User:Imetsia/The deleter role revote

This essay is a continuation of User:Imetsia/The deleter role. Please refer to that essay for background information and previous arguments.

Months have passed since the deleter-role vote was first proposed. Half a year later, the need to establish the role has only become more pronounced. RFV/D have slumped into disrepair, and administrators continue to be weighed down by deletions that other users could perform just as well. Ordinary contributors, stripped of the deletion right, are stuck having to hand trivial deletions over to their sysop colleagues. The whole arrangement all but ensures that our work is not executed as efficiently as it could be.

The establishment of the deleter role could solve all of these issues, thus giving ample justification to a revote. Still, there are some editors who will stubbornly oppose it. Apart from their substantive arguments addressed in the prior essay, two more will inevitably arise: that the revote is untimely, and that the proposal is too bureaucratic. These have been pillars of the cases against previous votes, but they both stand on shaky ground. The idea that revotes are always unacceptable runs contrary to consensus. And in this case, the anti-revote argument is exceptionally weak, as every factor in favor of a revote exists in quantity. Meanwhile, the claim that the vote is “bureaucratic” is self-defeating, and even its staunchest supporters have begun to abandon it.

In addition to addressing all of these issues, the revote also deadens the rest of the objections by the prior opposition. The nomination process is more robust, hidden revisions will remain inaccessible, and the right can easily be revoked if misused. With so much change to the proposal and with the other arguments having been discredited, there is no reason for anyone to vote against.

The need for the deleter role has only grown with time

Seven months ago, when the vote was first run, the creation of a deleter role was a policy of great importance. The closings of RFVs and RFDs limped along at a glacial pace; administrators were tasked with the dull assignment of deleting every entry with a misspelled title or other nonsense; and the work of administrators was less efficient because every little deletion had to be handled by them.

The picture is no different today. After opponents’ long-term neglect on the issue, the situation at RFD and RFV has continued to worsen, and administrators are still burdened with menial activity. If anything, our deletion forums have lagged behind even more. Entries collect dust for months on end when the result of a discussion is clear and the entry should be struck. Admins rightly prioritize matters more worthy of their attention, such as “blocks, monitoring… debates, [and] adjudication in disputes” over janitorial work. This is probably the right decision: The faith that other users placed in a sysop in his exercise of judgment and dispatch goes to waste when his time is spent on smaller concerns. Without the deleter role, these are problems that have plagued our disposition of entries at RFD and RFV for a long time. While they are not new issues, they continue to intensify in severity.

Take just one example of the decay at RFDN. It was over a year ago that we achieved consensus support for mass-deleting entries in Category:Italian cardinal numbers and Category:Italian ordinal numbers. Sixteen months later, not even half of the entries that should be deleted have been. It should be no wonder, however, because we have given the mammoth task of deleting those hundreds of pages to just one user. It is not for lack of trying that those entries have stuck around on the site to this day. Hundreds of pages have been deleted, but it is still not enough. And even after asking other users to help with the deletions, the problem remained, as no one answered the call. This lack of manpower means that the democratic decision of our community has been arrested for far too long. The permanence of the entries stands as both a violation and an insult to our consensus-building process. It is an indication that we have lost our way in the management of the entry-deletion power.

The deleter role would go a long way in solving these issues, and it is long overdue that we implement it. Without a more inclusive deletion power, RFV and RFD have been hopelessly backlogged despite the herculean efforts that we have employed in the past in the empty hopes of resolving the issue. Sysops have been constantly distracted by simple deletions, and non-admins have been working less productively because they have been unable to delete bad entries on their own. With time, what was once a mild annoyance has ballooned into an unbearable problem. Now is the time to finally solve it.

There is no excuse for further obstruction of this vote

So often when a good vote is presented, there is a group of users who emerges from obscurity to hinder the proposal in every way they can imagine. This comes in many forms, but the most prominent is a two-prong attack: First, complain that the vote is too early and that it must be deferred; then argue that its implementation would be too “bureaucratic.” These tactics have been the two millstones that obstructionists have relied on time and time again.

The typical objections to revotes are invalid here

So many users protest votes that are proposed in succession, insisting that no revote is legitimate unless delayed.

