Requests for Comment/Meta translation reform

The translation process on Meta, as implemented today, is truthfully less than ideal -- Important pages lack full translations in many languages, while trivial pages with low priority for translation have received translation attention in multiple languages. In addition, it must unfortunately be said that there are often translations being made with classic machine translators or due to lack of knowledge of the target language where a modern machine translation servuce such as DeepL or even ChatGPT could have provided a translation of better quality.

We must then ask ourselves whether we want to continue with the current translation process -- missing important page translations, partially-translated pages, and below-standard translations -- when users could simply use performant machine translation services instead and have better quality. Therefore, there are two main solutions to this issue: (if technically possible) integrate machine translation (preferably DeepL) on Meta and have translators simply review and correct output where needed, or keep the current model but only allow certain users to translate.

Because integration of translation services still has unsolved problems, I propose in the meantime we at least strive for better quality translations on Meta.

During the drafting phase a more radical idea, that of eliminating translation was proposed. There will therefore be two overarching options here: 1) eliminate translation on Meta, 2) reform and improve translation on Meta. The reform proposals will be presented first and the last proposal will be the one that calls for elimination.

Proposal 1 (Translator group)

 * Only users who have 'translator' rights are able to translate pages.

Rationale: As part of the following sub-proposals to establish rules for getting/keeping/removing translator rights, this consolidation of translation permissions allows those new criteria to be enforced.

Proposal 1.1 (Criteria for translator group)
In order to be appointed translator, a user must:
 * Translators are appointed at the discretion of Meta administrators. Translators will initially be appointed for a trial period of 15 days. If a Meta administrator is satisfied that they have performed well they will be appointed indefinitely.
 * confirm that they have read MediaWiki's documentation for using the Translate extension and that they understand how to prepare a page for translation;
 * have at least level 3 (Babel) competencies in both the target language and in English; [1]
 * provide a few examples of pages that they intend to work on if approved

[1] While users usually self-certify their Babel competency level, administrators may use their discretion in deciding whether the user actually fulfils that criteria.


 * If a translator has repeatedly translated pages poorly or otherwise misused their rights, their translator permissions may be revoked.

Rationale: These new criteria require that the user... The 15 day trial period allows the meta administrator to confirm all criteria are met and that a request isn't simply hat collection by the user.
 * 1) has the practical knowledge to use the translation tools
 * 2) has the language proficiency to create good-quality translations
 * 3) has a specific project that they plan to pursue

Proposal 2 (Translation administrators)

 * The translation administrator group is eliminated and the 'translationadmin' right is given to translators.

Rationale: Similarly to rollback, this group has limited scope and it's unnecessary to assign 'translationadmin' permissions separately if stricter translator criteria are approved in proposal 5.1. If translators misuse 'translationadmin' permissions, their translation rights will be revoked for misuse of permissions as per proposal 1.2.

Comments (2)

 * If DeeM28's proposal does not pass, while I would not be opposed to giving translation administration permissions to translators, I would like to see the translation administrator group revamped to include the  user right, to assisting administrators in removing unmaintained pages. It would still be a discretionary user group, but would have a higher level of competency and would shift from marking updated translation pages to removing unmaintained translation subpages. Dmehus (talk) 16:28, 21 May 2023 (UTC)

Proposal 3 (Pages excluded from translation)
Examples are: pages in the Tech namespace, pages that contain lists of users, guides for certain advanced user groups.
 * Pages which only present an interest to a limited/select group of users or which are very short should not be translated.

Proposal 4 (High traffic pages & Review) (added by DeeM28)

 * In order to ensure professional translation on Meta, only high traffic pages may be translated.
 * Administrators will maintain a list of high traffic pages that may be translated. Examples of such pages include: the Main Page, global policy pages, other pages of importance.
 * Translations may only be done if there are at least two users who have translator rights who are able to translate into the specific language.

Note: This proposal is subject to the translator group proposals (Proposal 1, Proposal 1.1) passing.

