ArchWiki talk:Translation Team

Add mention to translating Table of Contents

There is no mention of intructions for translating Table of contents pages. Personally, it took me a while to know it was available for translation after reading ArchWiki:Bots#Table of contents. How about adding instructions in here? -- Josephgbr (talk) 10:41, 19 June 2020 (UTC)

Agree. -- Blackteahamburger (talk) 11:25, 19 June 2020 (UTC)

In case further development of the bot involves upgrading the entry point, keeping all its ArchWiki documentation in the same page may ease keeping it in sync, I suggest just linking to ArchWiki:Bots#Table of contents as by the way we also wanted to do in Help talk:I18n#Checklist to add a new language. -- Kynikos (talk) 14:10, 20 June 2020 (UTC)

Regarding Translating Wiki Pages

Hello, I want to translate Archwiki pages into Bangla. But, wiki tells to inform translation team. Bangla language has no translation team. Now, whom should I inform? There is no interlanguage link, too. And also, how do I translate pages like Installation Guide, Arch Linux etc.? Would anybody please help me with these matters? I am new in ArchWiki translating.

—This unsigned comment is by FOSS ভক্ত (talk) 16:17, 2 September 2020‎. Please sign your posts with ~~~~!

The correct place for discussion about how to add a new language is Help talk:I18n#Add Bangla Language. – Lahwaacz (talk) 17:49, 2 September 2020 (UTC)

Swedish translations

Hi, I was told I should come here and talk about the state of Swedish ArchWiki translations.

As of right now there's only 2 translated pages, one of which is Main page (Svenska), which literally just excuses itself for being empty and nothing more. The second page is Installation guide (Svenska), which I took to fully re-doing as it was severely outdated, missing pieces and was seemingly written by someone who wasn't a native speaker.

After doing this page I'm quite exhausted and probably won't do much translating in the near future, at least not such a big project.

Either way I thought it'd be good to share my view on this and possibly get attention from others wanting to do Swedish translations! DerpishCat (talk) 19:28, 25 March 2021 (UTC)

Lack of best practices documentations

There are a lot of practices for translation that a lot of newbies are not aware of, such as the use of Template:Translateme. In fact, a lot of points are unclear. Information is also scattered everywhere, including ArchWiki:Translation Team, Help:i18n, Help_talk:i18n, ArchWiki:Contributing, Help:Editing, and Help:Style. As we know General recommendations exists, we really really need a Translation recommendations.

• Things like Help:I18n#Localized_redirects is utterly useless since the reader is required to read excessive articles to know how to create a redirection page. We can point them to a "centre" (like "Read here for more info").
• The fact is, Help:i18n is incomplete in my opinion. Contributors still need to read Help_talk:i18n to get more info.
• It is unclear whenever (partially) machine-translations are accepted.

So, I think merging those points into a (new) page might be good. — windowsboy111 (talk) 01:59, 2 May 2021 (UTC)

ArchWiki:Contributing, Help:Editing, and Help:Style have a broader scope than only translation. Not only translators edit and contribute to ArchWiki.

I agree, that Help:i18n does not look complete enough. It must be a complete guide for translators, the only place where they seek advice. We can merge critical topics from Help talk:i18n into Help:i18n, as well as the majority of ArchWiki:Translation Team sections, as #Create a new page and its translation and the sections about template usage.

After all, I see ArchWiki:Translation Team as a reference page which contains 1) links to important for translators pages (Help:i18n etc.); 2) technical issues like #Page list.

This allows to avoid both unnesessary splitting and duplicating content. -- Duodai (talk) 09:10, 7 May 2021 (UTC)

Interlanguage links are not visible anymore in the preview

"5. Preview the page with the new interlanguage link. / 6. Visit the interlanguage link you have just created" — After recent MediaWiki upgrades, this is no longer possible. I guess there is a bug somewhere in the vector-2022 skin (the vector skin works fine). (If anyone cares, it does not work in the monobook skin either.) -- andreymal (talk) 10:50, 18 July 2022 (UTC)

To me they still appear during the preview with vector-2022… but on the left, at the bottom of the main menu, where they were before the update of MediaWiki. --Erus Iluvatar (talk) 11:43, 18 July 2022 (UTC)

