Uživatel:Dan Polansky/English Wiktionary cultural problems

The English Wiktionary, as great a project as it is, suffers from certain cultural problems, increasingly more so.

In general, all editors are equal, but some editors are more equal. There are other problems as well.

• Nominally, processes are required to take certain actions and normal editors are chastised for not following them, while certain power holders--administrators--act as they see fit without process.
  • Normal editors delete templates via RFDO process. By contrast, CodeCat/Rua deleted many templates without a process. So did Ben Wing (User Benwing2), e.g. templates cs-decl-noun, cs-decl-noun-auto and cs-pron; these templates should have been kept and deprecated if required so that page histories remain neatly legible. Template deprecation mechanism has been implemented because I insisted on it repeatedly, and has proved its mettle.
  • There is RFM process for template renames. However, Ben Wing is systematically renaming templates like R:PSJC to R:cs:PSJC without using the RFM process, despite there having been considerable opposition in en:Wiktionary:Votes/2019-06/Language code into reference template names, which failed 8-10-3. He does so using what in Czech we call the "salami method", one thin slice at a time; he renamed templates one by one, originally falsely claiming that he aligned the templates with other templates when most templates were aligned in the other direction; after enough time of this unopposed, the alignment statement naturally approached truth closer, although he still did not dare to do the same for the English templates such as R:Webster 1913. Finally, fait accompli gets accomplished, following in the footsteps of CodeCat/Rua.
    • The fait accompli was supported by Theknightwho for R:fr:TLFi template.
    • Ben Wing deleted R:PSJC instead of leaving a redirect, creating unseemly redlinks in revision histories.
    • The problem with the against-majority template moves is that it would seem perfectly fine for anyone who preferred the name without the prefix to move it back to match their preference and then to try to align all templates by removing the language prefix as applicable, and the reason there were no move wars was only because such people abstained from imposing their preference on templates that had the language prefix.

• An administrator who repeatedly addressed vulgar insults to multiple other editors and called them other even if non-vulgar abusive names (civility), accused multiple others of dishonesty instead of nominally assuming good faith (assume good faith), revert-warred beyond 3 reverts during a day (3RR) and presented multiple Wikipedia pages in the joint essay/policy category (BLUDGEON, 3RR) as Wiktionary policies cannot be removed from his administrator role by far; instead, his victims are blamed not only by him but also by his many supporters. The display is disconcerting, evidence of what has unfortunately become of the English Wiktionary. Source: en:Wiktionary:Votes/2023-04/Second vote to desysop Theknightwho.

• Editors find it perfectly normal to post supports or opposes to votes with zero rationale or effectively rationale-free comments. This ought to be forbidden.
  • For example, in en:Wiktionary:Votes/2019-06/Language code into reference template names, none of the supporters originally posted a comment as part of the vote. Under some of these originally empty votes, discussion developed later.
  • In votes, instead of editors being accountable to issues raised about the rationality or conclusiveness of their arguments, they complain that they feel harassed when questions are asked to them.
  • Some editors explicitly stated that they had no duty to provide a rationale. They ought to have such duty.

• Sometimes, alegged consensus appears to be developed off-wiki, perhaps on the Discord server, but we don't know. Thus, when an editor asks about a certain practice, the response is not a link to a Beer parlour discussion or a discussion elsewhere but rather, a curious statement is made: "There is wide, wide, consensus in the community that we should not add dead links. Please do not add them further. I can call other admins if you like. Vininn126 (talk) 18:35, 1 April 2023 (UTC)". In fact, it was Vininn126 who was adding dead links despite reservations on his talk page. Therefore, the consensus must have changed, but we don't know where and how. Source: en:User talk:Vininn126#Dictionary links.

