One of the Nine Theses on Wikipedia series

☞ Read this essay on Wikipedia.
☞ Comment on Wikipedia. (And elsewhere on Wikipedia.)
☞ You can also leave comments below.

9. Adopt a legislative process.

Wikipedia’s processes for adopting new policies, procedures, and projects are surprisingly weak. The Wikimedia Foundation (WMF) has launched initiatives, but these do not establish major editorial policy. Incremental policy tweaks cannot deliver the bold reforms Wikipedia needs. No clear precedents exist for adopting significant innovations. The project is governed by an unfair and anonymous oligarchy that likes things just as they are. This stagnation must end. Wikipedia needs an editorial legislature chosen by fair elections: one person, one vote. To establish legitimate and fair governance, the WMF should convene a constitutional convention to create an editorial charter and assembly. This assembly would be empowered to make the sorts of changes proposed in these “Nine Theses.”

The Problem

Wikipedia has changed. Having originated or overseen many of the project’s first policies and procedures, I view the current system as a cargo cult: several policies have changed into forms different or even contrary to their original function.

The first eight theses describe some of the insanity. (1) Decision-making is done by an ersatz “consensus” that pretends to speak for a massive global community, while silencing dissent. (2) A once open, tolerant, global community is now dominated by Establishment commissars. (3) Many sources are blocked en masse for ideological reasons. (4) The neutrality policy itself has been upended, mocking and forbidding actual neutrality. (5) People take “ignore all rules” far too seriously. (6) Editorial leadership is anonymous, even after the project has become one of the most influential media properties in history. (7) Wikipedians disdainfully ignore public feedback. (8) Finally, Administrators routinely block accounts unjustly, tossing out serious editors peremptorily, because they can’t be bothered to deal with inconvenient participants.

What a mess.

These nine theses are my Hail Mary proposal to reform Wikipedia. These things needed to be said. Perhaps, perhaps, there will be some changes—but my hopes are not high. I have proposed many other fixes and tried alternative projects over the years, but there are in fact a number of people, including some very influential people, who like things just as they are. I know very well that, sadly, the system is unlikely to change very much. If it does change, it will probably be through external pressure, or pressure from the hitherto quiet rank-and-file.

All of that said, I am confident that most if not all of these nine theses would be popular among most of Wikipedia’s readers. The proposals are a matter of common sense, appealing to common civilizational values and sound rules of editing and publishing, absolutely consistent with a wide variety of political and religious views.

Still, if Wikipedians were to rally around any of these, one massive problem would remain: There is no legitimate, well-established way to ratify significant reforms. Suppose some of the more respected voices in the Wikipedia community were to agree that the project should repeal “Ignore all rules”—how would they implement such a change? How would they prove a new “consensus”? Twenty-four years after the rule was instituted, who would be properly authorized to declare a “consensus”? It is not clear.

Well, how have structural changes been made to Wikipedia in the last 24 years? In three ways, I think.

(1) The Wikimedia Foundation proposes a change, and it sticks. One example that comes to mind is the Universal Code of Conduct. [a] But this appears to be an exercise in self-protection by the WMF’s legal team, with little impact on day-to-day operation. Other attempts are sometimes objected-to by the community, then abandoned.¹ The most positive example here is the VisualEditor tool—useful, but not a fundamental change in editorial policy or governance. It was promised for many years, rolled out, and then finally made opt-in. So this seems like a possible approach. But some of the louder voices in the editor community resent even such modest efforts by the WMF as overreach.² So, the WMF has avoided making significant new changes over the complaints of the editor community.

(2) Iterate and rewrite old rules. We have already seen several examples of this. The most remarkable examples are policy pages, such as:

• Neutral point of view, [a] discussed under Thesis 4. It was entirely rethought.
• Notability, [a] which began as a simple and expansive set of broad principles, consistent with what is called inclusionism [a] (a preference for a larger set of article topics). But inclusionism gradually went out of favor. As the years went by, fewer people and topics merited an article of their own, so that, now, deletionism [a] (a preference for a narrower set of article topics) is the dominant view.³
• Some of the biggest policies introduced after I left were Verifiability,[a] which was introduced in 2003, and Reliable sources, [a] which did not emerge until 2005. These are an uneven mix of sensible rules, conveniently vague principles, and boneheaded regulations. Eventually, this free-floating body of policy added some real shockers—for example, regarding “consensus” blacklisting of valuable media sources about which there necessarily could be no consensus (see Thesis 3).