It works with vector, but with vector-2022 it is just empty. https://i.imgur.com/rmfAQf4.png (Btw, why are there two almost identical skins?) -- andreymal (talk) 11:55, 18 July 2022 (UTC)

Here's what it looks like on my side (new Vector on the left, old one on the right: https://imgur.com/a/e6ljapB) but maybe it's something to do with me using a custom CSS? The vector-2022 version is linked to MediaWiki's mw:Reading/Web/Desktop Improvements from what I understand. --Erus Iluvatar (talk) 12:13, 18 July 2022 (UTC)

Is this still an issue with the recent MediaWiki update? --Erus Iluvatar (talk) 20:07, 22 December 2022 (UTC)

Yes, I still don't see them in vector-2022 https://i.imgur.com/GMYF3yV.png -- andreymal (talk) 20:14, 22 December 2022 (UTC)

Welp, I'm out of ideas, I still see them on the left… The proposed switch to Timeless looks more enticing every update. --Erus Iluvatar (talk) 20:28, 22 December 2022 (UTC)

Add note for long pages

Since some translations can too long to do in one sitting, we should make clear in the rules, e.g. with a Template:Note, that it is possible to create a page in the User namespace to translate slowly before moving it in the Main namespace. It should also be explicitly noted that in that situation, unless the original text is being updated in the translation in real time, the Template:TranslationStatus should use the revision number from the original page when the copy before translation was made, to avoid missing updates to the original page in the translation.

See the draft below.

--Erus Iluvatar (talk) 12:47, 11 December 2022 (UTC)

It should also be possible to create a partial translation in the main namespace so that multiple people can work on it until it is fully translated. — Lahwaacz (talk) 19:46, 11 December 2022 (UTC)

Coordinated translation effort can also be done outside of the Main namespace, as it would avoids the need for intervention when the translation is not finished after years. See the history of KDE (한국어), General recommendations (Українська), Comparison of tiling window managers (Português): some page translations just stall. Would it be preferable to have the partial translation as a sub-page of the ArchWiki:Translation Team for the language? --Erus Iluvatar (talk) 20:53, 11 December 2022 (UTC)

See the related discussion at ArchWiki talk:Maintenance Team#About Partial translations. --Erus Iluvatar (talk) 07:00, 13 December 2022 (UTC)

Template:Translateme is created for this purpose. It is wiki after all, not a book. ArchWiki grows to such state because: Great User collaboration: Users in different parts of the world can open the same document at the same time making a collaboration of shared data much easier. Can we still call it a wiki if we only show fully translated wiki?

I give my -1 here because it make ArchWiki harder to collaborate. --Fengchao (talk) 03:58, 15 December 2022 (UTC)

As previous discussed, I am always against move any page into a Users page. It has below disadvantage:

1. It is hard to find, right now I only find some of them though a search and back links to some template. How could a user find stall translations or stall new pages to collaborate?
2. Even whan a new contributor find the page, the ownership is still not clear. User's page is usually owned by that person, like something of a personal code branch. Many translator will think like: "
3. It is hard to determine how to moving forward after the page in main is done by another person without finding this draft, delete it or archive it or just leaving it there?

--Fengchao (talk) 01:14, 16 December 2022 (UTC)

"Can we still call it a wiki if we only show fully translated wiki?" I sure hope so! Once again, a collaborative work does not require the end result to stay partially unfinished.

The solution for you issues on the pages in user space could be my proposal to use a subpage of the translation team's page.

--Erus Iluvatar (talk) 09:19, 16 December 2022 (UTC)

Draft

Note:

For long pages (e.g. QEMU, VirtualBox, ZFS), do not create a copy of the English page in the translation to work at it progressively, instead:

• Create the copy in you own user space, so that the finished page can be directly put in the Main namespace by using the Move button.
• Fill Template:TranslationStatus with the revision number of the English page at the start of your translation, so that any change made on the English page during the translation will not be forgotten when updating the page afterwards.

Address explicitly the fact that creating new pages should be secondary to maintaining existing ones

As seen in the French, Portuguese or Chinese translations, some contributors prefer creating new translations rather then maintain the existing one.

This leads to situations like Cursor themes (简体中文) where the page has been untouched for 8 years, deemed enough out of sync from the English content and redirected to it to avoid giving erroneous advice to readers, only for a new translation to be made from scratch 6 months afterwards.