Several users have staked out the position that any revote… is flatly impermissible unless delayed. Such arguments elide a useful distinction between an effort to overturn a decision on its merits and one to reconsider based on a procedural question. But it does not change the fact that legion users have already convinced themselves that they must presumptively oppose revotes in any form.

—Imetsia

The argument is an extreme one that prevents review of a procedurally-flawed vote for a long time. Thus, even a result that everyone agrees was ill-gotten must stand for several months for no other reason than to appease a limited conception of the voting process. That outcome is antithetical to Wiktionary’s principles. In our regulatory order, wrongful consensus is no consensus at all. And consensus that is “wrongfully slanted or… negated” or in which “some… procedure was not followed” should be overruled. The only way to overrule a wrong precedent is to have a revote. If we were to ban revotes, we would be left with no recourse to annul bad decisions.

In this situation, that is precisely the case. A discussion was put forward prior to the start of the vote, and the vote had a dormant period to allow for greater dialogue. Nevertheless, users neglected their duty to contribute to discussions and instead raised never-before-seen objections directly at the time of voting. Thus, the vote never had a chance to be amended to invite greater participation and support. That is a procedural failure, and it poisoned the validity of the vote. Even so, under extraordinary conditions that put the integrity of the process in serious doubt, select users refused to permit modifying the ongoing vote. This was, in other words, an example of procedurally-flawed and null consensus. It should have been recognized as such, and a revote should have promptly been allowed. But the typical crowd of users once again came out of nowhere with the explicit intent of preventing the right procedure.

Now, after so much time has passed since the first vote, there can be no more bad-faith argumentation against a revote. Even the most inveterate partisans have agreed that revotes are acceptable when the composition of the community has changed, views on an issue have evolved, or greater time has allowed for new revisions. With the deleter-role vote, all three criteria counsel revisitation of the original vote.

Over time, the makeup of our user base changes, and it is only fair that their viewpoints be considered on important matters of Wiktionary governance. The existence of a prior majority does not “compel unending adherence” to its decision. New users are entitled to weigh in on the issues, which may cause us to render a different consensus to displace the one before it. Such activity is both natural and desirable, as it allows us to update policies to be more congruent with the will of our community.

Since the first vote, new users have come onto the scene, and established editors have taken up new roles. The spate of editors who have recently joined our project is substantial. At RFV/D, the Beer Parlour, and elsewhere, emerging voices are greatly influencing the discussions held in those forums. Meanwhile, users that had already established themselves on the platform have been promoted. As many as three users have recently become admins since the conclusion of the deleter-role vote.Their status could make a difference both in raw vote count as well as their ability to lead by example. As other users have noted before, admins give credence to new ideas. In practice, proposals seem more legitimate when they bear the imprimatur of an administrator. Both the change in the constitution of our user base and the new editors that have joined the ranks of administrators would make ballot-casting a much different affair this time around.

Additionally, the rise of certain users gives a better idea of which editors might be fit for the deleter role. For some, the concept of a group of contributors competent enough for the deleter role but not quite good enough for adminship has been difficult to understand. The ascendance of particular users in our discussion forums, however, should give a clear example of the type of people for whom the deleter role would be ideal. It is that subgroup of editors who have demonstrated participation, insight, and judgement already in our discussion rooms that would benefit from a bestowal of the deleter role. They might not quite have the experience to be a sysop, but it would certainly benefit the server if they had the deletion power.

Second, changes in user opinion also suggest that it is time for reappraisal of an initial vote. The sole purpose of our votes is to survey public opinion among our editors and legislate accordingly. Consensus is our mechanism to issue or modify policy. So when consensus changes, policies must adapt to fit the new circumstances. Revotes then become the only way to ensure that our regulations are up-to-date and reflect the wishes of the very people under their rule. Some have argued that revotes are just a vehicle to game the process: Through them, a vote-creator can give himself as many chances as he wants to hope for a different consensus, and he may continue rerunning the same vote until he is “happy with the result.” But seen in a different light, revotes are vital to the consensus-making process. They assure outcomes more consistent with evolving public opinion, and they recognize the fact that consensus is not eternally fixed. Revotes should not only be tolerated, but welcomed, because they help codify changes in community opinion.

