Talk:Www.wikipedia.org template/Language sorting

From Meta, a Wikimedia project coordination wiki

Contents 1 Language ordering 2 Language sorting, round two 3 Language sorting, round three 4 Russian Wikipedia will have 100K articles soon 5 New 100,000+ section 6 Chinese Wikipedia will have 100k acticles 7 Suggestion: Removing the duplicate wikipedias from the mainpage 8 rethinking the top ten 9 Sorting by kind of alphabet 10 Italian Wikipedia 11 Translation of Search input

[edit] Language ordering

Moved from Talk:Www.wikipedia.org template, which was moved from Talk:Www.wikipedia.org template/temp

Just to clarify, I'd like to explain the system we've been using to arrange the language editions. The largest ten Wikipedia editions are listed according to their article count, from greatest to least, left-to-right, and top to bottom.

The rest of the editions are listed as follows:

1. Languages written in a Latin-based alphabet – or in an alphabet that corresponds more or less to the Latin alphabet, such as Greek or Cyrillic – are ordered alphabetically, according to the native language name or Latin transliteration.
2. Languages written with any other writing system are listed first, in alphabetical order based on the Latin transliteration.
3. Languages are alphabetized according to the identifying word in the native language name. Thus, "Bahasa Melayu" is alphabetized between "Македонски" and "Norsk (nynorsk)," rather than between "Asturianu" and "Bân-lâm-gú."
4. If the language is known under more than one name, the most commonly–used name is listed. In some cases where more than one writing system is commonly used, each writing system used at that Wikipedia edition is listed, separated by a solidus (/).

In addition, every line includes the lang and xml:lang attributes, to assist Web browsers with font-switching.

 – Minh Nguyễn (talk, blog) 6 July 2005 22:20 (UTC)

Actually, Minh, if I recall correctly it was a unilateral change and there was never a vote. Current policy on individual Wikipedias for interwiki links runs counter to this, and it seems absolutely absurd to me that you're basically saying "well all those strange-looking languages which I can't make any sense of, I'll just put them all at the beginning". Now, how about some consistency? There aren't enough languages in any of the lists that it would ever be truly difficult to find a language. However, I think that with the majority of the languages it makes almost as much or even more sense to order them by code than by your current system. When I changed the order on this page, it took me a good amount of time to do so, and I certainly didn't do it without a reason. It makes no sense that Hebrew comes before Catalan -- if you use the English name, it will come after Catalan, if you use the transliteration, it will come after Catalan, and if you use the ISO code, it will still come after Catalan. But in your dopey system, it comes before. Also, what's the purpose of the transliterations in the titles for the links? --Node ue

I think this system is quite reasonable. Ordering by ISO code is an esoteric practice most readers will find counter-intuitive. It always annoys me when I've sorted interwiki links alphabetically by the name that's displayed and then someone comes along and does it by code, so that we end up with weird anomalies like "Suomi" ending up among languages beginning with F just because it had "fi" allocated to it by some irrelevant people. ISO is a non-starter in my view. It's also silly and biased to go sorting language names alphabetically as if they were all written in the Latin alphabet. It makes no sense to find "Հայերէն" in the list as if it was spelt "Hayeren", because it isn't "Hayeren" it's "Հայերէն". There are few enough non-Latin names that they can be put

So what? Any reasonable person will expect to find "Հայերէն" AFTER Aragonese and Chuvash, because - guess what - its English name is Armenian, and its name transliterated is "hayeren". Why the hell would anybody look for it at the beginning? That makes _absolutely no sense_. ISO code ordering is not counter-intuitive, and whether you like it or not it's how things are done everywhere else in Wikimedia.

I would hope that the reader would be able to tell, just by glancing at the list, that non-Latin scripts are being placed up front. The problem with using ISO codes is not so much that it's counter-intuitive: it's hidden. But to solve that problem, we'd have to go the way of Wikisource's main portal. Ick.

The organisation of Finnish is an anomaly. It's one extremely illogical sorting out of a sea of very reasonable ones.

I currently see the following anomalies (if you factor in the transliterations): Estonian (Eesti – et), Finnish (Suomi – fi), Hungarian (Magyar – hu), Korean (한국어 / Hangugeo – ko), Southern Min (Bân-lâm-gú – zh-min-nan), Ido (io), Latvian (Latviešu – lv), Ossetic (Ирон æвзаг / Iron avžag  – os), Albanian (Shqip – sq), Filipino (Tagalog – tl), Tatar (Tatarça – tt), Anglo-Saxon (Eald Englisc – ang), Haitian Creole (Kreyòl ayisyen – ht), Kannada (ಕನ್ನಡ – kn), Maltese (Malti – mt), Northern Sami (Davvesámegilii – se), Tok Pisin (tpi). That's 17 out of the 95 or so in question. Some of these are only slightly misplaced, but they still serve to confuse readers who aren't familiar with an ISO scheme that may have been devised without any speakers of their language.

You're wrong about a few of these. Haitian Creole is going to be looked for under "Kreyòl". "Ayisyen" is just a qualifier to differentiate it from, say, Mauritian Creole or Seychellois Creole, in real life people generally just call it "Kreyòl".

Tok Pisin will be being looked for under "t", not "p". Unlike Indonesian and Malay, "Tok" is not a generic label that is discarded in alphabetisation. People are probably going to be looking for Northern Sami under "Sámegilii", alphabetically. Like with Haitian Creole, it's only there to avoid confusion and make it known that it's not also for, say, Inari Sami or Southern Sami. Outside of these, there are only a handful that are seriously out of place, Finnish, Hungarian, Korean, Southern Min, Ossetic, Anglo-Saxon, and Haitian Creole. Out of these, Finnish and Korean are in a list that is short enough and in large enough print that they will still be easy to find. Hungarian (H -> M) and Ossetic (O -> I) are still relatively close to where they would be in alphabetical order, and are still in the 1000+ list and shouldn't be that difficult to find. The only difficulties are Haitian Creole, Southern Min, and Anglo-Saxon. Haitian Creole isn't that bad. Many people will look for it under "h" anyways, as it is often ordered by its French name (Haïtien). Southern Min poses a real problem with confusion.

Anyhow, the ones that are "seriously off" -- Finnish, Hungarian, Korean, S. Min, Ossetic, Anglo-Saxon, and Haitian Creole, could perhaps be part of a compromise. For Finnish and Korean the need isn't so great, and I don't think it's really very great for any of the others either except S. Min.

And by the way, when you went ahead and changed the ordering scheme back to ISO codes, you forgot to place the bullets in the right places (such as the one after Chinese). You also forgot to move Southern Min, but I'm sure the Minnan speakers are now accustomed to people listing their language at the end, for no visibly apparent reason. And Simple English is in the same place, although that's understandable, since "simple" really isn't a language code, but "en" is. And Georgian (ქართული – ka) and Cashubian (Kaszëbsczi – csb) are in the wrong places, even according to the ISO scheme.

together without causing much trouble and without appearing to be randomly scattered about because of what their ISO codes happen to be. I think an exception can be made for Cyrillic and Greek which are sufficiently close and transliterate easily enough that it's reasonable to find "Русский" sorted under R, although I would also be happy to see all Cyrillic names sorted together at the

Oh, so Cyrillic and Greek are "sufficiently close" and "transliterate easily enough", yet Armenian, Georgian, and Arabic don't? How do you know this? Have you studied any of these alphabets? The simple fact is, Armenian is very easy to transliterate, and Հայերէն transliterates perfectly -- perfectly -- as "Hayeren", no ifs, ands, or buts.

Alright then, my mistake. If it is true that those scripts also transliterate that well (which does make sense given their geographical/cultural relationship to other languages in Europe and Asia), then I fully agree that they should go in the main list.

If the clause about "sufficiently close" scripts is a problem, we can just go ahead and move Cyrillic and Greek languages over to the first part of each list as well.

beginning if that's what people want. Now as for this comment about "well all those strange-looking languages which I can't make any sense of, I'll just put them all at the beginning", this is a non sequitur as far as I can see. What makes you think Minh or anyone else would put them at the beginning because he can't understand them? If anything, the ignorant course of action would be to pretend they were all written in the Latin alphabet and order them like that.

Well, so, it's NOT ignorant to pretend that Greek, Bulgarian, Russian, Ukrainian, etc. are written in the Latin alphabet? Where's the logic here?? At least what Minh has to say (see below) has some logic to it. Yours is totally off the radar.

Even if this wasn't done out of ignorance, it would be Latin-centric. None of your three systems in the Hebrew example are really very sensible, as I'm sure you realise. It does make sense that Hebrew comes before Catalan, because it's written with a different alphabet, and shouldn't just be lumped in with all the Latin-script titles. Seperating out the alphabets is the best way. Since we mainly have names written in Latin script, along with a few representatives of other alphabets,

Why should different scripts be sorted in a completely different section? This is discrimination.

I was afraid that the charge of discrimination would come up. I wanted different scripts in a separate part of the list simply for practicality. I'm not psychologist, but it seems that just by glancing at my version of the list, you get an idea of what's going on.

But how is it in any way practical to separate different scripts?