These back and forth are a waste of everyone's time and could be avoided if we warned translators at the top of ArchWiki:Translation Team#Create a new page and its translation that updating pages, in particular those flagged with Template:Bad translation or Template:Translateme, is preferred to the creation of a translation for a page which does not already have one.

--Erus Iluvatar (talk) 12:58, 11 December 2022 (UTC)

Above requirement assume a user will always show up and keep having interesting for a page after years. We could only put such requirement for maintainers, not a normal contributor.

A large percent of translators are one time contributor. For example today I buy a new Intel notebook, I read Intel Graphics and translated them into Chinese. Then monthes later my notebook is stolen and I will never read Intel Graphics and touch the translation again.

Not like French teams that has only one major contributors, Chinese team have 40 conributors list their names in the list. The team is too dynamic to give such requirement, it effectively means:

1. If you only have time to translate some sections, do not ever start.
2. If you are not sure you will show up again to maintain the page for a long time, do not ever start.
3. If you are most intrest to A but at the same time there are out-of-date page B,C, do not contribute to A but to B & C.

ArchWiki is a wiki, it will have a character of a wiki: Contributors come and go. Contributors are free or busy. They are free to contribute as long it is for good of other readers.

Below is what we should promote. From my point of view, it is the sprit and vitality of a wiki:

1. Start contribute right away, if you only have 5 minutes, it's great that you can just translate the installation section. Other contributors will start from what you left and improve the wiki step by step, section by section.
2. Start from what you are most familiar and most intresting one. Don't worry if you may not have time to maintain it later. Someone with same intresting with you will take it.

--Fengchao (talk) 05:08, 15 December 2022 (UTC)

You're right that it should not stop the contribution of one time translators, my initial take on this is clearly too one sided. However I still think you're too optimistic on the fact that eventually someone will continue the work. For high interest page it will probably be true, for most pages it means they stay stale for years. That is something I would like to help other translators to grasp, and pointing out the fact that some maintenance is required to end up with a coherent whole is a minimum in my eyes. Keeping things as they are now is clearly not working well enough. --Erus Iluvatar (talk) 07:04, 15 December 2022 (UTC)

Yes, some pages may stay there for years because no one is interest. It should be fix by attracting new contributors. I am not optimistic about keeping the page a perfect translation. I am indeed optimistic that very few out of date pages could harm our reader. There are an very active chat channel for Chinese Arch users and they guide users to the wiki page whenever possible. If someone are bitten by a misleading/out-of-date translation, the team will fix them quickly. --Fengchao (talk) 13:51, 15 December 2022 (UTC)

I don't see how leaving pages to rot will attract contributions from new translators, instead of trying to promote quality inside the existing teams. There is certainly a happy medium between waiting a decade to update a page and expecting translators to update a page in less than a week. --Erus Iluvatar (talk) 14:36, 15 December 2022 (UTC)

See here and here for KDE junior jobs. Usually such jobs are so easy that a developer could just fix by one line. It even takes more time for them to mark the bug as junior job instead of fix them right away.

You miss understand "leaving" here. It is not leaving the page there roting, it is just a re-allocate of contributors time. Everyone has limited time, they do not sit there do nothing. The time saved in fix a rarely read article could be used to translate a new article, or fix something in English page. With limited time we should focus on what a user need. Our chat channel is something like a user's survey. If an area questions, a contributor will put more time on it. But it does not means we should just delete untouched area. It is a long tail.

Let us use kernel as another area, there exist many arch/drivers rot, no one touch them, no one even have a hardware to test them. But they are siting their to serve a tiny users. Such code usually stay their for years before they are cleaned up or drop when Linus Torvalds are somehow sure no one in the world still need new kernel for the hardware. Whe Linus do not just fix them? Time limit so he must find and attract a large team to maintain it.

--Fengchao (talk) 23:52, 15 December 2022 (UTC)

As answered on our other exchange, the KDE example is not a good one, since we do not point new contributors to easy pages as far as I know.

The Linux kernel has been criticized loudly for removing support for "only" decade old hardware, but it was justified since said hardware was really not majorly used or popular. There are even instance of code spending less than a decade in-kernel. The rule of thumb mostly is "are there still developers with the hardware" and not "are there still any users with it". A quick search for "removal from linux kernel" yields thousands of results.