• Theknighwho, Chuck Entz and DCDuring raised concerns with someone doing mass closures of RFV. It was claimed that one month of discussion period is not enough despite process specification. If one month is not enough, the process should be changed to require, say, three months. It was claimed that RFV closer should take effort to find attesting quotations, which does not match my recollection, but if it is so, the process description should say so. While taking effort to cite before closing a RFV as failed is ideal, requiring the closer to take the effort would have a chilling effect on the RFV process administration. The open atmosphere of the project would ideally require that expectations are clearly stated rather than relying on "common sense" of non-practicioners. I have never seen the concern-raising administrators Theknightwho, Chuck Entz and DCDuring close a single RFD or RFV, but I cannot be sure. I find it peculiar for the non-practitioners to complain, not for the first time, about practitioners. RFDs and RFVs are notoriously underadministered. Sources: en:Wiktionary:Beer parlour/2023/February#Disallowing mass closures; en:Wiktionary:Beer parlour/2023/March#RFVE Mass Closures (Again).

• The blocking policy en:WT:BLOCK is virtually empty. The binding policy says:
  • "The block tool should only be used to prevent edits that will, directly or indirectly, hinder or harm the progress of the English Wiktionary.
  • "It should not be used unless less drastic means of stopping these edits are, by the assessment of the blocking administrator, highly unlikely to succeed.
• The non-binding text of the policy is there for confusion; naive users read the page and expect the text to apply, but it does not. Yet, administrators sometimes quote the non-binding text as if it had a policy force.
  • The experience shows that, as a slight hyperbole, any text in Wiktionary namespace that can be misused as policy will once in a white be so misused, regardless of its policy status. The remedy is to eliminate as much text from that space as harmlessly possible.
• Solid remedy: write a real blocking policy that is reasonably detailed and significantly curtails administrative powers. One option is to prevent a block longer than a month of an established editor unless in extreme well specified circumstances, unless resulting from a vote to issue a longer block.
• Quick and cheap remedy: remove all non-policy text from en:WT:BLOCK. That would at least prevent the confusion and misuse of the non-policy part. It was tried in en:Wiktionary:Votes/pl-2015-11/Short blocking policy, which did not pass.
• The sorry state of affairs was brought about by en:Wiktionary:Votes/pl-2010-01/New blocking policy.
  • The discussion is here: en:Wiktionary:Beer parlour/2010/January#Pulling our act together.
  • The discussion mentions in italics the problem that "Wiktionary does not have a friendly atmosphere".
  • In the act of supreme irony, the problem of unfriendly atmosphere (to which I attest) was pseudo-solved by giving the unfriendly administrators even more arbitrary power. Mind boggles.

• Abandoning of the rhyme pages was proposed, to be replaced by categories.
  • I explained multiple times how, for Czech, it leads to information loss of nearly all rhyming content, since the category mechanism would only work if the entries existed, but for Czech rhyming, the pages contain great many inflected forms and non-created lemmas; they often consist mostly of redlinks.
  • Despite this explanation at least two times before, there is again a proposal: en:Wiktionary:Beer parlour/2023/May#Retiring the Rhymes namespace.
    • It does not link to previous relevant discussion, an administrative misconduct. It pretends the arguments against do not exist and have not been collected.
    • By what standard is it allowed that the post creator pings selected editors? Seems improper. Pinged possible supporters: Fenakhay, Mahagaja, Theknightwho, Thadh, Surjection, Equinox.
  • Previous discussions: en:Wiktionary:Beer parlour/2021/August#Retiring Rhymes:, en:Wiktionary:Beer parlour/2021/August#Automatic rhymes, en:Wiktionary:Beer parlour/2013/December#Rhymes categories again, en:Wiktionary:Grease pit/2013/October#Convert rhyme pages into categories?.
  • People previously expressing reservations: Dan Polansky, Krun, Paul G.
  • en:Wiktionary:Rhymes is a starting page; en:Rhymes:German/aχə is an example of German page with some redlinks; en:Rhymes:Czech/anskiː has fairly many redlinks, all lemmas; en:Rhymes:Catalan/ɔnia has some redlinks; en:Rhymes:Danish/akə has some redlinks; en:Rhymes:Czech/atsɪ consists mostly of redlinks.

• When a template previously widely used gets deleted, it makes the revision history hard to read since instead of the template, there is just the template name redlinked.
  • This is a problem for Wiktionary more than Wikipedia since Wiktionary is highly templatized.
  • There is no culture of pushing toward template deprecation instead of deletion. I was the only one repeatedly pushing for deprecation, from what I remember.
  • Deprecation happened e.g. for en:T:term, but it should be used much more widely.
  • When templates are renamed without function change, the old name should be kept as a redirect.
  • Deprecation may involve simplification of the functionality to provide decent even if imperfect output, if it helps break dependencies on other templates that underwent change.