Comments (4)

 * I personally would just like to see translation suspended and eliminated, per your well-articulated rationale in Proposal 8 and my comments on the outmoded Translate extension and its sub-optimal design as well as the evolution of AI and universal translation tools rendering the necessity for human translations by a very small, and infrequently active, core group of translating volunteers obsolete. For relatively stable and high-traffic global policy and the main page, I can see a weaker case to retaining them. I wouldn't be so keen on restricting approval to certain languages, though, if someone maintains them. Administrators can already delete unmaintained translation subpages, and this was something I tried to do regularly as a Meta administrator. It's just not being done regularly. Dmehus (talk) 16:25, 21 May 2023 (UTC)
 * My main issue with this is that I'm not sure what's seen as "all relevant pages". Personally I think under this proposal it would be okay to only have the Main Page translated for example. Though if there's a narrow definition of all relevant pages that could be interesting. Reception123 (talk) ( C ) 17:31, 21 May 2023 (UTC)
 * Thank you for your comments, Reception123. I wouldn't be opposed to keeping the Translate extension, and to repurposing the moribund  user group that includes the ability to mark pages for translation, but I'd still like to perhaps repurpose the   group to include additional janitorial/mopping permissions, so long as we have additional criteria for granting and clear but also flexible criteria for revoking (i.e., edit-warring, recidivism following administrator warnings or guidance, etc.), to include the ability to remove unmaintained or no longer maintained translations. If we do limit to specific pages, I would think Miraheze, Help center, and most of the key global policies would be good to have translated. I don't think we necessarily need to have translations of all the Meta user group pages, particularly as ChatGPT and future AI models (i.e., Google's Gemini) continue to evolve. Basically, any page that is not frequently edited. We may also want to consider having a new page protection level, associated to that revamped patroller group. Perhaps revamp Patrollers as  Janitors? Dmehus (talk) 17:45, 21 May 2023 (UTC)
 * Thank you for your comments. I have taken them into account and have a new proposal that allows translation for important pages but still contains the approval requirement. The reason is that without the approval requirement translations can not be fully professional because a real risk persists of users overestimating their translation capabilities and producing poorly translated pages without check as administrators evidently are not able to properly review translations into languages which are foreign to them. Requiring two translators for one language will mean that one will be able to review the other. In regard to your new Meta Janitors group proposal I believe that is an idea that ought to be considered but it might be best not to have too many things decided in a single RfC and to first decisively decide the issue of translation in a direction or another. I invite you to make changes to my proposal if you believe they are useful and also to propose a different one altogether if you are unsatisfied with my idea. --DeeM28 (talk) 14:42, 22 May 2023 (UTC)
 * I do not wish to presume that the lack of response signifies agreement but I have deleted my former proposal text and only kept the current one. I hope it is acceptable and only would require minor changes. I strongly encourage other users to also comment on this as I believe it would be a significant change that requires wide support. --DeeM28 (talk) 13:36, 25 May 2023 (UTC)
 * I hope you don't mind but I've made some changes in order for the proposal to be more interlinked with the original translator group idea and for an additional language approval to no longer be required as long as there's two confirmed translators. In addition, I've also separated out the removal part into a separate proposal to not make this one dependent on that as well. I hope that's acceptable. Reception123 (talk) ( C ) 15:13, 27 May 2023 (UTC)

Proposal 4.1 (Removal of current translations)

 * Any current translations that are incomplete or otherwise poorly done will be removed. This includes incomplete or poorly done translations of high-traffic pages.