In this case, the policy positions of editors have been in flux. For one thing, the already slow deletion process has gotten worse and worse, making an already difficult situation truly dire. That alone will probably have changed the minds of a few skeptics who did not even so much as acknowledge the troubling state of RFV/D before. In addition, both prior to and during the run of the first vote, the count of users supporting the idea had been oscillating wildly. In the discussions preceding the vote, I originally stated that “there have been at least two instances [in which] users have wanted to give over the deletion power but withhold the block power;” but I had to quickly amend the comment after Fay Freak had said something to the same effect. In the course of less than an hour, the headcount of supporters had already changed even prior to the vote’s inception. When the vote did start, another person also switched sides. With the preponderance of support-to-oppose votes in the balance, and with such large and unexpected swings in allegiance, reconsideration of the prior vote is in order. Considering also the ever-worsening conditions in our deletion forums, all signs point to the urgency of realigning our policies with public opinion. This too should point us in the direction of a revote.

New factual developments since last December could also change public opinion. Since the last vote, we have learned that fourteen other wikis also have a deleter role. It has worked just fine for them. While opponents project all sorts of chaos and difficulty due to the deleter role, none of their concerns have come true on other wikis. Deleters have proven to be responsible users that help get things done more efficiently on those platforms. That fact provides powerful ammunition against the key argument from the opposition the last time: that the deleter role would give users too much power.

Third, lapses of time and modifications to a proposal figure heavily in considering revotes. The two are closely related, as greater time allows a vote-creator to amend his proposal to accommodate the prior opposition. Revotes give us the opportunity for compromise and conciliation. Thus, they generally put forward more finely-tuned policies that appeal to a greater swath of the voting population. Even a proposal that was defeated soundly the first time may come back in a better form and attract significant support the second time.

The deleter-role vote is a prime example of the benefit of more time when writing policy. After the original vote, we have benefitted from all the responses that the voting community has given us. Accordingly, as detailed below, this revote announces a number of changes. It is a proposal that many editors who had found themselves in the opposition will likely embrace. It addresses the key concerns of those who voted in opposition last time. With these edits, the revote should gain 63% approval by the most conservative estimates, within striking distance of the passage threshold. And it may well draw as many as 83% of users to support in a more optimistic scenario. So there are substantial hopes that a revote would pass with all of its modifications, ever more reason to welcome a rethinking of the original proposal.

A second vote on the deleter-role would have abundant justification. It would be so inherently sensible that even those predisposed against revotes would have to make an exception this time. Every indication supports rexamination of the initial proposal: We have a different base of users, opinions have been rapidly changing, and a great deal of time has passed since the first iteration. Given the wealth of reasons for revoting, even the editors most prejudiced against it will no longer have their preferred excuse against this vote.

Opposing revotes for its own sake violates consensus

We should seek to overturn wrongful votes as quickly as possible. Not only do our policies require us to do so (see above), but our users have mandated it time and time again. Consensus is the binding principle in all corners of legislative activity on Wiktionary. Here, editors have overwhelmingly rejected the idea of disallowing revotes. Opposing revotes for its own sake is a practice that we have shunned on multiple occasions, but it is an impulse that continues to hold influence over a particular group of users. Not only does this gum up the works of our democratic processes, but it is itself a repudiation of the democratic principle.

We have voted twice before on the idea of banning revotes within a certain time period. The first effort, to limit a revote within a week of the initial vote, failed. Users were uncomfortable with setting unnecessary limits on the creation of votes, thus stunting our ability to reverse bad policy. Even a period of one week was too burdensome for a majority of users, let alone the multiple-month periods that some people have floated more recently. Unsatisfied with this result, Victar proposed another vote to the same effect, and it failed once more. Ironically, this vote against revotes was itself a revote. And this time, another supermajority reaffirmed the need for quick revotes.