 – Minh Nguyễn (talk, blog) 21:28, 21 July 2005 (UTC)

there is no great need to specify an arbitrary order of alphabets whereby we'd have Devangari, then Arabic-based, and so on. The ordering of the language names not written in the Latin alphabet is always going to be fairly arbitrary, so ISO codes are as good as anything here I think. Feel free to propose a better way of sorting the non-Latin names, but just mixing them all in with the Latin ones and relying on ISO codes, Latin transliterations, or English names, would be a dopey system. The system we have now seems fine to me. Trilobite 07:13, 21 July 2005 (UTC)

Well, it seems extremely rediculous to me, that Hebrew comes before Catalan. The Hebrew word for Hebrew is, when transliterated, "Ivrit". The ISO code is "he". The English word is Hebrew. The only tripped-up sorting order that places Hebrew *before* Catalan is yours. --Node ue

The change I made was mainly for consistency's sake, because the languages were previously ordered according to the whim of the last person to edit the page. The reason I went with the "rules" above was that everyday users of Wikipedia have probably never heard of ISO language codes, and thus wouldn't think to look for Finnish next to Esperanto and Indonesian. Also, the number of languages written in the Latin alphabet (henceforth to be called "Latinate languages" for brevity) on the portal is dwarfed by the number of non-Latinate languages. So I lumped all the non-Latinate languages together, hoping that part of my scheme would be fairly obvious.

The fact is, Finnish is really one of only a couple of languages that aren't readily findable in this sorting order. Yet, Finnish, being in the 10000-articles-or-more section, should not be difficult to find -- it's big enough that anybody should be able to find it. For most of the other languages, ISO order is compromise in a sea of chaos. In a relatively recent vote on the English Wikipedia, people voted narrowly to sort interwiki links by ISO code because there really is no other logical way.

But are we just going to say "tough luck" to speakers of those smaller non-Latinate languages, and force them to search for their language in "a sea of very reasonable [orderings]"? At least with the scheme that I used, they wouldn't have to search through such a long list.

It's unfortunate that this is one of those issues that can't be solved through consensus. The narrow vote, though, indicates that the ISO order perhaps isn't a great way to arrange the languages, though it may have been ever-so-slightly the best one available to the English Wikipedia community.

The "narrow vote"? You mean, you and Trilobite voicing your opinions in opposition to mine? And the option you favour is also available to en.wiki, so why would one be better there but not here?

As for the transliteration tooltips, I know you're not fond of using the transliterations in the first place – I'm still not sure why Tim Starling brought them back. I just put them there so that people looking at the code would have a clue how it's all arranged. The tooltips would serve just as well as comments – that's probably what I'll change it to when I have the time.

What do you mean, so they have a clue how it's all arranged? Why would comments be nessecary? People should be able to tell which Wikipedia it is by the ISO code. If they can't, they can always look it up. The script doesn't really make it any easier -- how is the average non-Hungarian-speaking Wikipedian to know that "magyar" means "Hungarian"? Certainly, it's just as hard to figure out "Magyar" as it is to figure out "Հայերէն", if you don't know Hungarian or Armenian. In fact, it might be marginally easier to figure out Հայերէն, because Armenian writing has a distinctive "feel", while Hungarian is written in Latin, like a heap of other languages on there.

When I said "they," I meant "those who edit the HTML code." And how do you expect the everyday user to "look it up," when they don't even realize that we're ordering by ISO code? They'll just glance at the list, not be able to find the language they want, and give up. (It took me a couple months to even notice the Vietnamese Wikipedia's link on the English Wikipedia way back when.)

We're dealing with such diverse languages here that, no matter the ordering scheme we use, there's going to be a seemingly random rule in there. I wouldn't mind if we ordered the non-Latinate languages by ISO code, since that part of the list is short, so it's relatively easy for the reader to pick a link from anyways.

Yes, but what seems most random to me is sorting languages by script. It has plenty of strange placements -- Hebrew before Catalan, Arabic before Afrikaans, Farsi before Asturian. And many, many others. Yet, with ISO, there are only a handful, Finnish, Hungarian, Hebrew, Japanese, and Korean are the only ones that come to mind; all but one of these are big enough that they are in lists that are so short they should still be easy to find (Hungarian is not; yet hu... and ma... are close enough it still shouldn't be that difficult).

I'm not proposing to completely sort by script; just to divide the list into two sections, since that seems more practical IMHO. Hebrew may come before Catalan, Arabic before Afrikaans, and Farsi before Asturian, but it would seem obvious for a speaker of Hebrew, for example, to look in that first part of the list, since the second part clearly contains only languages written in the Latin alphabet (well, it would, given my proposal to eliminate the "sufficiently close" rule).

I really think that's just more bull. That is, for lack of a better term, discrimination. You are assuming that these people don't know the Roman alphabet. Also, languages like "hebrew" are in lists short enough and in large enough fonts that they won't have to go searching at all.

But I didn't want Tamil speakers, for example, scratching their heads as they search the entire (microscopic) 100+ list for their language, just because they don't happen to know the Latin transliteration. Speakers of Tok Pisin, on the other hand, would be able to at least figure out that tpi: is listed under either T or P. [By the way, we've got to make the text of that 100+ list a little larger. It's already a bit difficult for me to read the Latinate part of the list; just think of how Bengali (বাংলা) must feel.]

I can guarantee you that there is not a single Tamil speaker on the Internet who does not know that the transliteration for, and English translation of, the name of their language is "Tamil". There aren't a lot of people who are literate in Tamil and yet don't know the Latin alphabet, and probably none of these people have access to computers. I agree with making the 100+ list bigger, but Bengali should, if you have a good font, be about the same size as any other language.

So Tamil was a bad example, but that's probably not the case for every language out there. And my only Bengali font is one of those all-purpose Unicode fonts, so that could be the problem.

But it _is_ the case for every language out there. To use the Internet, knowing the Latin alphabet is pretty much a requirement as the vast majority of URLs use it, as do many input methods (especially for Indic languages).

Well, I suppose I don't have much room to speak about ease of navigation: I was the one who compiled the list of languages at vi:, and it's currently atrocious, because it once doubled as a statistics page. I've got to clean that up as well.

I'm certainly still upset about the rejection of _my_ proposal for the template, which had ALL of the current contents fitting into one 600x800 window. And why on Earth do we have a huge image at the top right now that says what is unmistakably English (not some universal interlanguage, no, but English) stuff about Wikimania?? This is a _portal_ page. Anybody who clicks on their appropriate language will get information about Wikimania immediately upon their arrival, and it will be in the proper language too. --Node ue

You mean this? It probably got rejected because most people were worried about how the page looked at the time, not how functional it was.

That's certainly true. But has that changed at all? --Node ue

Oh... by the way, I meant this one, not the one you linked which is much much older. --Node ue

Yeah, the Wikimania banner should contain just the name and logo, since it points to a website that's been translated into several languages. I think the "Wikipedia" wordmark belongs there, because that is the project's official name, isn't it? And most users will have to type in www.wikipedia.org to get to the site anyways, so they're probably aware of its Anglo-centric name. :^)

And I think we should stop threading our comments; this is going to get messy real soon.

 – Minh Nguyễn (talk, blog) 21:28, 21 July 2005 (UTC)

Let's please stop threading the comments:

1. Pardon my ignorance, but when you simply call Haitian Creole "creole," isn't that a lot like saying "language," as in "English language"? Anyhow, I'm sorry for not knowing more about these languages.
2. Even though you classify those five or so languages as a "compromise," it still doesn't help people actually find their language, and that's what I'm most concerned about, since most newcomers are first directed to the main portal, which is where they either find their language or don't. If they don't find their language easily enough, they'll just give up – at that language's expense.
3. People I know always ask me for help with their computers for simple things like finding Paint in the Start Menu. Finding Paint's location in the Start Menu is kind of obvious me, and it's probably not that hard for you either (unless you're on Mac OS or Linux, of course). But for the multitude of computer users who aren't computer savvy, even searching through the Start Menu becomes a hassle. If they can't figure out their own computer, which uses their language, how are they to find a small link amidst the heap of strange-looking characters?
4. True, most Internet users probably have some understanding of a Latin-based alphabet, but why make people go through that hoop of hunting their language down in a longer list? By splitting each list up into no more than two sections, users don't have to look through a list of 42 languages, they would effectively have to look through a list of either eight or 30, for example (depending on the language). And in the second part of the list, the alphabetic sorting becomes more straightforward.
5. If setting all the non-Latin scripts to the front is discrimination, wouldn't that be some kind of positive discrimination for the speakers of those languages? :^) Actually, most of those languages were already being listed at the front before I came along and started maintaining that page. All I did was to make the sorting more consistent.
6. When I said "narrow vote," I was referring to the en: vote that you mentioned earlier:

   In a relatively recent vote on the English Wikipedia, people voted narrowly to sort interwiki links by ISO code because there really is no other logical way.

7. But yes, me and Trilobite voting against you would be a narrow vote. :^) Actually, I was mistakenly thinking of the vote regarding en:Template:Wikipedialang, the vote that you invited me to, coincidentally. :^) That template lists the languages' English names as well as the native names, so the situation is a bit different there, since it would kind of make sense to sort according to the English name if the English name is what appears first in every link.
8. Oh, something I forgot to reply to:

   ISO code ordering is not counter-intuitive, and whether you like it or not it's how things are done everywhere else in Wikimedia.

9. The Vietnamese-language projects arrange the languages according to how their displayed. Thus, "Bahasa Indonesia" is under B. That was the consensus there, and I know vi: isn't alone in that practice. (be: da: et: nn:). It's not completely intuitive, but at least our reasons aren't hidden behind an edit page – you can literally see why we put Indonesian under B, even if it doesn't make incredible sense. Another compromise, right? :^)
10. Your proposed design has some good ideas in it. (I take it that your design was more of a proof of concept, right?) In particular, I like that you replaced the "search" words with a magnifying glass. I'm surprised that it wasn't part of Catherine's design in the first place. Unfortunately, the background is dark enough that, on my monitor, it makes it slightly harder to read the foreground text. Also, I've been meaning to propose that the book icons become stacks of pages – that what the numbers more or less represent; the books would belong in Wikibooks' portal page, if they ever decided to make theirs look more like Wikipedia's.
11. Speaking of the design, I really liked the look of Forseti's design. (Though I would support replacing his background with the one in your proposal.) For awhile, there was an effort to merge Forseti's design with Catherine's, but it looks like nothing ever came of it.
12. Again, please don't thread the comments; I can figure out what you're saying if you post it all below this message. Thanks.