As answered in our other exchange, I would not be opposed with keeping the existing status quo where maintainers are allowed to redirect (and not delete, as discussed) unfinished translations after two years. This way, the limited time of the translation team is picked by the maintenance team when pages have visibly been abandoned. --Erus Iluvatar (talk) 07:05, 16 December 2022 (UTC)

What is abandoned? A wiki do not have ownership. How could you defined abandoned? Whether a page should be kept or not should be judged by the content, not edit time. If the content is still useful to some user, it should be kept, no matter how long it stay there with no change. If some part are out of date, just delete out of date sections. After discussion so much, I feel your are just finding some ways to delete. But I have to say again that you only have right to delete contributions you understand. No one have right to delete contributions/sections without fully understand the content. We used to have rule that a page should flag for deletion/arching for at least 2 weeks. That is for content based deletion. Even when a maintainer are sure that the content is useless/out of date. It need to review still. Delete just by out of date time or last edit time: No. Delete by understand the deleted text with some review time: Yes.

--Fengchao (talk) 13:43, 16 December 2022 (UTC)

A translation not finished in 10 years I consider obviously abandoned.

Ok, if you use this time as a criteria, I am fine. But I check the history same pages are remove because it is only out of date for 1~2 years. That is what I am against. --Fengchao (talk) 03:18, 17 December 2022 (UTC)

I will repeat here my apology for the pages where I made a bad judgment. I do not have time right now to re-reck every single one of them but I hope they are far from the majority of the redirections I made.

Now that we've established there can be a reasonable time frame to expect for finishing a translation or updating an out of date page, we just have to find the exact number: while I took 10 years as an example for undisputable redirection, I hope you do not see that number as the right amount of time to wait in each case.

Right now unfinished translations are explicitly given a 2 year time frame, while out of date translations are written in a way that invites the maintainer to check not only the time frame (6 months as a bad translation) but also if the page is majorly desync-ed or not.

What would you change the current guidelines to in regards to the amount of time left to a translation team in each case?

--Erus Iluvatar (talk) 07:18, 17 December 2022 (UTC)

As I already mentioned on other discussions, we should only take action based on the content, not by time.

There is no ownership of wiki pages, if a page is only out of date on some small parts, it should be kept there. Ten years no updating is fine here. If a page contains wrong/misleading content, just mark them as bad translation is Ok.--Fengchao (talk) 13:53, 5 January 2023 (UTC)

As answered below, it is not a matter of ownership. I will stop wasting your time and mine on this subject since we clearly have incompatible views on it (but I will point out that you are not giving me consistent answers), until other translation teams have more on the topic. Thank you for exchanging with me. --Erus Iluvatar (talk) 15:18, 5 January 2023 (UTC)

It is not a matter of ownership. After our discussion, I feel you are just clinging up to every scrap of translated content regardless of its value to the readers. It's like spreading butter too thin, better have a smaller bread slice but have it fully covered, than a huge one with parts of it without substance.

The rule you are talking about has not changed in how it works: content to remove should be flagged for a week, so that people have time to discuss if needed.

Right now translations have to be flagged for months without evolving before a maintainer can touch them. When a translation has been flagged for update for a full year or has been unfinished for two years, I think we can safely say no one is going to take care of the page any time soon and it's better for the readers to be redirected to up to date information.

--Erus Iluvatar (talk) 14:22, 16 December 2022 (UTC)

To update the guidelines, we need each translation teams input. For Chinese team we will keep the page as long as it is still useful to end user. If some sections are missleading oro too out of sync, we just change those sections into English if no one have time/interest to update translation. But that is un-related anymore becasue Chinese wikis are moved to a seperated location. --Fengchao (talk) 01:23, 18 December 2022 (UTC)

Good, so since your translation team is no longer affected we can be keeping the existing rule in place while we wait for the input of other translators. --Erus Iluvatar (talk) 06:49, 18 December 2022 (UTC)

I add a note into the create translation section. We could encourage maintain existing page first, but should not forbid new page creation just because there are out of date pages. --Fengchao (talk) 08:02, 23 January 2023 (UTC)

Thanks ! It's probably a good step to bring awareness to this topic :) (as answered 2022-12-15, "You're right that it should not stop the contribution of one time translators…") --Erus Iluvatar (talk) 08:08, 23 January 2023 (UTC)

Change from Template:Translateme to Template:Bad translation

I re-read ArchWiki:Translation Team#Dealing with old translations carefully, and I think most of the dispute comes step:

Change Template:Translateme to Template:Bad translation if the translation has become critically outdated.