Users in both cases recognized the necessity of allowing votes in succession. In 2016, it was a matter of avoiding useless constraints. In 2021, editors univocally championed the cause of allowing fast revotes. Anything to the contrary would “prevent us from… immediately overturn[ing] wrong decisions”, leading to a situation in which “badly structured votes [could not] be reversed or amended.” “[I]t would make it impossible… to overturn a vote (within a time period), even if it became very clear that the decision was wrong.” Very relevant to the deleter-role proposal: “If a failed vote inspires a new proposal that is a great improvement and well-liked, it would be mad not to allow an immediate rerun.” This is just a brief collection of the many opinions among the supermajority who opposed the quick-revotes ban. There are still more who echoed the same concerns. This is a striking endorsement of quick revotes on the part of our community, and all editors should commit to following this consensus.

The long history and strong tradition of issuing revotes confirms our community’s acceptance of them. Revotes are an ancient practice, dating to the earliest days of Wiktionary’s history. Since then, we have repeatedly voted on the same issues many times. If, as opponents argue, revotes were an unfair manipulation, then one would not expect so many revotes in our history. The proliferation of repeat votes since our founding is difficult to square with opponents’ distaste for them. Malgré them, second votes have been commonplace for years, and large majorities of users approve of them as foundational to the creation of consensus.

That is not to say that all immediate revotes are good. “[A] vote rehashed and reproposed immediately after its conclusion is generally to be resisted.” Many revotes could be issued with the sole intent of disrupting the system and should therefore be struck down. However, what separates an abusive revote from a legitimate one is two things. They both relate to the character of the first vote that it seeks to uproot: “the nature of the[] error” and “the quality of the[] reasoning.” Both criteria had been amply satisfied for the deleter-role vote: (1) The initial vote was not only wrong, but a procedural nonstarter, so the error was particularly grave; and (2) the logic behind the deleter role was very convincing, encouraging a fast redo.

For just these reasons, we should have had the opportunity to revote quickly. In a perfectly just and rational world, we would have aborted the first vote and issued the necessary modifications in short order. We would have been able to overturn an “egregiously wrong and deeply damaging” negation of consensus “at the earliest opportunity.” We were not able to do so because of the typical crew of revote opponents — people who rail against repeat votes for their own sake. Their position is untenable, because it advocates for the entrenchment of wrongful decisions, no matter how mistaken or procedurally defective. They also disregard the fact that revotes are deeply rooted in our history and tradition, and that multiple majorities have affirmed the legitimacy of repeat votes. Thus, they not only advance a nonsensical argument, but they also do so in defiance of preexisting consensus. Fast revotes are a logical and necessary part of consensus-building, and one should have been allowed for the deleter role.

The bureaucratism argument is likewise dishonest and invalid

In any discussion about a new policy, the argument that it would be too “bureaucratic” is almost a given. Sometimes editors simply want to oppose a proposal, but they struggle to think of a rational basis to vote against. Naturally, then, they turn to the lazy excuse that the proposal would lead to bureaucratization. These arguments are dishonest and opportunistic.

In the first run of the vote, Hazarasp gave his version of the bureaucracy line:

Aside from the very sound points raised by others, this is exactly the kind of overelaborate, ill-advised bureaucratism we should be trying to avoid here at Wiktionary. While it isn't likely to have much of a encumbering effect in itself, the passing of this vote could play a role in legitimising the foisting of greater and grosser bureaucratic loadstones upon Wiktionary in the months and years to come.

—Hazarasp (parlement · werkis) 08:30, 12 January 2022 (UTC)

The most amazing and radical part of his argument is his admission that it is founded on a false premise. Hazarasp begins with a superficial attack on “overelaborate, ill-advised bureaucratism,” but he immediately undermines his own claim by confessing that the proposal is not “likely to have much of an encumbering effect.” The whole theory against “bureaucratic” ideas is that they would create administrative hassles and complicate our current rules. But the concession that the proposal would not burden us to even a minimal extent then contradicts and annuls the crux of the argument.

There is a more fundamental and troubling problem with the bureaucracy line too. Those who espouse it have no problem with obstructing good policies — even ones that, by everyone’s admission, would be very useful — just because they are “bureaucratic” to some degree. They would have no qualms about blocking sound legislation, so long as they may pin their frustrations on “bureaucracy.” The argument becomes even more sinister when considering how broad it is. Its writ is “potentially unconfined, as any new policy is ‘bureaucratic’ insofar as it requires… new procedures, guidelines, etc.” By Hazarasp’s logic, we would need to halt every new-policy vote from here on after. Our register of policies would need to be frozen in time, just to appease the concern that adding or amending a policy would create administrative complications.