 – Minh Nguyễn (talk, blog) 01:50, 22 July 2005 (UTC)

I would like to number each paragraph of yours so I could have numbered responses to each. I've done this, if it's not OK feel free to remove the numbers from your comment.

So, here are my responses:

1. It may seem this way, but how many other creole languages do you think most Haitians are familiar with? Most have heard of other languages, such as French, English, Spanish, etc, but the only creole they really care about is Haitian Creole. So, the usual name for their language is "Kreyòl". "Ayisyen" is only used when it's nessecary to distinguish it from other creoles, as is the case here -- we don't want somebody coming to ht.wikipedia and writing articles in Mauritian Creole, which also goes by "Kreyòl" (in its case, "Kreyòl morisyen" is the more specific name). In Tok Pisin, "tok" is a general name for languages (as in "tok inglis", English), yet in an alphabetical list it will still be expected as a "t" rather than a "p".
2. With only a few exceptions, ISO order matches alphabetical, or it is only off by one or two. Many of the exceptions are languages with over 10k articles, so the font is big enough that they will still be easy to find (Finnish, Korean, Hebrew).
3. With Paint, it's something that you just have to look for, or know in advance. There's not even a hint of any logic to it. However, in ISO order, it's at least pseudo-alphabetical, so things should be realatively easy to find. Also, I don't know about you, but it's my guess that text in the Armenian script will jump out from all the other alphabets to an Armenian speaker.
4. As I noted in my previous numbered point, text in the script of your native language often seems to jump out from a jungle of text in other scripts. For me, text in the Roman alphabet jumps out from a long page of Arabic text, even though I can read Arabic to a degree. If you've ever seen the movie "Spirited Away", Chihiro is able to pick her parents (transformed into pigs) out of a group of seemingly-identical pigs. Presumably this is because of familiarity -- even though most people wouldn't be able to tell, she was able to because she was familiar with them, so they "jumped out" at her. With most non-Latin scripts, there are between 1 and 5 languages using the same script on this page, so if they all "jump out", it should not be difficult to find the right language from a maximum of 5 choices.
5. Positive discrimination, perhaps. But it will possibly be confusing to people who use the Latin alphabet, who see that each list starts off with strange characters (which may even appear as question marks if your browser doesn't support the script).
6. --
7. I actually meant one regarding interwikis.
8. --
9. You're right about that. But still, at the majority of Wikipedias, the consensus is to order by ISO code. One of the other brilliant Interwiki ideas was, on the Hebrew Wikipedia, to remove all Interwikis except to English, and /maybe/ French, Spanish, and Russian.
10. I'm pleased that you like my design. I agree about the books, but I'm afraid images of stacks of pages would be difficult to make out what they were supposed to be. I did however replace Catherine's ugly blue books with some nice, better-looking books. I'll try to fix the background. The entire point of my design was actually so that all of the languages would fit in one page, without modifying font sizes much. I'd also like to remove the search box as I think it's vile and inappropriate for a portal, but alas the consensus seemed to be that we should keep it. The other main point was to remove the "Wikipedia" image, because I don't think it's language neutral, and it also takes up quite a bit of space.
11. I really favoured Forseti's over Catherine's, but obviously EVERYBODy hates it... since his was entered into the vote a couple of days later, it's not exactly fair though.

    Node ue 22:06, 22 July 2005 (UTC)

1. Alright, you're the language expert. If you say so... :^)
2. You'll still have people scratching their heads, because how many readers of Wikipedia will have even heard of the ISO language codes? If we use the codes, we'll have to make the codes visible, without making the portal look as clumsy as Wikisource's. Any ideas on how to provide this information subtly? I'll try to come up with something, but getting the portal substantially changed would be an uphill battle.
3. By putting the shortcuts in folders, the Windows team (intentionally or otherwise) implied that there was some logic to the placement of shortcuts. But how many everyday users actually know why it's called "Accessories"?
4. No, I've never seen that movie before, but you're quite correct about different scripts "jumping out" – assuming that the lists are reasonably short, and that some of that text gets enlarged. The latter issue can be resolved easily, but shortening the lists by breaking them up into more article ranges is probably a bad idea. If a user doesn't know the transliteration for, say, Kannada – I'm just randomly using an example; I have no idea if Kannada-speakers mostly know the transliteration or not – it'll take a little while to find their link. But, then again, putting the non-Latin scripts up front might not help much more, since now many of those scripts will look alike. I just hope that most visitors will have the patience to find their language.
5. I don't really think that seeing those strange characters (or boxes) at the beginning of each list would necessarily confuse people. I think most people will be aware of the existence of other languages, and of other scripts like Chinese and Japanese and Arabic. I don't think they'll be that surprised that other scripts like Gujarati and Armenian exist.
6.  
7. Yes, my mistake.
8.  
9. I don't know if it's brilliant; it's probably very easy to maintain, but it might make readers think that there are only two versions: Hebrew and English. (You'd have to change "English" to "Other languages; please scroll down on the page you arrive at.") For a one-link solution to work, the link would have to lead to an "interwiki wiki," a middleman, in which each page lists only links to different versions of the same article. That wiki, though, would still have the same ordering issue as we do, and I'm stumped as to how they would name each page.
10. Regarding the wordmark, see my comments below. Regarding the space that said wordmark occupies, I liked that your design took care of that issue by presenting some names of Wikipedia in commonly-used languages. Regarding the search box: I know many people who'd be lost if the searchbox were removed from just the portal. Plus, it's probably more convenient for speakers of the ten largest languages who have dialup. I really have no idea what segment of the Wikipedia-reading population that is.
11. Yeah, I'd really like for there to be a revote. Forseti's sure fits with the Monobook skin better, since the two designs share the same background. The MediaWiki developers used Forseti's design for their blackout message before, and I have a clue as to why.

 – Minh Nguyễn (talk, blog) 02:08, 23 July 2005 (UTC)

(After edit conflict with Minh.) Agreed with that. My own comments have been chopped to pieces and replied to in bits out of context, and then all the rest of what's been said gives no immediately obvious indication as to who said it. Firstly, my view on the Wikimania banner is that it shouldn't be there at all. The word "Wikipedia" above the globe, which Node removed at one point, I do not regard as language-specific. Yes it's the English name of the project, but since it's also used for the URLs of all of them I think it's reasonable to regard it as the underlying universal name of Wikipedia. Versions like "ויקיפדיה" and "Uiquipedia" are just attempts to render something like "Wikipedia" in a way that's appropriate to the language being used. If Wikipedia had been started in Asturian maybe we'd have had http://en.uiquipedia.org/ and written it as "Wikipedia" for the English logo, but it would still have been reasonable to have "Uiquipedia" on the multilingual portal because all the language versions would have been at xx.uiquipedia.org and would have based their own spellings of the name on that word. Now as for Cyrillic and Greek, as I said, I would be happy to have them sorted at the beginning (actually I would prefer this). I know very well that Armenian and others can be transliterated just as Cyrillic and Greek can, so fine, let's have full seperation of alphabets instead of one rule for Cyrillic and Greek and another for Armenian. Please don't paint the mixing of these with Latin script as my suggestion (that is "totally off the radar" to use your rather inflammatory words) when all I said was that I was prepared to tolerate it - I don't like it and would prefer full seperation. We could order by alphabet first (though the order will be necessarily arbitrary - perhaps we should do it by least represented to most), then by alphabetical order within those alphabets. As for seperating alphabets being discrimination, this is nonsense, and they are not "in a completely different section", they are just at the beginning of the list instead of being arbitrarily mixed in with those in the Latin script according to their ISO codes. ISO codes are meaningless to the average visitor. We cannot accept it as reasonable that "Suomi" is sorted under "fi". Seperating alphabets makes it clear at a glance that you don't need to go searching through the list looking for your non-Latin language name under its transliterated form, its English name, or its ISO code. It is perfectly reasonable to seperate alphabets instead of trying to come up with some universal ordering system based on ISO or Latin transliteration. Another thing I have noticed is your altering of language names. While I'm all for accuracy I think it would be best if we took them from the names used for the interwiki links, and then argue seperately about what the interwiki links should say instead of just changing it here, resulting in inconsistency between the portal and what's used throughout all language editions as the text of interwiki links. For example, you changed "Kernewek" back to "Kernewek / Karnuack", but "Kernewek" suffices as the interwiki link. If Cornish speakers who want to promote different orthographies object to the bias of the spelling we use for the interwiki link they can argue that elsewhere and once they get it changed, the portal can be changed. If you want to be neutral and use all orthographies, you should also have included "Kernowek" since this is the Unified Cornish Revised version used even on the main page of the Cornish Wikipedia. Finally, if you are going to go editing the portal, please at least keep it tidy by remembering not to leave out the "•" at the end of lines and all the rest of it. Thanks. (And if you're going to reply point by point, please add numbers or something instead of chopping the comment up, since this isn't the mailing list and it's not so easy to keep track of who's said what). I hope we also get some input from other people here. Tim Starling? Trilobite 02:52, 22 July 2005 (UTC)

I'm only going to reply to a couple of points here because I would say the rest are not worth responding to, or I have already responded.

