict views on whether the events qualify as genocide?

The difference is very important, because substantive questions, of type (1), can be so controversial that we can never arrive at any sort of consensus about them. But editorial questions, of type (2), are more tractable, if they are understood as being subject to rules about neutrality, and especially if neutrality itself is understood as the method of characterizing disputes rather than picking winners and losers.

In short, then, consensus processes must aim to settle editorial questions, not substantive ones. They must prioritize fair representation of all significant views on substantive questions, as NPOV requires, rather than endorsing a single position on substantive disputes.