I found two examples are not critically outdated from my opinion:

• Example 1: zh-hans Laptop page. While the translation does have its defects, they are not critical and it should be marked by Translateme instead.
• Example 2: Help:i18n (Español). This page mainly contains two parts: Guidelines sections is very useful to every Español readers. Languages section is mainly about how to create a new language/other interlanguages links that few Español readers should ever care. It is now flaged by Bad Translation and if no Español users wants to waste time on updating the Languages section in one or two years, it will be redirect to English page. We delete important up to date part because of a less important out of date part.

At the same time, Environment variables (Español) is indeed critically outdated because it still contains non functional ~/.pam_environment.

With current process, Template:Bad Translation is like Template:Deletion flag before. So we should be more careful and only change TranslateMe to BadTranslation when the page is indeed critically outdated. Or is it too subjective here? --Fengchao (talk) 13:37, 4 January 2023 (UTC)

Edit: Another related discussion for Arch Linux (Ελληνικά) here

I'll leave the Greek and Spanish translation team handle the respective pages where you have questions for them. Since I am the one who flagged the first two examples you picked, I will re-iterate my previous apologies if I made a bad judgement on some pages, but:

• if the words "Bad translation" have been misunderstood on the Laptop page, I'm glad to see they have produced the intended effect in the end, see Special:Diff/757215 and Special:PermanentLink/757246#Thanks !
• the potential redirection of the whole Help:i18n (Español) page while only a single section is flagged (even if it's with Template:Bad translation) is a baseless speculation: unless I've majorly messed up, no page were redirected when only a single section was flagged (with either Template:Translateme or Template:Bad translation).

I'd be happy to review together some pages that have not been updated in years, so that we can discuss of an objective set of criteria where we can clearly delineate the separation between a page needing a refresh in translation (i.e. Template:Translateme) and a page where its value is dubious to the user (i.e. Template:Bad translation).

In the meantime, I hope we can leave the judgement on a case by case basis to the person flagging the page (using these templates on translations is not a translation team only affair, as far as I know).

P.S.: You still have not answered me on a branch of our discussion above as to when you consider that a page can be called abandoned, nor on ArchWiki talk:Maintenance Team#Procee update for bad translations. We have pages that harbor the Template:Bad translation for years or stay unfinished for years, I think we can do better.

--Erus Iluvatar (talk) 17:52, 4 January 2023 (UTC)

What is wrong is not the "Bad translation" word its self. The translator update the page in the fear that if he don't, what he/the team contributed will vanish in a year. You are glad here but the translator had a really bad feeling. They complain ArchWiki value more on Perfectionist insead of usefulness. And this is just one reason. If the page is not wrongly flag as Bad Translation, the contributor may spend time on what he think is more useful. Follow your logic, I could add Archive flag on most out of date pages, and I am sure someone will fear their contributions will vanish and put more time on editing pages. Such logic is totally wrong. --Fengchao (talk) 14:15, 5 January 2023 (UTC)

I'm terribly sorry that pointing out their mistakes made someone feel bad, but I can not simply ignore all the contributions who need an improvement and where I am not able to fix them myself just because it might hurt someone's feelings in the process. I'm sad that I worded badly enough my remarks such that they were received as a personal attack, while I'm actively trying to simply state the factual errors. How should I flag a page where a translator removes the Template:Bad translation and updates the Template:TranslationStatus stating the page is fully translated up to the revision from 2022-11-17 from the English page when it is not the case? Template:Translateme? Which will promptly be ignored for years? Do you have suggestions on how I could improve my requests in the future?

There is to me a whole palette of nuances between doing a perfect translation and doing a bad job, but I feel like you don't see things that way and see what I call doing things right as being a perfectionist. There is a saying in French that roughly translates to "What deserves to be done, deserves to be done well". A partially updated (i.e. which states it is up to date with the English page while some sections are missing) translation is a disservice to our non-English readers in my opinion.