1. "Wikipedia" is not language neutral. You make it sound like translations of the name are all fake somehow. My refutation is this: All Wikipedias which have a localised name, use it in ALL contexts except for the domain name. If what you said were the case, it would be like the Southern Min and Vietnamese Wikipedias, where the name used is "Wikipedia" but in the article on "Wikipedia" it explains that it's pronounced "ūi-ki̍h-phí-lī-ià" and "Uikhipéđia", respectively.
2. On the mainpage at kw.wiki, it uses ALL orthographies, with a separate introduction section in each. The interwiki link is a non-issue, wrong Interwikis stay wrong for ages without being changed.

   -Node ue 22:06, 22 July 2005 (UTC)

1. Regarding the wordmark: Likewise, the logo itself isn't completely neutral. The Romanian Wikipedia uses an Ă in one puzzle piece instead of an Й – they insisted on that modification, despite your efforts otherwise – and the Sicilian Wikipedia changed the shape of the logo entirely. You wouldn't propose including multiple versions of the logo, would you?
2. Keeping only one version of the logo on the main portal is a compromise (a no-brainer, really). You're apparently willing to compromise on the sorting of languages by ignoring the few "anomalies" for the sake of the "sea of very reasonable [sort positions]." So why not compromise on the wordmark as well? From a cursory glance, it seems that the vast majority of languages maintain "Wikipedia" as the project name, and many (like French) make only small changes:
   • 29 of the 69 languages at Wikipedia raster name use "Wikipedia" verbatim. Four more just add an accent mark. Four more just replace or add a letter, but the name is still recognizable either way.
   • Of the 67 translations at Wiktionary's Wikipedia entry, 26 are "Wikipedia" verbatim. Four more simply add an acute mark over the e. Two change only one letter. Four change two letters, while leaving the name still recognizable.
   • Of the 102 languages listed at Wikipedia logo in each language, 35 are verbatim, six add only diacritic marks, five only change or add one letter, and 12 change only two, keeping the name still recognizable.
3. In all, that's about half of the languages listed. I suppose the wordmark isn't absolutely essential, but it's an important part of the branding. You rarely see the logo without it. Plus, I don't know of any Wikipedia project that has changed their W favicon to a U or V. If people bugged the developers about it enough, they could probably change it, but I'd think that most Wikipedias are trying to associate themselves with the English Wikipedia, and the W certainly helps in that regard.
4. I don't know, perhaps I'm just wary of removing the wordmark because I remember the days before the puzzle ball. Back then, the wordmark was more unique to Wikipedia than the logo, to me at least. Ah, the good ol' days. :^P
5. By the way, "Uikhipéđia" is never mentioned in the Vietnamese article. You probably got that from a talk page or something, where I tried to explain to someone how I'd go about pronouncing it. In fact, others might pronounce it quite differently, something like "Ưikibiđia" or "Uikibéđi." And the more "correct" way (for some, especially in the academic community) might be to transliterate it from Chinese, in which case it might be "Hoài Cơ Bách Khoa," which wouldn't fit too well in the logo.

 – Minh Nguyễn (talk, blog) 02:08, 23 July 2005 (UTC)

1. There was actually never a vote at the Romanian Wikipedia, it seems to mostly be due to the persistence of Danutz. I'm still convinced that this is due to some degree of Slavophobia, as Romanians are resentful of the communist era which they associate with Russians. The Sicilian Wikipedia's logo has only been changed temporarily (in fact, I fixed it for them because they were having some troubles with it), this is for their "festival of quality" which has a focus on Sicily, similar to other temporary logo changes at fr:, it:, etc. for holidays and the like. So certainly with the possible exception of ro.wiki the globeball is neutral. I think that most people have come to associate it with Wikipedia, and as noted before the background is intended as a replacement for the raster name because I feel it's a waste of valuable space to have such a big thing on top which isn't language neutral at that.
2. Yes Minh, the majority keep "Wikipedia", but if you weight the count based on the size of the Wiki, you'll find that in the top 29 (by article count), 12 change it. 4 of these are still in the Latin alphabet. Only one qualifies as "minor" (French using Wikipédia). In my background, however, I chose one language from each script (English for Latin, Ukrainian for Cyrillic, Hebrew for Hebrew [as opposed to Yiddish, which uses a different name], Arabic for Arabic, Hindi for Devanagari). For Devanagari and Arabic, there is little or no variance between the different names in these scripts. This may not be "neutral", but it comes closer than having only one name. Also, it leaves some scripts out, for example Armenian and Georgian, but the Wikipedias in these languages are still relatively small. I did however design it when Georgian was still much tinier, so perhaps the addition of Georgian is in order.
3. -
4. -
5. -
6. I don't think most of them are concerned about associating or dissociating themselves with the English Wikipedia. The only example of a duality of names (one for the local version, another for the entire project) is at the Galician Wikipedia. They call themselves "Galipedia", but refer to the project as a whole (ie, including other languages) as "Wikipedia", and according to them Galipedia is a Wikipedia. There was a similar thing going on at the Minnan Wikipedia, but that was due to transition from a different domainname and is pretty much gone now. I don't think most people even notice the favourites logo, and the majority of people can't even see it because they use a stupid browser (i.e. IE).
7. I remember that. But then, at that time, the concept of non-English Wikipedias was still relatively new, and iirc this was before the English Wikipedia was moved to en.wikipedia.com.
8. Hoài Cơ Bách Khoa... now there's a laugh. That's actually what the Korean Wikipedia did, going for "Wikibaekkwa". And, while the Chinese name sounds rather nice in Mandarin (weiji baike), it sounds absolutely awful in Cantonese (waigei baakfo) and perhaps in Classical Chinese (uigiə bækkuɑ).

   Node ue 16:13, 23 July 2005 (UTC)

1. Yeah, whenever I choose photos for the Main Page at the Vietnamese Wikipedia, I always have to be careful about the images I select, because many people (especially in Southern California) can get very bitter about Communism and the Vietnam War. Now, if we ever move to using your background image, I'd suggest that you get the image cleaned up a bit – you know, some hinting etc. The font might be a little too stylistic – you could just rehash the wordmarks in Nohat's multilingual logos, and gray them out a bit. Also, with the wordmark gone, the page seems to be pushed up a bit too far. Maybe just allowing for a little margin at the top would make it more aesthetically pleasing.
2. Your statistics are still pretty much consistent with mine – about half keep the name untouched (17/29 = 58.6%). I want others to get into this discussion, so we can have outside opinions; the fact that you've tried to represent major scripts is a plus for your side of the argument.
3.  
4.  
5.  
6. Well, Vietnamese Wikipedia is very concerned about associating ourselves with the English edition. In the past year, a few projects have popped up claiming to be Vietnamese-language Wikipedia editions, when in fact they just happen to use MediaWiki. One project in particular, Vkpedia, actually touted itself as the Vietnamese edition of Wikipedia, when in fact they are not. Some major Vietnamese news outlets featured stories on Vkpedia as such, comparing them to the English Wikipedia without so much as mentioning the Vietnamese Wikipedia, which was founded more than a year after we started.
   
   Earlier this year, Vkpedia threatened to severely undermine our project, when some of our prominent contributors decided to jump ship. Anyway, Vkpedia has been dormant for a few months now. I suppose that imitation is the sincerest form of flattery, but imagine the surprise and confusion that people who've heard of Vkpedia must have when they stumble upon our Vietnamese edition.
   
   Just to clarify, I don't have any problem with Wikipedia editions branding themselves slightly differently. On the Vietnamese translation of the Wikimedia News page, I always mention the local Wikipedia name when I can, because it's fun to observe these projects' variety (independence, really). The Vietnamese Wikipedia finds it important to present itself as the official Vietnamese edition, though, and I wouldn't be surprised if others find it important as well.
   
   As for the favicon: well, that majority is decreasing with Firefox isn't it. :^)
7. IIRC, the concept of separate, non-English Wikipedias was considered a temporary hack, until the developers could figure out a better way of doing things. I've been tracking the development of the Devmo (Mozilla Developer Center) wiki, where they've actually implemented a single signon etc. for their five or so language editions.
8. Actually, I just checked on it again, and it turns out that my transliteration was incorrect (yay for using the Unihan database…). It's apparently "Duy Cơ Bách Khoa", which is at least shorter! [1]

How's about we try to get more people involved in this discussion – maybe hold yet another vote – before we change the language sorting policy on the actual portal? I've invited Tim Starling so far; I'll also post to the Vietnamese wiki communities.

 – Minh Nguyễn (talk, blog) 02:14, 25 July 2005 (UTC)

Responding to Node:

1. "You make it sound like translations of the name are all fake somehow." Not at all, they are perfectly valid, and it's good to have localised variants of the name. This should be encouraged. However, as I said above, "Wikipedia" is the underlying universal name on which the others are based. Rather than justifying this in an anglocentric way, by saying that the English Wikipedia was the first and should therefore take priority somehow (which I don't believe), we can say this is valid because it's the URL of the whole site. I would use localised variants in ALL contexts except the domain name and the multilingual portal, which after all is the page that's located at the domain name with no language specified, so really all we are doing is providing a way for people to remember the domain name of a site which they can otherwise know by a localised name. Furthermore, as Minh says, we compromise on the logo even though this has some variants.
2. Maybe you ought to set about arguing for the interwikis to be changed then. I'd certainly support you if you're trying to fix those that are wrong. After all, this thing we're arguing about here is only the portal - a single page, although a very visible one. Meanwhile if someone writes an article in a certain language where the interwiki link has been got wrong, every equivalent of that article in every other language that links to it has its interwiki link wrong. Sorting this out is a higher priority, and getting it right allows us to be both accurate and consistent at the multilingual portal. Your response to my Cornish example made no sense. The main page uses four orthographies, with three variants of the word between them. The article on Cornish is located at Kernewek, and the interwiki links say "Kernewek". As far as I can see then, the options for the portal are "Kernewek" or "Kernewek / Kernowek / Kernuack".