Of course, no one who pushes the bureaucracy line really believes in that. It would be preposterous to stop all future policies in the effort to avoid minor administrative snags. In reality, the biggest proponents of the argument have not used it to prevent all new policies from passing. Instead, they use the argument from time to time, “whenever the occasion calls for it, to defeat helpful proposals.” It is an “opportunistic tactic” that editors sometimes employ when they cannot convincingly address the issue on its merits.

Perhaps this is why Hazarasp distances himself from the bureaucracy line by pointing to the bogeyman of “greater and grosser loadstones” downstream. He grants that the deleter role is not “encumbering,” so he must find a way out of his contradiction. The only one he can come up with is a slippery-slope argument. It is the classic claim that one small infringement on the anti-bureaucracy principle will encourage greater transgressions down the line. The specifics matter here. Hazarasp’s ostensible worry is that supporting any bureaucratic policy will “legitimis[e]” other policies in this direction. Thus, this is an “attitude-altering” slippery slope that he claims problematic.

The trick with an attitude-altering slippery slope, however, is that they necessarily “turn[] on the justification the public would infer from the [regulation] and accept as ‘precedent’ for the future.” That is, the slippery-slope argument only works if people generally understand the deleter role to itself endorse bureaucracy as a principle. Hazarasp has not made that showing. Far from it: The deleter role would take so much time-consuming work out of the hands of administrators, while at the same time liberating the mass of ordinary users who could use the deletion tools. In addition, there is little “bureaucracy” involved in the establishment of the role. No new policy will take up space on a project page; there are no forms to fill out, forums to occupy, or technocratic business involved in creating the role. The slippery slope is a figment of Hazarasp’s imagination, because there is just no way that the creation of the role can be understood as a validation of “bureaucracy.”

The bureaucracy line has been in terminal condition from the very beginning. The argument does not make sense, for any principled application of it leads to absurd results. Its proponents have thus had to deal in the business of finding creative workarounds, such as using the doctrine selectively or handwringing about future hypotheticals. But such efforts further enfeeble an already weak argument. They are a recognition that the doctrine is prostrate with fatal problems. Rather than strengthening their position, these last-ditch attempts to find a saving grace convey just how contrived and poorly conceived the argument is. We can no longer allow editors to get away with the bureaucracy line and thus imperil the prospects of a vote’s passage.

Having now exposed the anti-bureaucracy case, the entire edifice of dubious arguments against the vote collapses. A revote is more than warranted in this case to correct a wrongful consensus. The type of error and the soundness of the deleter role both amplify the need for a redo vote. Even entertaining opponents’ animus towards revotes, every single factor that could call for a repeat vote is present in abundance for the deleter role. Our policies and our precedents are further proof that another vote is required. Lastly, the bureaucracy line is a thin reed to support obstruction of the deleter role. Its reasoning has been unconvincing since day one. Now, an already unimpressive argument has become a pale shadow of its former self, as even its supporters have seemed to lose faith in it. Put another way, there are no good excuses to stand in the way of this vote’s passage.

Changes to the deleter-role proposal eliminate any other objections

Opponents last time threw out a bevy of concerns about the proposal. None of them are valid. Nonetheless, those arguments led a large-enough group to oppose the vote and impede its march toward the two-thirds threshold. We could continue to spill ink to describe just how misguided their views are, but the last essay accomplished just that. The reality is that those objections — no matter how erroneous they may be — prevented the success of the previous vote. It is time to “contend with the practical situations of the moment” and reshape the vote to adjust to prior objections.

As a result, this second proposal is not as strong as the first. It stays true to the overarching design, but it incorporates some modifications. In this situation as in others, we could have brooked no refusal and run with the exact same proposal repackaged as a second vote. But just as one can “avoid[] the consequences by denying the principle,” so too is the inverse true: By taking a doctrinaire approach, we would have had to accept a predictable defeat a second time. Principles are not a passe-partout that provide a full or precise answer to every question, and they must be attuned to the circumstances. Plowing on with a failed idea under the theory that we cannot take no for an answer is the wrong way to go.