Template:Translateme is the my answer. Ignored for years is perfectly OK if it is not critically out of dated and do not contain mislead info. And I also do not want to waste my/your time on how to view a 90% translated page. For me it is not perfect because 10% of it does not align with English page but the 90% translated part is valuable to its reader. And for you, the 10% ignored part is a disservice. Let us just follow current rule, and only mark a Bad Translation when it is critically outdated. --Fengchao (talk) 12:33, 6 January 2023 (UTC)

For the translation of Laptop, the fact that a whole addition to a section was missing in the translation has no incidence for the reader since it's not making the page "critically outdated", if I'm reading you right? I'll agree to disagree and will restrict myself to Template:Translateme until we can find a common ground to define "critically outdated". I will ask again, do you have any advice on how to invite a translation team to finish or improve a page, without it being badly received?

--Erus Iluvatar (talk) 14:21, 6 January 2023 (UTC)

Regarding flagging mechanically pages for archival, you are (once again) putting words in my mouth. The more probable course of action would be that the user flagging page mechanically would receive a gentle warning on their talk page to cease their unhelpful behavior, if most of the flagged pages do not deserve to be archived. Depending on the reaction from the contributor, a revert from all their recent edits could be done instead of wasting everyone's time to evaluate on a page by page basis if the content is really outdated enough to deserve archival.

Nowhere in this thought experiment does anyone become coerced to contribute in any form. I'm not sure I'm following your logic.

--Erus Iluvatar (talk) 15:16, 5 January 2023 (UTC)

From my point of view, a admin's job is to refining a contribution, if there is something wrong, if it is not perfect, improve it or mark it. But any valuable part of it should be kept even if it is only 10% of the contribution. Revert a contribution because 90% of it is bad is wrong. Once a contributor save its changes, the 10% useful part become the Property of Arch wiki community. No one, including a admin, have right to revert such edits without refining the useful part out. --Fengchao (talk) 12:33, 6 January 2023 (UTC)

You have also been removing parts of a translation when you updated an out of date page. Please teach me: how do you define when a contribution is valuable or not, so that I avoid mistakes on this subject in the future?

Edit: It looks like there is no consensus on this subject, see Special:Diff/455394/455395.

--Erus Iluvatar (talk) 14:21, 6 January 2023 (UTC)

For anyone who do not bother clicking on my removing parts of a translation, I summarize here: I sync the whole page with English. During the sync, I changed thousands of words and below 6 translated words are removed by my action:

1. 安装 -> Installation 2. 配置 -> Configuration 3.提示和技巧 -> Tips and Tricks, 4. 使用 ALSA -> Using ALSA 5. 参阅 -> See Also, 6. xbmc中文插件库安装方法 -> xbmc Chinese plug-in library installation method

And I admit here that there is something wrong with my removing of above 6 words. I should translate them back if I have time. Hope you can give other redircted translations the same attention as above 6 translated words/Sentences in the future.

--Fengchao (talk) 23:29, 6 January 2023 (UTC)

To be clear, I'm not blaming you for it: I'm probably the first here to make mistakes when editing to fast. For the full information of someone following this exchange, the linked page is a fully untranslated copy of the Kodi page with the Chinese suffix, where the only edits to the content have been in 2019 to copy/paste the whole text from English (the linked diff) and the page creation in 2015 which only translated the section titles.

I had just re-read our previous exchange on the maintenance side where this page had popped up, which is why I used it as an example of situations where you too had removed some translated content when performing a maintenance task. What I'm trying to convey is that you're explaining that keeping pages with partial or out of date translations is a "good compromise" and your main argument is that we should not throw out content indiscriminately, but that every page should be scrutinized before acting on them, while I'm advocating for simply opening the English page and the translation side by side: if they visibly differ (and there's no need to speak the translated language there) in their structure, that's something that needs to be addressed either by updating the translation if it's a few details missing (i.e. less than 20-50% of the page, depending on the context), or if the whole page is too desync-ed a redirect to the English page is probably the right move. You have explained regularly that our time for contributions is limited: the latter approach is faster and leaves less room for errors (like the CDemu page where the translation still contains things that were removed from English in 2009).

I'm still open to discussing a general definition of "critically out of date".