I still haven't seen a good argument against alphabet seperation, though I'm quite open to one if you want to put the case instead of just saying perfectly reasonable points are not worth responding to. This seems obviously preferable to ISO codes with all their anomalies. I suspect you are too eager to see a desire in people to be language-specific, as suggested by your comment about the motivation for alphabet seperation being that the non-Latin names weren't understood by whoever was doing the seperation and could therefore be sidelined somehow. If you believe this, you are mistaken. I think all three of us in this discussion actually have pretty similar ideas about the equality of languages within the project (let's remember after all that many people at en: were strongly opposed to setting up a multilingual portal in the first place, claiming that en: was the flagship edition and all others were subsidiary, which I vehemently disagree with), but you're getting carried away and claiming things like the word "Wikipedia" cannot be used to name the project as a whole because there happen to be some localised versions of it. Imagine a hypothetical printed finished Wikipedia some years down the line, with a stack of volumes constituting a comprehensive encyclopaedia of similar size in every living natural language. Volumes of the Hebrew edition would have "ויקיפדיה" on the cover, and volumes of the Asturian edition would have "Uiquipedia" on the cover. If we stuck all these in a big Wikipedia library however it would be legitimate to have a sign reading "Wikipedia" on the door. This is all getting away from the original issue however, which is do we seperate the alphabets somehow (my preferred option), do we mix all alphabets and order by Latin transliteration (I could tolerate that and I suspect this may eventually be the way it gets done), or do we mix all alphabets and order by ISO code (bad idea that leads to silly anomalies)? I really hope someone else weighs in on this discussion. Trilobite 03:52, 23 July 2005 (UTC)

Incidentally, I consider this issue to be bound up with the ordering of interwiki links. As more and more language editions grow, lots of articles are going to have large numbers of interwiki links, and we need a better way of organising them. I would like a set order to be built into the software, just as we build the names into the software instead of specifying them when editing the page, and then whichever order the links were found in the source text, they would be displayed in the sidebar according to the predefined pattern for sorting, whether it's by ISO or Latin transliteration or seperation of alphabets. Node mentioned consistency: well let's have some consistency between the portal and interwiki links. This makes it all the more important for others to get involved in the discussion. Trilobite 04:08, 23 July 2005 (UTC)

Yes, I also think it's likely that the languages will eventually be ordered by transliteration, and there are a lot of arguments supporting that practice, not the least of which is consistency with the interwiki links. Has a bug ever been filed, requesting a software-defined order for interwiki links? If not, I'll gladly file the bug report.

I think environmentalists would be on our case if we attempted to publish the entirety of Wikipedia's content. :^) Following up on your library analogy: it would help if, in the next few years, Wikipedia builds up a brand around its logo. In order to do that, we'd have to use the logo religiously. Once the general public becomes as familiar with the puzzle ball as it is with Google's G, or the Windows logo, or Apple Computer's logo, then we can start omitting the project's name, because people won't have to be reminded of the name: the logo will immediately call the project to mind, in whatever language the person is used to. But for now, we need to help people associate our logo with our name.

I guess if we somehow made the portal less cluttered – I'm not sure it's even worth trying, since we have to list 100+ languages now – then we could go for a very simplistic, Apple-like style, in which the puzzle ball is very dominant in the design, and the wordmark wouldn't have to be.

Another thing: Node, I know that you were trying to pack most of the portal into one page with your proposed design, but even with my large resolution, it barely fits. I think that fitting all the languages in one screen would only work if we moved to a grid layout, something like Windows Explorer's list view (that would be overwhelming to the visitor, though).

 – Minh Nguyễn (talk, blog) 05:21, 23 July 2005 (UTC)

Minh you make some good comments. I don't know if a bug report has been filed about interwiki link ordering, but I've just brought the subject up on IRC to see if any of the developers who were around had an opinion on it. Didn't get too much of a reaction but there seems to be a slight preference for ISO, and not too much enthusiasm for a software-defined order (even though interwikis are already being sorted unofficially, and somewhat inconsistently). I think if they considered the matter though, people would appreciate that ISO anomalies are confusing to the casual reader ("suomi" sorted under "fi", "español" and "esperanto" appearing to be the wrong way round etc.), and that whatever we choose, we will need to specify an order at some point, either by policy or in the software, because we have dozens of Wikipedias getting bigger and bigger by the day. It might be worth filing a bug to try and start some discussion on this. The more I think about it, the more I'm coming round to alphabetical order by Latin transliteration. Trilobite 05:37, 23 July 2005 (UTC)

[edit] Language sorting, round two

Moved from Talk:Www.wikipedia.org template

Trilobite, I'm a bit uncomfortable with the revision that you've made. To me, seeing each and every script listed separately looks even worse than it sounded when you described it. Is there any rhyme or reason behind the ordering of the scripts now? See, with the way I had it before, there were two variables involved in the sorting of languages: the name of the language, and whether or not it was written in Latin. With your system, that second variable is no longer boolean, if you will, so things just get more complicated.

I suppose that, to the ordinary user who doesn't know each of those non-Latin scripts, our two systems make no difference whatsoever. But now, how can you argue to Node that Armenian (hy:) goes after Urdu (ur:), but before Chuvash (cv:)?

For simplicity's sake, I was quite alright with your other proposal of doing away with language segregation altogether, which is what I eventually did in my test version of the page.

In any event, we should just choose one sort method and stick to it; I think any constant reshuffling of languages will only serve to confuse the users.

 – Minh Nguyễn (talk, blog) 03:32, 22 August 2005 (UTC)

By the way, I merged some changes I made at my test version into the main test version. Hopefully none of these changes will be so controversial, but you never know. :^) – Minh Nguyễn (talk, blog) 03:43, 22 August 2005 (UTC)

Hi Minh. If you remember the situation before all the discussion about language ordering (and right up until yesterday in fact, because the main portal's ordering had never been changed), we had language names in Latin script ordered alphabetically, with Cyrillic and Greek names interspersed according to their Latin transliterations, on the grounds that these looked relatively similar, had straightforward transliterations, and could be slotted in easily. It was quite rightly pointed out that other scripts, such as Armenian, also transliterated to Latin perfectly well, and it was unreasonable to seperate some scripts but not others, just because they looked less like Latin. What I've done is gone down the route of consistent script seperation. The main consequences of this are the grouping together of Cyrillic language names right before Latin instead of strewn about according to transliteration, and the grouping together of other scripts instead of mixing them all up and ordering by ISO code. The ordering of these grouped-together scripts is necessarily going to be somewhat arbitrary, I think, because unlike alphabetical order within a particular script, there is no established sequence to work from. This doesn't mean it has to be totally random, however. You asked if there was any rhyme or reason behind the ordering. I'll give you an example: Devanagari and similar (e.g. Bengali) are sorted together, followed by a group containing other Indic scripts whose characters are more curved in appearance and don't possess a 'washing line'. I think this is considerably more logical than having all Indic scripts mixed up arbitrarily with all the other non-Latin words. Now you're right that there are problems in script seperation. I've just noticed that while in the 1000+ section Arabic and Persian are grouped together at the beginning, in the 100+ section Urdu appears much later on. This is a relic of ISO ordering and is admittedly inconsistent. We can certainly debate how best to order the scripts and refine the system being used at the moment. As for doing away with segregation, there are certainly good arguments for that, but trying to unify the sorting of all these words written in wildly varying alphabets, abjads and abugidas, some of them written right-to-left (I dread to think how we would deal with Mongolian vertical scipt!) is inevitably going to be like trying to fit a square peg into a round hole. If we go down this route I think sorting alphabetically by Latin transliteration is the only viable option. It would have the disadvantage of being Latin-centric, although it would be fairly sensible because the majority of words in the list are written in Latin script, and the issue at hand would be how best to fit a differing minority into the sea of Latin-script words. Since it appears that your test portal doesn't slavishly follow ISO, but sensibly corrects for anomalies like Estonian/Esperanto and Finnish, I think I'll refrain from discussing the "ISO only" idea for now, which I would really object to. One last thing: I'm not sure it's a good idea to get rid of the tooltips. Lots of people have poor Unicode support, and being able to put their mouse over a word they can't see and get "Samskrta" is going to be more helpful to them than "????????". I agree with your enlargement of the 100+ section. I had wanted to do this myself but thought someone might object on the grounds that it gave undue prominence to Wikipedias with a very small number of articles. Personally I would have all three lists in the same size text, because I think the numbered headings make it clear to the reader that at this stage they're not going to get the same amount of content in Breton as they do in French. Cheers. Trilobite 19:32, 22 August 2005 (UTC)

[edit] Language sorting, round three

Moved from Talk:Www.wikipedia.org template

Samuel, before moving the Moldovan and Mongolian links to the Latin alphabet part of the list, please read our (rather lengthy) discussions above. Anyways, just moving those two links would be quite inconsistent, since you left many other Cyrillic alphabet links in their places. – Minh Nguyễn (talk, blog) 07:19, 10 November 2005 (UTC)

Aha! Very well, it isn't a terribly big deal. As long as people can find what they're looking for... Sj 12:34, 21 November 2005 (UTC)

[edit] Russian Wikipedia will have 100K articles soon

According to this prediction, it'll be in August. Shall we add it to those around the logo? MaxSem 18:46, 27 July 2006 (UTC)

How? It doesn't look to me like there's room. Why not keep it as top 10 languages? Looks better that way. --Spangineer^[en] [es] (háblame) 19:29, 27 July 2006 (UTC)

Right: most of the top 10 editions were featured around the logo long before they reached 100,000 articles, so getting to 100,000 isn't why they're up there. – Minh Nguyễn (talk, contribs) 04:51, 28 July 2006 (UTC)