In revising our original idea, we do what others before us refused to. We have listened to the criticisms of the opposition and adjusted the vote. Compare this to the opposition the last time around, who did not even bother to take responsibility for the mess that they had made. Contrary to their better judgments, they obstructed the vote based on concerns that were never raised prior to its start date. They engaged in voting by ambush in a manner that violates our policies and our precedents. We offered a neutral way to solve the problem, but in response, they plugged their ears and dug in their heels to continue to stonewall the vote. Despite all these wrongs dealt to us by the opposition, we have turned the page and now propose a revote that addresses their objections.

The nomination process

The nomination process was one of the major complaints from opponents of the first vote. Even though their concerns are dubious, the revote is generous enough to attend to them. Under the new proposal, three admins will need to approve any nomination.

This new process will satisfy even the most rigorous of standards. Requiring just one admin to cosign a nomination is an excellent bulwark against improper candidacies. “We regard even one administrator as particularly judicious” because they “got their title only after a supermajority trusted them” with it. Having two admins approve a nomination is an even more ironclad method, as years of practice have proven at Wiktionary:Whitelist. Logically, then, conditioning access to the deleter role on the consent of three sysops will prove to be an impossibly secure approach. The chances for abuse of the system are next to none. The only way for a user to “fall through the cracks” would be to deceive no less than three administrators, each one of whom would have had to show superior judgment in order to become an admin in the first place. To trick one sysop into doing one’s bidding is hard enough; to do so three separate times will be a chimera.

The revocation of the right

Once a user earns their deletion rights, they do not get a free pass to delete all the pages they want. Instead, even the slightest misuse of the right is a ground for its revocation. Any administrator may take the deletion abilities away from them at will, without even the need to hold a discussion. Deleters do not enjoy an absolute right, but one that is conditional by design. This acts as a persistent check on the deletion power and a safeguard against abuse of the right.

Hidden revisions

Another one of the claims championed by many in the opposition is that deleters would be too powerful if they held the ability to hide revisions and view hidden revisions. Again, one of the more suspect objections raised, but the new vote addresses this too. Under the new proposal, deleters will not be able to access hidden revisions.

This should extinguish Surjection’s claims, which many other users signed on to. Although Surjection blindly proclaims that technical problems prevent mixing and matching permissions this way, this worry is based on falsehoods. Our technology allows us to tailor the role to our liking and choose only the permissions that we need. Furthermore, Surjection claims that we should withhold deletion rights from non-admins out of the concern that they may view sensitive information. Under the current system, editors can still stumble upon pages with sensitive information, but they are just powerless to do anything about it. Were the deleter role implemented, those same users could actually delete the pages. Because of the particular apprehension with viewing hidden revisions, we have proposed removing that ability. All in all, there is no more reason to fear that deleters may inappropriately view sensitive information.

With all these measures now put in place, the case against the deleter role has been reduced to rubble. No matter how doubtful the objections were before, we have attended to them anyway. That reflects our commitment to doing everything possible to pass a highly useful proposal. There is no more room for antics from the prior opposition. We have solved every single problem that they had with the first iteration of the vote. Now one can be certain that those who vote against it this second time are doing so maliciously. What more could they want after we solved every single complaint that they raised last time?

* * *

Despite our best efforts to ensure democratic policymaking, the sad reality is that some votes fail unjustly as a result of procedural or ideological failures. One of the major reasons for revotes is to correct those prior mistakes and cast off the wrongful decisions that they bring about. There are some who argue that we should let these errors lie, or at least stick to them for a few months before correction. There are others who demand that we should not pass any proposals if they produce even the least bit of administrative resistance. Do not believe them. For too long, these arguments have succeeded in obstructing useful policies, but we should not put up with them any longer. The deleter role is one of many policies that has fallen victim to the anti-revote and anti-bureaucracy lines, and Wiktionary is worse off for it. Now, seven months after its initial debut, it is long past time that we right that wrong.