From these examples it can be seen that iteration has led to major changes; the massive changes made to the neutrality policy are the best example. But such changes require many years, and they are not necessarily changes that anybody would have actually agreed to, should the matter have come to a vote early on. Slow-moving chaos is, after all, the nature of such change, not unlike a years-long game of “Telephone.” [a]

(3) Somebody writes an “essay,” and people start citing it as if it were policy. Indeed, many essays now are cited as de facto policy.⁴ You can skim through the following sample if you like:

• A good example is “The duck test,” [a] which is a rationalization that Administrators use to block accounts that they suspect, but have not adequately proven, are actually sockpuppets. Though just an essay, this is cited frequently to block accounts permanently—even when accounts are never proven on rigorous, technical grounds to be sockpuppets. Circumstantial evidence becomes sufficient and is often used for partisan gatekeeping. (See Thesis 8.)
• Then there is “Arguments to avoid in deletion discussions,” [a] a long list of vague, unofficial rules about how to argue for and against deletion of articles and text. An oft-cited section is commonly abbreviated as “WP:OTHERSTUFFEXISTS.” [a] This is frequently invoked, in a really shameless way, to justify inconsistency across articles.⁵
• Another essay makes a virtue of the sheer meanness of many entrenched editors: “Wikipedia is not therapy.” [a] If you complain about the stress that Wikipedia conflicts are causing you, some petty editor may cite this rule at you.⁶
• Here is an essay that elevates a practice that I disagree with rather sharply: “Why most sentences should be cited.” [a] I find the extreme proliferation of footnotes absurd, especially when they are irrelevant—for example, when uncontroversial facts are commonly known.⁷ [a]
• Finally, “Assume good faith” [a] is one of the silliest, most misbegotten, and oldest essays that became a “guideline” in Wikipedia-land. Guidelines function as a sort of “policy lite,” but they are typically treated as binding. In effect, “Assume good faith” requires editors to pretend to be naïve; this disables legitimate and serious criticism of manipulative behavior. The “rule” empowers sociopaths. You must not question their motives; you must not notice their subtle insults; you must not observe that your treatment is Kafkaesque; you must participate in the pretense that they are not engaged in a stupid game. The guideline makes Wikipedia function as a humiliation ritual.

These, then, are examples of how opinions, first expressed in essays, can be arbitrarily dignified as a policy or “guideline.” In this way, essays are quite good at allowing individuals and other minorities to make policy tweaks. But they rarely if ever accomplish wholesale change.

In short, there is nothing in place to make significant changes. With the occasional exception of Wikimedia Foundation proposals, there have been few, if any, significant changes after 2006 or so that were not directional tweaks or gradual iterations of existing rules. There is simply no precedent for Wikipedia to embrace radical new ideas like ending decision-making by consensus or enabling competing articles.

There is a striking absence of formal governance structures. The Arbitration Committee is the closest thing that exists, but its function is judicial rather than legislative. I noticed this lack of formal governance long ago.⁸ I never took the opportunity, while at Wikipedia, to start anything like a legislature that could authorize new projects, policies, and procedures.⁹ Wikipedia is aware of this state of affairs, but rarely treats it as a problem. There was a discussion about formalizing governance in 2008, under the heading “Governance reform,” [a] but nothing came of it; it was relegated to the dustbin of wiki-history.

There is also such a thing as “WikiProject Democracy,” [a] but this is just a small group of interested individuals who set up a page pointing to a collection of functions on Wikipedia that happen to be, roughly speaking, democratic. The platform has a fair few procedures driven by input and “votes,” but no process for adopting big new policies and projects. If you ask Wikipedians at present, they will tell you that any big changes would require “consensus,” which is a vague and unworkable idea in itself (see Thesis 1), and laughable as a proposal about how to make significant changes to the rules.

Nevertheless, many Wikipedians seem quite comfortable with how things are at present—despite how evidently broken the system is. This is not surprising, because those at the top of the system are comfortably ensconced. This is consistent with sociologist Robert Michels’s iron law of oligarchy.¹⁰ As Michels put it:¹¹

Organization implies the tendency to oligarchy. In every organization, whether it be a political party, a professional union, or any other association of the kind, the aristocratic tendency manifests itself very clearly. The mechanism of the organization, while conferring a solidity of structure, induces serious changes in the organized mass, completely inverting the respective position of the leaders and the led. As a result of organization, every party of professional union becomes divided into a minority of directors and a majority of directed.

Consequently, we can anticipate that Wikipedia’s “aristocratic” leadership will oppose giving new types of democratic power to the rank-and-file. Probably, there would have to be something like a popular uprising that, as it were, seizes power. How such a popular uprising might work, in the context of Wikipedia, it is very hard to say. (And, no: I will not be leading such an uprising.)

Underlying Wikipedia’s avoidance of democratic governance, there is one fundamental, structural problem: true democracy requires one person, one vote. And that, in turn, requires that Wikipedia user accounts—at least some of them—be paired reliably with real human identities. In other words, if you want to vote on important matters, we must know that you have just one vote; for that to be the case, somebody we trust has to know who all the voters are.