OK, I got the idea. MaxSem 11:22, 28 July 2006 (UTC)

How many languages do you think we need before we add a 100,000 article section. Jeff8765 03:26, 31 July 2006 (UTC)

I'd suggest including the top 10 editions and Russian in a new 100,000-article section, so that the list won't be so small. That would also leave room for a million-article section in the future. The 10,000 section is only going to get longer over time, so we need to add a higher-level list and maybe – maybe – drop the 100-article list. That way we also don't have to figure out how many blue book images to place at the beginning of each section. :^) – Minh Nguyễn (talk, contribs) 08:44, 8 August 2006 (UTC)

Actually, there is room at the bottom of the logo (in between the Portuguese and Spanish Wikipedia links) to add the Russian wikipedia to those around the logo. --FreshFruitsRule 21:03, 15 August 2006 (UTC)

I tested it out and on this revision of www.wikipedia.org template/temp I made a version which has the Russian Wikipedia in between the Portuguese and Spanish Wikipedias (on the bottom of the logo). I tested it out using that hixie site, and it appears to work well. The permanent link to the version I made at temp is at: [2] --FreshFruitsRule 22:44, 15 August 2006 (UTC)

It looks alright, but it's only a temporary measure. The Chinese Wikipedia already has over 80,000 articles, and I don't think it'll make sense to keep adding to the already-overcrowded logo section. I think we should keep the logo section limited to the top 10 editions and create a new 100,000+ section that contains English through Russian. It'd be a bit redundant, but it allows for expansion. – Minh Nguyễn (talk, contribs) 03:37, 16 August 2006 (UTC)

So, it finally happened. Let's decide something... MaxSem 17:10, 16 August 2006 (UTC)

• I think the best option is to keep the top 10 around the logo, and then do what the English Wikipedia does: break the list up into 50k+, 25k+, 10k+, 1000+, 100+. The shorter each section is, the more readable and less cluttered it is. --BRIAN0918 19:15, 16 August 2006 (UTC)

• I agree with Brian0918 that the Russian Wikipedia should not be put around the logo (it would look too cluttered), but I think that we should at least add Russian to the search part of the Template, as that can fit. This would make the search section on the front page look like this (with the Russian term for "Search" added beside "buscar" and the Russian language added to the drop-down list:

FreshFruitsRule 00:02, 17 August 2006 (UTC)

• I think what Russian Wikipedia should be put around the logo, because is no rule about top 10. Also the quantity of big wiki (100k+) still is not too much... --Swix 05:30, 17 August 2006 (UTC)

  True, there isn't a hard-fast rule about the Top 10, but that's how we've always done it, and this practice prevents overcrowding of the logo area. If I remember correctly, when we switched to this portal design, most of the languages in the logo area didn't even have 100,000 articles yet: they were included only because they were the Top 10.

  Although the addition of the Russian Wikipedia doesn't pose a big problem at the moment, the Chinese Wikipedia is getting quite close to the 100,000-article mark: what will we do then?

   – Minh Nguyễn (talk, contribs) 07:34, 17 August 2006 (UTC)

Dear colleagues, I would suggest to redesign the logo a bit. I think it would be a good idea to add Russian and Chinese wikipedias, because a picture with Russian cyrillics and Chinese logographs would visually express the international nature of the project better. Now only Japanese hieroglyphs do (others are written in latin script). To save space, we could try to remove articles number information, considering that it have to be constantly updated, what is not good for a logo. Dr Bug (Vladimir V. Medeyko) 05:59, 17 August 2006 (UTC)

The English Wikipedia recently had some discussion about representing the project's international nature on the list of language editions at their Main Page. The problem is figuring out what languages would expresses this international nature best. Some would propose displaying Cyrillic and Chinese characters as you do; others might argue that Devanāgarī (Hindi etc.) and Arabic and Hebrew would serve just as well. But we only have a little space around the logo, and I think that including Russian and Chinese due to their different writing systems would only lead to overcrowding in this top section, because everyone will believe their language deserves a spot.

I believe the ten Wikipedia editions with the largest number of articles deserve their placement around the puzzle ball, because they've worked hard to generate all these articles. If anything, the fact that most of the Top 10 languages are from Europe indicates that our non-Latin editions have a lot of work to do.

Also, note that the Japanese ideographs 日本語 are also valid Chinese ideographs, and that the links and article counts that surround the puzzle ball are not part of the logo: they surround the logo.

 – Minh Nguyễn (talk, contribs) 07:34, 17 August 2006 (UTC)

Just in case, Russian and Chinese both are United Nations official languages (two out of six) --217.67.117.64 19:43, 17 August 2006 (UTC)

Please don't thread responses like that; it makes it much harder for others to know who said what and participate in the discussion. And Arabic, as it happens, is an official language at the UN as well. Placing 13 languages around the logo would make the page look a bit cluttered. – Minh Nguyễn (talk, contribs) 02:10, 18 August 2006 (UTC)

In case you didn't realize, Russian is a European Language as well. --FreshFruitsRule 11:31, 17 August 2006 (UTC)

Yes, I realize that, but I was responding to the suggestion that the languages that surround the logo should be restricted to ones that convey our internationalization, by including a wider variety of writing systems. Again, the Top 10 languages are there because they've worked hard to become the largest ten editions. If we use verifiable numbers (such as ranking or article counts) as our criterion for inclusion, we avoid all sorts of POV discussions.

For a brief period of time, I think the language list at the bottom of en:Main Page listed not only the ten largest editions, but also the ten most-spoken languages worldwide. That was a novel idea, but many of the most-spoken languages happen to have very small Wikipedia editions, so linking to them so prominently would be unfair to the Wikipedia editions that represent fewer people but have more articles.

 – Minh Nguyễn (talk, contribs) 02:10, 18 August 2006 (UTC)

Perhaps if we were to make the logo larger in size, we could increase the amount of languages surrounding the puzzle ball to 15 so that we could list the top 15 languages instead of the top 10, which would include Russian, Chinese, Finnish, Norwegian and Esperanto as well as the current top 10. --FreshFruitsRule 11:51, 17 August 2006 (UTC).

Increasing the size of the logo might introduce issues with lower-resolution screens. Some people still use 640×480, for example. – Minh Nguyễn (talk, contribs) 02:10, 18 August 2006 (UTC)

A couple of things:

• Could you please add Russian to the search box as indicated above.
• So if we don't put Russian around the logo, it will now be forever stuck in the honorable "10 000+" zone below. May I suggest we create something like "100 000+, but language not elite enough for the logo" zone, or something to that effect. Csman 17:50, 17 August 2006 (UTC)

My proposal above was to create a 100,000+ section that included all the languages with 100,000 articles or more, since "100,000+ except the ones shown above" is hard to convey in a multilingual fashion. I viewed this as a nice compromise, because the Top 10 "elite" still got prominent placement around the logo – which I think looks kind of neat – while the other 100,000+ languages aren't stuck with much smaller editions like Georgian.

Regarding placing Russian in the search menu, what do we do when a couple other language editions reach 100,000 articles? Remember that we also have to put each language's translation for "search" right above the search box, and the more languages that are listed above that box, the more it just looks like a big jumble of unintelligible words (to non-hyperpolyglots, that is). To me, the best solution for the search box would be to get rid of the "search" translations and place a small magifying glass image to the left of the search box.

 – Minh Nguyễn (talk, contribs) 02:10, 18 August 2006 (UTC)

How about Image:Magnify-elian.png? – Minh Nguyễn (talk, contribs) 02:36, 18 August 2006 (UTC)

Not sure I understand. Does the dropdown menu go away too with this magifying glass? If you are concerned about "search suche..." on top, we could just put those words into the existing dropdown to have "Search in English/Search in French..." instead of the current plain "English/French...". That way, all those words on top of search box will be nicely hidden in the dropdown and everyone will still know it is search. Csman 06:35, 18 August 2006 (UTC)

That sounds good to me. – Minh Nguyễn (talk, contribs) 23:43, 18 August 2006 (UTC)

I think we must put all lang editions around the logo and in search box, who reach 100,000 articles, because they've also worked hard to generate all these articles. --Swix 05:01, 18 August 2006 (UTC)

I agree. The current design gives some languages way too much credit and others almost none, creating an impression some languages are not "kosher enough" as I sarcastically mentioned earlier. Everyone works hard on their WP, it just didn't catch on early enough in some countries. Csman 06:38, 18 August 2006 (UTC)

Hence my proposal: keep the Top 10 ring as is, but add a 100,000+ list to reward the editions that have worked hard enough to reach 100,000. – Minh Nguyễn (talk, contribs) 23:43, 18 August 2006 (UTC)

One of Russian users created a logo with the Russian text in the bottom. In my opinion, it in no way looks cluttered. More, I feel, it's even better than the current logo: with the Russian text in the bottom, the iwiki texts look like laurels of fame, while current logo looks cropped. Take a look, please! Dr Bug (Vladimir V. Medeyko) 09:12, 18 August 2006 (UTC)

That's a great implementation! And for those people worried about more languages having 100000+ articles, the only one that's close to it is Chinese, and it still has about 1-2 months to go. The rest of the languages will get there no earlier than 6 months from now and I think that there'd be more than enough time to add each new "major" language to the top of the front page.

Also, having that section reserved for the Top 10 makes it look like a competition for which language has the most articles, and this would definitely influence the creation of poor quality stubs to increase those article counts.