• Entries usually lack verification evidence.
  • For many entries, systematic bot-like effort could be taken to make sure entries have external links to good monolingual sources, but it was not done on a wider scale.
  • People feel no requirement to provide some form of evidence that what they created exists.
  • An economic form of evidence could be introduced, a note to the effect of, attested in Author1 Year1, Author2 Year2, and Author3 Year3. That would alleviate the need of laborious markup.
  • My solution was to either provide a link to a dictionary or indicate in edit summary where the entry is attested.
  • Another solution is the use of template GBK for Google Books search, restricted to the particular language, function available in Czech Wiktionary GBK template; Czech Wiktionary now also has ČNK for a Czech national corpus search. Then, the reader can quickly verify at least that the form exists.

• Editors who systematically enter mistakes are sometimes not effectively dealt with. Blocking them gets controversial.
  • Minimum competence and carefulness is all too often not required.
  • However, some editors received messages to the effect that they should not edit in languages they do not know or have not studied, a contradiction.
  • en:User:Luciferwildcat could not be blocked via a vote, per en:Wiktionary:Votes/2012-07/Blocking of Luciferwildcat; he was blocked thereafter without a vote by user "-sche", who opposed the block in the vote.
  • en:User:Razorflame edited and created entries in many languages, received multiple complaints from other people on his talk page about his mistakes, violated copyright of a Czech dictionary as a result of poor understanding or carelessness as per his talk page message, was blocked multiple times, which resulted in a controversy per en:User talk:Msh210/Archive/Razorflame and en:Wiktionary:Beer parlour/2012/June#Indefinite or long-term blocking of registered well-meaning editors. The block log includes e.g. 'No proof given of "harm to the progress of Wiktionary"' unblock, referring to the dysfunctional and anti-specificationist pseudo-blocking policy that resulted from dismantling the real blocking policy.
  • en:User:Pass a Method/en:User:North Atlanticist Usonian was creating unattested entries again and again, and engaged in other problematic editing. It took a long time and many rather fruitless posts to his user talk page to render him harmless.
  • en:User:Embryomystic entered countless unattested entries, and was blocked only once for very short time for an unrelated thing. There was a time when I made a point of systematically sending his entries to RFV, but I eventually gave up, especially given that I seemed to be the only one to really care about this kind of misbehavior. There is more information at en:Wiktionary:Votes/sy-2019-12/User:Embryomystic for admin.
  • en:User:Rasmusklump showed a pattern of bad editing as per his talk page and was blocked not even once. On the other hand, my method of regulation via user talk page seems to have worked reasonably well.

• The users that seem to be most connected with the problematic culture are at the same time very valuable contributors. Theknightwho and Ben Wing seem remarkable in their contribution to modules, templates and technical aspects; so was CodeCat/Rua. Ben Wing's rollout of automatic Czech declension into Czech noun entries is impressive. Chuck Entz seems to be doing a remarkable job in long-term editing monitoring job, reverting bad edits in considerable volumes.
• In the case of Theknightwho and CodeCat/Rua, template editor rights would seem sufficient. This could be true for Chuck Entz as well; one does not need to have administrator tools to perform systematic reverts.

• All this is consensual in so far as, for some issues, hardly anyone is complaining or raising a concern, and for other issues, votes trying to remedy the issues are failing. It is therefore not a problem limited to individual administrators but rather one with the cultural attitudes of the supermajority of editors who support these administrators or do not dare to speak up. It is hard to see what can be done. Perhaps stewards of the foundation should intervene and start, at a minimum, to enforce the code of conduct against a project that has collectively gone astray and whose administrators fail to enforce elementary decency, provided the code of conduct actually prohibits some of the problematic behaviors described.