For some Wikipedians, this is a bridge too far. They would never agree to any system that puts their anonymity at risk. But this is not, actually, a sound reason to object to the plan, since identities need not be public. The confirmation can be made privately by sufficiently trustworthy people. Moreover, not all accounts need to be voting accounts. Generally speaking, if you are unwilling to reveal your identity to anyone, you will have no leg to stand on if you try to defend a right to vote.

I can imagine Wikipedians taking issue with this. They might say, and I would agree, that voting should be restricted to active accounts. But then they might propose a “clever” addition. Namely, if an account has not been outed as a sockpuppet,¹² then we should “assume good faith” and treat it as a distinct individual for purposes of voting. The problem here is that we really have no clue as to what proportion of accounts are, in fact, sockpuppets. I suspect that there are many more of them than Wikipedians ordinarily assume. If the percentage of accounts that were sockpuppets were in the low single digits, perhaps the rule would be acceptable. But probably not. After all, if some voting scheme were adopted according to which each active account had one vote, then the number of active (successfully hidden) sockpuppets would most certainly increase, for voting purposes if nothing else. “Assume good faith” is simply an insane rule to follow when it comes to the high stakes of voting.

This might seem to be an impossible problem for Wikipedia, but insofar as we are talking about a serious legislative body, there is a simple and practical solution to it; see below.

Like some other theses, such as Thesis 1, Thesis 5, Thesis 6, and Thesis 7, the present thesis would have Wikipedia grow up, join the real world, take responsibility, and win back some of the legitimacy that it has—sadly, but deservedly—lost over the years.

The Reasonable Solution

Without a means of legitimizing major changes through an online plebiscite, it is hard to know how Wikipedia might come to adopt any of the nine theses. So, this is perhaps the first thesis to pursue in practice. Discussion could be launched by grassroots Wikipedians or by the Wikimedia Foundation Board of Trustees. And the first thing to discuss, I would say, are these five goals:

1. Wikipedia should adopt rules for a constitutional convention. Such a convention would set the rules for the democratic governance of Wikipedia qua editorial organization (not as a legal entity, since the WMF is already legally constituted). But first, there would need to be rules for conducting and ratifying the convention itself. While the convention rules could be drafted collaboratively, on the wiki, I do not think there is any way to legitimate the adoption of the rules on the wiki. In my opinion, only the Board of Trustees could legitimately adopt the rules that would govern the constitutional convention.

2. The primary goal of the convention is to settle on the method and procedure for a Wikipedia editorial assembly. It might accomplish many other things, including the adoption of one or more of the theses proposed, but what is most needful is the establishment of a legislative body, which I would call an assembly. Defining rules would address how to elect members; how often they would meet; what their scope would be; term lengths and limits; what procedures the body would follow; the procedures for calling a plebiscite; the openness of deliberation; and other matters.

3. Both the constitutional convention and the assembly should conduct business at face-to-face meetings paid for by the Wikimedia Foundation. Among other advantages, this solves the problem of ensuring one person, one vote, at least in assembly voting.

4. The unique identity of those voting for members of the constitutional convention and the assembly should be confirmed. The Board of Trustees, with advice from the community at large, should adopt a system that requires that the identities of voting accounts be known to some small, diverse body of trustworthy individuals,¹³ preferably from outside of Wikipedia and the WMF, so that the voting is conducted according to the essential democratic principle of “one person, one vote.” Those Wikipedians unwilling to prove their real-world identities (and ownership of associated Wikipedia accounts) would be able to continue participating in the wiki as usual, but they would not be able to vote for members of the convention or the assembly, or in plebiscites.

5. A first order of business should be the rewriting of policies and the adoption of policy pages, guideline pages, and essays as “official policy” or not. If there is some important distinction between “policy” and “guideline,” it must be formally defined and adopted by the convention or the assembly. Over the years, many details of policies and especially of guidelines have piled up, which should probably be simplified or deleted. The sheer complexity and even incoherence of some areas of Wikipedia policy and guidelines is a much-bemoaned problem: the assembly would be the place to fix such problems. Finally, the vast majority of essays should be deprecated. I strongly advise the assembly to require that all essays, henceforth, be made subpages of their primary authors’ “user pages.” Any essay intended to become official policy should be explicitly adopted by the assembly. A rule should be adopted to the effect that editors should not cite essays in arguing for their edits, or they should not be treated as having any weight.

It can be anticipated that the existence of a democratic representative assembly will lead to factionalism and new problems. But that, in my opinion, is an improvement over the stodgy, oligarchical status quo and a small price to pay for restoring common sense and democratic legitimacy to one of the world’s most powerful media platforms. Perfect unity is not the goal; legitimate governance is.