I say we don't encourage this kind of behaviour, and set up everything so that the users would concentrate on the quality and not the quantity.--Lenev 14:44, 18 August 2006 (UTC)

I agree with the comment above, and that having only the top 10 languages featured on the www.wikipedia.org page makes it seem like a competition for which language has enough articles to be featured on the main page, and that will influence the creation of poor quality articles. I believe this could be resolved by listing more major languages around the logo or by listing the best quality language editions rather than the best quantity language editions. around the logo. --FreshFruitsRule 23:08, 18 August 2006 (UTC)

I can say I agree with both Lenev and FreshFruitsRule that listing the top 10 language editions will make it seem like a competition, but I have to say I object to listing the best quality editions, because although that is a good idea, how would we measure the quality of a language edition without bias? The only statistical way I can think of is measuring the length of the articles in each edition, but that would not be a good method because an article that is long isn't necessarily good quality. --216.106.103.3 23:15, 18 August 2006 (UTC)

I'm not saying that Vladimir's proposal has any problems now, but I think we should also plan for the when the other languages get to 100,000. Every time we move these languages around, we likely confuse scores of users, forcing them to look around for their language again. We have to move languages around when we promote them, of course, but let's not change the layout too often. Say we move Russian up to the top, then move it back down again when Chinese or some other language reaches 100,000. Yes, it won't be for a few months, but it's still going to happen, and we'll be sending our visitors searching each time we shuffle things around.

If we create a complete 100,000+ list, we make the changes more predictable. I figure we'll eventually have to create this list anyways, once having 100,000+ articles is as ordinary has having 10,000+ articles is now.

Incidentally, I think listing all the 100,000+ wikis around the logo encourages the creation of low-quality articles (such as those year stubs...). If we list only the Top 10, a wiki (let's call it "A") might be able to rise to the Top 10 by creating stubs, but with that method A will quickly fall out of contention, because the wikis that focus on quality don't rely on stubs so much; they'll quickly outpace A. We saw that happen, to some extent, with the Italian Wikipedia: they created an article on every French and Italian placename and quickly rose to fifth (?) place, but other wikis soon caught up, sending it back down to eighth place. If we list all the 100,000+ wikis around the logo, a wiki that creates several thousand stubs to get to 100,000 will remain at the top even if it never grows beyond that.

 – Minh Nguyễn (talk, contribs) 23:43, 18 August 2006 (UTC)

I suppose, we should elaborate a picture, that will include wikis in all languages, but in font size according to the wiki's size (logarithmic articles number). I will allow us to remove the article count, while preserving information about the wiki size. I think, this will solve all the problems above. And as a temporary solution, while we are creating the new design, I'd propose to add single Russian link for PR reasons: this may do some hype in the Russian media. What do you think? Dr Bug (Vladimir V. Medeyko) 22:15, 23 August 2006 (UTC)

So then, what should we do? We could do as I suggested above and place the wikipedias with the highest quality around the logo and then order them by quality. This would encourage the creation of many high-quality articles rather than many low-quality stub articles, which would in the long run help resolve Wikipedias greatest problem: Low-quality articles. --FreshFruitsRule 23:59, 18 August 2006 (UTC)

How do you suggest we gauge "quality"? We do have up-to-date statistics on the number of words at most of our projects [3], but the numbers for the English Wikipedia haven't been updated since June, and Erik Zachte's statistics often aren't updated for months at a time, making them a bit less reliable than the article counts. Anyhow, I like your idea, but we'd have to refactor the entire page quite a bit. For example, we'd have to replace the book icons with some kind of image that conveys the notion of "word". Any suggestions? Also, we'd have to publicize this discussion a lot more if we want to make a substantial change like abandoning the article counts; I don't think the community would appreciate us changing the page like that without wider consensus. – Minh Nguyễn (talk, contribs) 00:31, 19 August 2006 (UTC)

People, these are all valid points and we definitely need to continue this discussion on the quality/quantity problem. However, the Russian Wikipedia has already reached about 101000 articles and it's still in the 10000+ list.

Now we have two options, to put it near the logo OR to create a 100000+ section for the non-Top 10 wikipedias. Soon we may be adding the Chinese & Finnish wikipedias to that list.

For now I will try the second option and we'll see how it looks, but I have a feeling that Russian's going to look pretty lonely at least until the other languages catch up.

P.S. I'd like to hear people's opinion on this.--Lenev 19:37, 21 August 2006 (UTC)

I've edited the Temp (http://meta.wikimedia.org/wiki/Www.wikipedia.org_template/temp), put the code into a HTML generator and look.--Lenev 19:58, 21 August 2006 (UTC)

Russian does look pretty lonely right now. Would anyone be willing to try my suggestion and include the Top 10 languages in that 100,000+ list as well? It's just ten extra links. :^) – Minh Nguyễn (talk, contribs) 19:14, 22 August 2006 (UTC)

Yeah that sounds good, try it.--Lenev 14:24, 23 August 2006 (UTC)

I've tried it on the temp page, and it looks good to me. I'm a little concerned, though, that the row of book icons for the 100,000+ list might be too long: it might wrap to the next line when viewing the page with lower resolutions. – Minh Nguyễn (talk, contribs) 16:46, 23 August 2006 (UTC)

I've decreased it by 1 image (from both sides) and I don't think it would be a problem for any resolutions anymore. Personally, I think it's good to go. --Lenev 19:30, 23 August 2006 (UTC)

Looks good. Csman 22:16, 23 August 2006 (UTC)

Thanks for working on this, Lenev. I think it looks good, but I'm not comfortable making these changes live, since I'm a relatively new admin and it's clear from the discussion above that there are some users here who disagree with this approach. I'll publicize this at Meta:Babel. – Minh Nguyễn (talk, contribs) 05:29, 24 August 2006 (UTC)

Don't you think each opinion (excepting administrators, of course) represents a community of wikipedians of a language? Do not hesitate to approve! Mashiah Davidson 01:30, 26 August 2006 (UTC)

Some time ago I had a discussion here [4] on the amount of links surrounding The Globe. It looks like if the independent member is ok with extended design of the logo. Mashiah Davidson 01:50, 26 August 2006 (UTC)

The best solution for the moment Chineze wikipedia reach 100 000 is to move english at the top of the logo and shift other languages. The final logo gets having 12 items like a clock-face. Note that en:Sexagesimal tightly connected to Babylon. Mashiah Davidson 01:50, 26 August 2006 (UTC)

That's exactly why I've hesitated to change the portal: some people want Russian (and eventually Chinese) added around the logo, and others (including me) would like to have a 100,000+ list instead. I've pushed for the 100,000+ list because it's inevitable that a lot of languages will eventually reach 100,000 articles, and continually stuffing them in the logo area isn't a very future-proof solution.

Also, many of us at Meta are speaking as individuals, not as representatives of a whole Wikipedia community. Certainly we've got embassadors etc. here, but I'm certainly not speaking for the Vietnamese Wikipedia community (which never had a discussion on the subject), and Dr. Bug and Csman probably aren't speaking for the Russian Wikipedia. These are our personal opinions, which is why I've sought more opinions.

 – Minh Nguyễn (talk, contribs) 07:08, 26 August 2006 (UTC)

You know that in countries like Russia there are a lot of people who don't feel free writing English. This is the reason why we may assume that each person using english rarely represents their compatriots here.

I am a bit surprised your actions are caused by disagreement. I see that we have two opinions each is well-argued. Is this enough to put to the vote?

What do you think about the question: "What is of higher priority twelve around or two lonelylanguages?" Mashiah Davidson 15:12, 26 August 2006 (UTC)

This question don´t looks in the future. The decision we make now should solve the problem for the next years. It makes no sense, when we discuss each time a new language reached the 100K level. I prefere the idea of Minh Nguyễn with the Top 10 around the ball and a new 100,000+ section. So it looks like Wikisource and Wikiquote with the top 10 solution too. By the way: for wikibooks I prefere a Top 10 solution too. -- W like wiki 16:24, 29 August 2006 (UTC)

(Breaking indent.) Wow, I hadn't even noticed the Wikibooks portal! That page is a strong argument in favor of my proposal; it shows just how messy the Wikipedia portal could become if we keep adding more languages around the logo. Mashiah, my proposal is not to have "two lonely languages", because the 100,000+ list will include the Top 10 editions, meaning that the list will contain 11 links. – Minh Nguyễn (talk, contribs) 21:11, 29 August 2006 (UTC)

I'd say it is actually a pretty strong counterargument, since it shows that other project's admins think this looks good enough to go on the portal. But that doesn't matter (構いませんよ), I don't mind your proposal -- it will give us (ruwiki) another incentive to grow and another goal to achieve. Csman 18:03, 30 August 2006 (UTC)

Hey, we could put the languages in a circle around the Wikilogo. English at the top leaving a place for Russia, China at the bottom. Well that's just an idea, I'm not a web page design expert though... --Yamakiri 11:00, 7 September 2007 (UTC)

[edit] New 100,000+ section

Okay, a week after my post at Babel and with the support of at least three others here, I've added the complete 100,000+ list (which includes Russian). This approach is a bit redundant, since it provides two links for each of the ten largest editions, but this is done to make the lists more predictable in the future. Feel free to argue away; this portal is still far from the perfect way for visitors to select their language. – Minh Nguyễn (talk, contribs) 01:55, 31 August 2006 (UTC)

I think it's great. Keep up the great work! :-)

James F. (talk) 23:11, 31 August 2006 (UTC)

Wow, hard working. After a long discussion like this I was a bit afraid, that it will last for ever without any changing. Good work. Good decision. Tanks Minh Nguyễn, Lenev and all the others. -- W like wiki 23:23, 31 August 2006 (UTC)

Yeah, this portal tends to accumulate lots and lots of talk, such as the discussion about language sorting last summer that never really got resolved, and the earlier attempt at merging Catherine's design with Forseti's that never went anywhere. :^) – Minh Nguyễn (talk, contribs) 04:39, 1 September 2006 (UTC)

I concur; thank you for your good work! Korg + ⁺ 14:46, 5 September 2006 (UTC)