--Erus Iluvatar (talk) 06:47, 7 January 2023 (UTC)

A page is critically out of date when readers will not get the package installed and configured. Missing a section or several sections is fine. But if it contains misleading or wrong informtion, it is critically out of date. It should be judged by how many people will be served harmed if the page is kept there. So it depends on readers segmentation. And that is why Chinese team select to keep a page even when half or more parts are just English. 50% Chinese readers in Section 2 readers will benifit and a redirect is just means nothing. And that is why French team select to just redirect it because French readers could just read the English page. --Fengchao (talk) 14:17, 7 January 2023 (UTC)

Yes, it is much clearer to me now why you want to keep translated content seemingly at all cost. I will say again that I'm sorry if I redirected pages that were useful to your readers.

I still think partial translations are not as beneficial as you make them to be, but you changed my mind for old fully translated pages: the can be kept when they do not provide wrong/misleading (as opposed to "benign" out of date) information, though we should probably find a way to clearly alert the reader when the translation is more than 2 year old (in my experience it is a common threshold on many pages where information stops being relevant, in particular for the Troubleshooting/Know issues sections). I hope I exercised enough judgement when I went on a "cleaning" spree on old translations and not many (ideally not even one) page were redirected when they were fully translated, simply a little out of date but still visibly useful.

How can we improve the situation about partially translated pages though (especially those where less than 10% of the page has been translated)? There is probably some kinds of per-team structure that can be set up so that they do not become common, though I agree with your previous comment that we should not force anyone to contribute in a way they do not want to (and I want to publicly apologies for the pages where I could be read as bossing others around on updating them). See the state of hardware pages before we got Help:Laptop page guidelines, copy/pasted old cruft accumulated on most of them with a sprinkle of lspci/lsusb dumps: I believe we should lead by example, and try to get as many partial translations to a complete state as we can.

We also have the issue of non-responsive translation teams (e.g. the Italian pages were once numerous, but no one seems to remain nowadays to update them).

This brings me back to my proposal at ArchWiki talk:Maintenance Team#Procee update for bad translations for creating Template:Keep translation, that way translation teams have a year to review if a page is still useful and should be kept, while allowing non-maintained translations to be gently put to rest when they get really out of sync.

It should be fix by manpower. At least for Chinese, it is not that no one want to do it. It is just no one have enough time to do it. 1. More users -> 2. More readers -> 3. More contributors -> 4. More regular contributors. It is a Funnel. Make installation more simple and at least provide installation and configuration translations to attract new users/readers. Make the initial contribute as easy as press a button and just start(Step 1). Do not revert a edit even if 80% of it is wrong, help refine the 20% and a contributor will learn along the way(2 to 3). Protect a contribution as much as we can. Guide them to become a regular contributor. --Fengchao (talk) 14:51, 9 January 2023 (UTC)

Edit:In light of your explanation on the ease with which Chinese readers handle English, I'm puzzled at the reasoning behind edits like your recent one on the Chinese wiki where the page is updated with the un-translated content from the English version. Would that not be an issue for your readers in "Segment 2"?

--Erus Iluvatar (talk) 19:47, 7 January 2023 (UTC)

Yes, some users can not understand what those text are. I only have a small amont of time for Archwiki. So I have to leave less importand part to other team members or anyone who does have time. We are always in short of manpower so it is back to the funnel again. Most users only need Installation and general configuration after all. So only a tiny percent of them will be influenced. --Fengchao (talk) 14:51, 9 January 2023 (UTC)

So you have a "dynamic community of tens of contributors" that is "too dynamic so we have to choose the Bazaar model", but not enough manpower for finishing translations? That does not add up. --Erus Iluvatar (talk) 15:30, 9 January 2023 (UTC)

Sorry, I do not catch what you said. Maybe you misunderstand what Bazaar is or do you really ever read The Cathedral and the Bazaar? In Bazaar mode, everyone just do what they are intrested in and it seem like a mess(a Bazaar) from outsider. Lack of manpower is a problem of most open source projects, is Arch full of manpower? Is ArchWiki full of manpower? I don't think so. It is a 80/20 problem, 80% of reader's need is fulfilled by 20% of page content. So I am only intrested in the 20% part right now. I do not want to translate a section that most readers do not need. Anyhow if one day a perfectionist show up and want to translate those sections he/she will be welcomed too. --Fengchao (talk) 12:42, 11 January 2023 (UTC)