I should address a few potential objections to this. First, some might say that this proposal threatens to make the project more of an oligarchy than it already is. That would be a puzzling response, considering that Wikipedia right now is not a democracy even in concept: the proposal is to make it more of one, in part because it is already very much an oligarchy. Besides, until we know who the actual leaders of the project are (by acting on Thesis 6: “Reveal who Wikipedia’s leaders are“), we cannot really know the extent to which it is currently oligarchical. Many of those with their ears to the ground have assured me in recent years that the number of voices that really “matter” on Wikipedia is shockingly small. All told, the number of influential accounts is, perhaps, 600 to the low thousands.¹⁴

Second, some might say that this proposal would violate the principle “Wikipedia is not a democracy,” [a] as if this were an inviolable bedrock principle. It is not, of course. It was not an original principle of Wikipedia, I can assure you, and the suggestion (as the just-linked policy page states) that decisions are to be made by “consensus” is bankrupt.

In any event, I and many others who originally developed the project always thought of Wikipedia as essentially democratic. The project now has the funding—and potentially the organization and maturity—to make it actually democratic, as it should be. This is the necessary first step to making Wikipedia finally a just, neutral, and robust community, capable of reliably summarizing everything that humanity knows.

One of the Nine Theses on Wikipedia series

Footnotes

1. An example of that was explored by Thesis 7: the public rating system.[↩]
2. An example of this is when the WMF banned an admin, “Fram.” See Ashley Rindsberg, “How the Regime Captured Wikipedia,” [a] Pirate Wires, Aug. 5, 2024.[↩]
3. I myself am close to being a radical inclusionist. There is no reason for people not to start articles about many quite trivial topics, as long as they are of good quality and the privacy of private individuals is respected.[↩]
4. This is another instance of Wikipedia acting like a cargo cult. The very idea of “essays” emerged from my own practice, in 2001, of spending 15-30 minutes a day unburdening myself of random thoughts and putting them on my user page. Later, these essays were removed to where you can still read them, on my user page on meta.wikipedia.org.[↩]
5. In other words, an editor wants to make an exception to what really ought to be a general principle that holds sway across many existing articles. If you argue that other articles exist that have similar features to the one under dispute, then this “policy”—which is only an essay, not actually policy—will be cited. The fact that “other stuff exists,” i.e., articles that follow a pattern, does not mean that the pattern should be followed in the present case. That is supposed to end the dispute; in the hands of influential editors, it often does.[↩]
6. Cf. Andrew McMillen, “Wikipedia Is Not Therapy!” [a] Wired (August 15, 2016). This is the story of “Elliott,” who almost committed suicide after his account was permanently blocked, on which, see Thesis 8. I feel compelled to point out that this is yet another example of how Wikipedia changed—for the worse. Until fall of 2001 or so, Wikipedia really was open, welcoming, and fairly friendly. It gradually became less so. On my “user page” [a] (i.e., my public presence on the wiki), for many years, I have had these two pieces of advice: “May you continue to be open and warmly welcoming, not insular, … [and] to show the door to trolls, vandals, and wiki-anarchists, who if permitted would waste your time and create a poisonous atmosphere here.” Wikipedians, who take their nasty and tiresome game-playing very seriously, clearly did not heed such injunctions.[↩]
7. As a style, this is found only on Wikipedia and, I suppose, some academic writing such as law reviews. It is typically useless to the end user, and a distracting artifact of edit-warring on Wikipedia. The sheer clutter can also pose a serious impediment to easy editing; sometimes I wonder if that is intentional.[↩]
8. See my memoir, “The Early History of Nupedia and Wikipedia,” reprinted in my Essays on Free Knowledge: The Origins of Wikipedia and the New Politics of Knowledge [a] (Sanger Press, 2020). This is why Citizendium.org wrote and adopted a community charter, [a] which had legislative authority, in 2005–6.[↩]
9. I recall idly thinking I should propose some governance structures, but before I could, Bomis ran out of funding for my position in the collapse of the Dot-com Bubble.[↩]
10. As some researchers have argued. See Shaw, Aaron, and Benjamin Mako Hill. “Laboratories of Oligarchy? How the Iron Law Extends to Peer Production.” arXiv preprint arXiv:1407.0323, July 1, 2014. https://arxiv.org/abs/1407.0323. [a][↩]
11. Michels, Robert, Political Parties: A Sociological Study of the Oligarchical Tendencies of Modern Democracy, trans. Eden and Cedar Paul. New York, 1968: Free Press. Originally published 1911. Quoted from the Wikipedia article “Iron Law of Oligarchy.”[↩]
12. I.e., an extra account run by someone who already has an account. On some of the issues with sockpuppetry, see Thesis 8.[↩]
13. I.e., not closely associated—in order to make bribery and corruption more difficult.[↩]
14. The latter, if one includes active accounts that have “Autopatrolled” rights.[↩]