Some votes that tried to remedy the issues:

• en:Wiktionary:Votes/sy-2017-11/Desysopping CodeCat aka Rua
• en:Wiktionary:Votes/2023-04/Second vote to desysop Theknightwho
• en:Wiktionary:Votes/2019-09/Replacing de-sysop votes with confirmation votes, as a result of my posts to votes for which I was attacked, posts that raised the issue of it being too hard to desysop administrators. In the light of the votes on CodeCat and Theknightwho, this 2/3-requirement for confirmation would have been the right policy: even with that policy, clearly problematic administrators would not be removed from power.
• en:Wiktionary:Votes/2020-02/De-sysop votes to pass by simple majority. This one passed; at least something. Even so, not a single administrator was ever desysopped for misconduct as a result of a vote in the English Wiktionary as far as I know, a red flag.

• It is probable that admins form something like a caste, an analog of priestly class. Good sourcing is pending.
  • Thus, an admin is less likely to vote to desysop another admin than a non-admin is.
  • In the Czech Wiktionary, there was a group of admins such that when there were desysopping votes, no admin has ever cast a vote to desysop another admin. This is consistent with the hypothesis above. That is not conclusive, being merely a single case, but suggestive.
  • If it is so, the policy of admin confirmation votes requiring 2/3-consensus to pass, equivalent to desysoping votes requiring 1/3+1 to pass, seems advisable to control the power that is out of check. It would be similar to some extent to a policy that admins are elected for one year, not indefinitely.
  • In the English Wiktionary, admins hardly ever block other admins.
  • In the English Wiktionary, admins hardly ever admonish other admins.
  • Adminship that is de facto for lifetime is not a form of absolute power, but is concerning. It is not a good thing.

• As a manner of self-defense, I'd like to make some notes about what is going on in the English Wiktionary about the Czech entries:
  • Ben Wing engages in great activity in expanding inflection in the Czech entries.
  • This may seem to be in time sequence after my block and related to the alegged "obstructionism" of mine.
  • However, I was absent from the project in 2021 and the first half of 2022; at that time, Ben Wing had all the opportunity to do what he is doing now.
  • As a matter of fact, I thought for a long time the Czech noun inflection should be done via modules. I would not have opposed the effort at all.
  • When someone else started to implement cs-IPA as a pronunciation module, I welcomed it. When CodeCat/Rua was trying to nudge editors into entering manual IPA markup in Czech entries by placing user-visible requests there, I was pointing out it was uneconomical since the pronunciation is so regular that it should be done by automatic means, not wasting manual human labor. I favor such automated means.

The foundation is in the process of introducing universal code of conduct:

• Meta:Universal Code of Conduct

If enforced by external agents, perhaps stewards, it could in principle make a difference. The foundation has the opportunity to enforce a much more friendly atmosphere if it so chooses.

I must have seen a Wikimedia Foundation page suggesting that wiki projects such as Wikipedia and Wiktionary should be self-governing. However:

• Either the Universal Code of Conduct applies, or the project is fully self-governing. Both cannot be the case.
• Possible solution: Employees of the foundation should enforce Universal Code of Conduct, especially against administrators not installed by the foundation.
• The measures employed do not need to be draconian; they should not be. There is hardly ever a need for indefinite blocks. One month long blocks including those of transgressing administrators elected by pseudonymous curious-named editors or foundation-driven desysoppings seem to be all that is required, for a start.
• Otherwise, the wiki projects are liable to maintain their own culture of what is acceptable and what is not, at odds with the Universal Code of Conduct.
• As an analogy, if an aerospace corporation says to Federal Aviation Administration (FAA) that compliance with safety regulations is ensured by employee-elected compliance inspectors, FAA is unlikely to be happy.
• If an organization pledges to uphold a code, it is the organization's responsibility to enforce it and to be able to show to external auditors what specific steps and measures it is taking to enforce it.
• In general, enforcement of a code cannot be delegated to those against whom the code is to be enforced.

Having been harassed and hounded by Theknightwho and indefinitely blocked by Ben Wing, I can hardly be neutral. Therefore, the kind reader will carefully investigate my report for bias, inaccurate statements, misleading statements, errors of emphasis and other deficiencies. The conflict of interest I have is obvious. On the other hand, so is the conflict of interest of both Theknightwho and Ben Wing, who love be to able to run free and do as they please without traceable consensus, and have no use for a consensus and process watchdog such as myself.