No, what I'm saying is: you explained before that you don't have a visible organization on your translation team because it's too big (and you used the Cathedral/Bazaar metaphor as an example, which I am familiar with) but this is in contradiction with your explanation above about a lack of manpower to explain the lack of fully finished translations. Unless what you meant was that the solution to my issue with unfinished translation will spontaneously emerge once someone new joins your teams and wishes to tackle that part of translations (but then it is in contradiction with what you said as "it is not that no one want to do it. It is just no one have enough time to do it"). --Erus Iluvatar (talk) 14:04, 11 January 2023 (UTC)

There is no contradiction except the large and small words. The team is too large to require everyone having the skill/time to fully translate a page before start a contribution. At the same time, the team is still too small to fully translate all pages in Archwiki. So many translators choose to ignore some sections that is not too important/too hard to translate. The discussion is too long and loose focus and not related to other teams so I plan to stop it here. --Fengchao (talk) 14:21, 12 January 2023 (UTC)

Sound good to me, we've been running in circles for a while, let's wait to see if anyone has more on this topic. --Erus Iluvatar (talk) 14:42, 12 January 2023 (UTC)

Reader segmentation

Different language may have different reader segmentation, I thought that maybe it could improve our understanding of our different attitude and action. We could do a reader segmentation as below:

• Segment 1: Readers who could only read English, they just read English. Not related here.
• Segment 2: Readers who could only read their local language, English pages means nothing to them.
• Segment 3: Readers who could read their local language and English, but English is foregin so they will read local language first and refer to English when their problem is not solved.
• Segment 4: Readers who could read local language and English very well, they will read English first because sometime a translation may be out of date.

For Chinese Archwiki readers, I give my rough estimation:

• Segment 2: 50%. Although all students in China learn English, most of them are only familiar with daily English. They could not read English technical documents (ArchWiki and upstream documents) well.
• Segment 3: 40%. Many readers could read English page by using a dictionary. It is time consuming. Arch Linux is already very hard to install. So if there are no Chinese translation, most of them will stay with Ubuntu.
• Segment 4: 10% or less. I put my self in this segment. They are potential contributors.

I am curious about other languages for example, how is French readers' segmentation like. Does it differ a lot with Chinese readers? --Fengchao (talk) 02:01, 7 January 2023 (UTC)

I think there is more than simply the composition of our reader base that explains how we approach translations and general maintenance.

To answer your question, a guesstimate would be (in the tech space, not for the general population) 70% in the fourth category (but they don't see the need to contribute to translations since they dont depend on them), 30% in the third (but they usually do not need to refer to a dictionary, they will probably put the whole English page through an online translator if they are missing some words, so once again they do not depend on translations).

I don't think I've interacted with a single person unable to read English well enough that they need to rely only on French documentation.

Thank you for the insight on the marked difference in our target for translations though: the "translate the world" approach reminds me of France a1 or 2 decades ago where e.g. every single movie had to be translated (mandated by law), which in turn (though this is probably not the only factor) led to low English skills for most people, while other European countries only provided captions and had better English speakers.

--Erus Iluvatar (talk) 07:57, 7 January 2023 (UTC)

Thanks, that could explain a lot. I knew for the first time that most Franch could read English so well.

Chinese readers face a much larger gap, and the quality of AI translation is very bad and few readers will use whole page translation. English is compulsory from elementary to graduate but still many programmers will never read a English technical book or even a manual. And that's why we have different view on redirecting an out of date page to English. For French readers, few are hurt. But for Chinese readers in segment 2, it means from 80% down to 0%.

--Fengchao (talk) 13:59, 7 January 2023 (UTC)

Yeah, machine-translation is very bad with ideo/pictographs. To be fair, outside of the tech bubble, the percentage looks much more like what you described, with few people being fluent in English, and those who speak/read it only are taught about daily English and not much more. I have a much clearer picture of why you are so adamant about keeping in place partially translated of very old translations, as you see them as better than nothing, given than most of your readers will not go to the English page. Thank you very much for your patience with me on that topic. --Erus Iluvatar (talk) 16:25, 7 January 2023 (UTC)