Proposal 5 (Elimination of translation) (added by DeeM28)
This proposal is bold and may even be considered by some as "radical" which is why I should explain in a more detailed fashion. I believe people generally on Meta are afraid of saying quiet things out loud, of attempting to change significantly a certain function that has been done for a long time or to take an important and decisive decision. The proposer above identifies the issues related to translation and proposes some methods in order to fix these issues. I can agree that the issues may be improved by those measures but I do not think that it will be significant. I would like to say that this proposal in no way has anything against multilingualism or the fact that Miraheze is clearly a global platform. I think it is necessary to eliminate translation is because it is very likely that it has a very minimal usefulness due to its incomplete and imperfect nature. With the rise of powerful translation websites and extensions/add-ons for browsers it cannot be said that users who do not speak English are not able to very easily have pages translated. If we were able to have all pages translated in a professional, coherent and full manner I would be glad to be able to continue but the issue on Meta is that pages are often only partially translated or not translated at all or translated poorly or some pages have translations and some do not. I do not believe it is possible at this time for Miraheze to properly fix this issue. Translation of all pages is a feature that is likely inspired from Wikimedia's Meta wiki but is unsuitable for Miraheze Meta in my view. I understand that this proposal is a long shot and is likely to be opposed by the translator aficionados here on Meta but I do not see a reason for why it should not be presented as an option in case someone finds favor with this idea. I of course propose to leave pages that are already translated up so that translators time has not been wasted out of respect for their work. I also take this opportunity - in case anyone has time to fully read my text - to enquire to system administrators whether they are able to provide statistics regarding how many people actually use the translated pages daily. --DeeM28 (talk) 14:46, 21 May 2023 (UTC)


 * Translation on Meta is generally suspended for an indefinite period of time.
 * Pages that are fully (100%) translated in a language and are done so in a professional/rigorous manner will remain and may be updated as needed. All other translations will be removed.

Comments (5)

 * I would like to give my vote before RfC starts. I am absolutely against this proposal; I am a person living in Turkey, I know English at en-3 level. For both other users and myself, it would be logical to continue reading and translating each page in the native language of the person. For your information Hey Türkiye  Message? 18:00, 21 May 2023 (UTC)
 * In addition; I can't let my translations go to waste. There is a labour; there is a work done. You have spent hours and minutes for this translation. Reading English does not always benefit people; there are people who do not speak English. That's why we do what we do; translation in accordance with that language. I do the same. I translate the text into my own language to understand it better, then I edit it if it contains a loss of meaning - otherwise the text cannot be read. One thing is that there are many users here (including me) whose first language is not English. So I wanted to use my strongest opposition to defend our rights, and I wanted to make it clear in this comment why I voted the way I did. As soon as the RfC starts, my vote stands - I will not edit the proposal even if there is no edit in it. I respectfully announce. Hey Türkiye  Message? 18:13, 21 May 2023 (UTC)
 * A bit bold indeed, but I was actually thinking the same thing here. For one thing, the Translate extension is showing its age. Moreover, rapidly evolving AI models and deep integration within search look set to render English Wikipedia and even the multilingual Wikipedias redundant. In short, Wikimedia Foundation looks set to be facing an existential crisis in the next several years, as it would not surprise me to see English Wikipedia's traffic plummet, as Google gradually begins replacing English Wikipedia infoboxes in search results pages with continuously improving AI-generally supported search results. Nearly all users on Meta Wiki have some degree of English language knowledge, and AI, as the proposer notes, continues to evolve and get better and better, rendering the need for translations obsolete. It also artificially increases edit counts and page counts, on pages which are edited too frequently to be translated by a very few number of volunteers. Dmehus (talk) 16:19, 21 May 2023 (UTC)

Comments made during drafting phase (Removed proposal)

 * This seems like the inverse of Proposal 8.1, which I would tend to prefer. Rather than have a list of excluded pages, we should maintain a list of included pages for translation. Requests to have pages added to that included list should be made at the applicable venue, Administrators' noticeboard or Stewards' noticeboard, as applicable, and must include a compelling rationale. So rather, perhaps we could revamp Proposal 7 with the aims of Proposal 8.1, rather than have risk having proposal and counter-proposal bloat? Note, most pages in the  namespace are already not translated, though there may be occasional non-technical guideline that could be useful and, in this way, such a page could be added to the list of included pages, if there's a compelling rationale. Dmehus (talk) 18:23, 21 May 2023 (UTC)