[edit] Chinese Wikipedia will have 100k acticles

Chinese Wikipedia will have 100k acticles. I personally think that is is the right opportunity to put the link of Chinese Wikipedia (and Russian as well) around the logo. I don't think that the rule should be fixed forever that only top 10 languages can do so. Did you know that there are more than 1.3 billion people using Chinese as their first language? So, as Chinese Wikipedia has a new milestone, I think it is the time to break the old rule. -- Kevinhksouth 09:43, 12 November 2006 (UTC)

Did you read the discussions above? Soon it won't be practical to include every language that has over 100,000 articles around the puzzle ball. Then we'll have to make a decision: limit the number of languages in the logo section to 12 (which would seem very arbitrary to the possibly vocal Finnish, Norwegian, and Esperanto speakers who will get left out) or reduce the number back to ten (which means eventually removing Russian and Chinese). Or we can redesign the page, but the last attempt to update the portal design didn't go very far. So which would you prefer, if we include Russian and Chinese in the logo section? – Minh Nguyễn (talk, contribs) 20:53, 12 November 2006 (UTC)

I think it is reasonable to add Russian and Chinese in the place which contains major Wikipedia languages (around the logo in this current design). 145 million people are first-speakers of Russian, while 1.3 billion people are first-speakers of Chinese. Also, they both are the official languages in United Nations. In conclusion, because of the popularity and importance of both languages, they are worth being considered as major Wikipedia languages. From the capture above, I think 12 languages are still acceptable to be put in. Nevertheless, if most people think that it is not practical to put 2 more languages around the logo, I support redesigning the page. -- Kevinhksouth 03:12, 13 November 2006 (UTC)

While I think you have a point about Russian and Chinese being major enough, I think that having 12 languages around the logo really stretches this design, turning what's currently a very simple, friendly listing into a somewhat cramped layout. (Just two additional languages can do that.) I'd very strongly support a redesign, but you'd have to help me convince the rest of the Meta/Wikipedia community that a redesign is necessary. I'm certain that, by now, many Wikipedians have grown attached to the current design, so it'll be an uphill battle to convince everyone that we've started to outgrow it. – Minh Nguyễn (talk, contribs) 04:56, 13 November 2006 (UTC)

What you're asking for is certainly possible with the alternative that Forseti proposed a bit late into the design competition way back when. You can still see what his design looks like to the right. (To test a live preview of the design, follow the directions at Talk:Www.wikipedia.org template/temp, using this template code, replacing $1 in that code with the source code of the portal body.) Basically, at this point, the "ring" idea no longer works, because the webpage has only so much room for a ring of languages. The only other option I see is to remove "The Free Encyclopedia" or the article count from each of the Top 10 languages, so that there is enough room. But then why would we even have the languages around the logo, since we could easily just keep the 100,000+ listing below it? – Minh Nguyễn (talk, contribs) 05:03, 13 November 2006 (UTC)

22-month time is not short in my point of view. However, many people in the Wikipedia community, who are having vested interest already, do not prefer to have a change. It is too difficult to convince the community to have a new design. Sigh... Moreover, so far only Minh Nguyễn join this discussion, and other people just ignore it. (Even no Russian or other Chinese Wikipedians come out to give comments!) I think I have to give up now. I will be back when there are people suggesting having a redesign. (P.S. Forseti's proposed design is also quite good, and I think that putting 12 languages in the top area still looks pretty good. ) -- Kevinhksouth 17:00, 13 November 2006 (UTC)

People usually don't pay attention to this page. I posted a message about this issue at Meta:Babel some time ago, but got no response. Maybe you could try there, or at the Russian and Chinese Wikipedias' embassies or "Village Pumps". – Minh Nguyễn (talk, contribs) 09:05, 14 November 2006 (UTC)

I don't think we should redesign. But if we do limit it to ten languages over 100,000 articles, then I think we should have the languages with the greatest number of speakers in those ten spots. So, even if Finnish and Esperanto make it to 100,000, I think we'd be well justified in not including them in the ten featured on the default page. But keeping Russian and Chinese off simply because they arrived late is absurd. If anything, Wikipedia is far more valuable to these countries (and their hundreds of thousands of speakers) than otherwise groovy languages like Dutch.—w:en:User:Perceval

Using this criteria (and the List of languages by number of speakers) we would have the following list of languages on the front page: Chinese, English, Spanish, Portuguese, Russian, Japanese, French, German, Italian, Polish. Dutch and Swedish (beautiful languages though they are, and likewise accomplished in their article counts) would be left off. I think this is both practical (as far as our aims of spreading information) and reasonably fair.—w:en:User:Perceval

I guess that's fair. (And if article count is no longer the only criteria for listing a language at the top, then we should get rid of the article counts in that section.) However, changing the criteria for the Top 10 is almost like changing the page's design, so you'd need to propose your idea on a well-traveled page and solicit opinions there, before any administrator here would be willing to make that significant a change. – Minh Nguyễn (talk, contribs) 07:47, 5 January 2007 (UTC)

I emphatically believe that the Forseti design should be implemented. It's worlds better than the current design, and it scales. Andre (talk) 07:07, 24 November 2006 (UTC)

[edit] Suggestion: Removing the duplicate wikipedias from the mainpage

I think the top ten languages wikipedias (English-Svenska) should be removed from the "100 000+" list since they are also listed around the wikipedia-logo above. As far as I know, they've only been listed so that the #11 place (the Russian wikipedia) wouldn't be "lonely" in this category. Now there are sixteen editions with 100 000+ articles, #11-#16 should be enough so that the top ten isn't needed anymore to be redundantly listed twice. --85.176.237.171 08:07, 12 January 2008 (UTC)

I agree. Waldir 05:45, 13 January 2008 (UTC)

Quotes from above:

• "I'd suggest including the top 10 editions and Russian in a new 100,000-article section, so that the list won't be so small." (User:Mxn)
• "The Russian Wikipedia has already reached about 101000 articles and it's still in the 10000+ list. Now we have two options, to put it near the logo OR to create a 100000+ section for the non-Top 10 wikipedias. Soon we may be adding the Chinese & Finnish wikipedias to that list. For now I will try the second option and we'll see how it looks, but I have a feeling that Russian's going to look pretty lonely at least until the other languages catch up." (User:Lenev)
• "Russian does look pretty lonely right now. Would anyone be willing to try my suggestion and include the Top 10 languages in that 100,000+ list as well? It's just ten extra links. :^)" (User:Mxn)

and from from #New 100,000+ section:

• "This approach is a bit redundant, since it provides two links for each of the ten largest editions" (user:Mxn)

With all this, I think it is safe to update the list by removing the redundant links. If noone pronounces on this for some time (as usual, sigh), I'll remove them myself from the /temp page. 87.196.212.165 18:05, 23 March 2008 (UTC) (I didnt notice I was logged off. signing now.) Waldir 18:07, 23 March 2008 (UTC)

[edit] rethinking the top ten

I'd like to relaunch the top ten discussion. I have read this talk page from top to bottom, and I think I can summmarise the ideas presented here (under my personal opinion, evidently -- I hope you don't mind :P )

Starting off, I think that the criteria for showing around the globe as a 100K+ article count should be definitely put out of question. The list will just keep growing bigger and bigger. It has been noted many times that this solution is far from ideal, and besides there's no written rule regarding that, so people will keep asking why are some wikipedias not there, until we define a clear set of rules of inclusion. A top-ten approach is robust and scalable (If we choose a relative rather than absolute measure, we avoid stuff like "a wiki that creates several thousand stubs to get to 100,000 will remain at the top even if it never grows beyond that"). We just need to figure out which criteria we'll use to generate the "top ten" list. However, it would perhaps be useful to keep 100K+ as a threshold to avoind strange results from some of the criteria below. Here's a quick summary of the proposals made, spread across several sections of this talk page:

• top ten by article count (current) -- rewards effort, but might create competition, encourage stubs
• top ten languages (Minh Nguyễn, Perceval) -- usability (many users from those languages would quickly find their wikipedia), internationalization, most of them have over 100K, but unfair to hardworking top ten by article count that are left out.
• top ten by average article lenght (FreshFruitsRule, 216.106.103.3, )-- not a good measure of quality, and people could still try to trick this. Alternatively, 1 − Stub-ratio could be used.
• top ten by highest ratio of speakers per article (me) -- probably fair, since wikipedias in languages with few speakers that manage to reach a reasonable size are worth being promoted, but strange results would appear, especially if we don't include the 100K threshold. Alternatively, users per article ratio could also be used, perhaps with better results.
• tag cloud? (Dr Bug) -- I personally dont find them readable, but they became famous for a reason. plus, all wikis could be listed by alphabetical order making it easier to find a specific language. Not sure how to order non-latin characters though. We should probably use the Interwiki sorting order

I'll make a table to allow comparison among these different ideas. Till there, I'll be expecting your thoughts on the matter. Waldir 05:44, 13 January 2008 (UTC)

So happy to see that someone has relaunched the top ten discussion. I personally support top ten languages most, since it is the best criterion to determine which languages are "top". The option which I oppose most is top ten by highest ratio of speakers per article. As you can see, Volapük, which has only 30 speakers, will be the toppest, but Chinese, which have nearly a biilion speakers and over 10k articles, will be very low and never get into top 10 (or even top 50). -- Kevinhksouth 07:57, 14 January 2008 (UTC)

I'm compiling some data here, please edit the page and/or provide some feedback so the page gets ready for moving to a proper discussion place. Waldir 23:05, 14 January 2008 (UTC)

I personally like the idea of a tag cloud. As weighting criterion for the font size I would use an averaged index of the four criteria article count, number of native speakers, average article lenght and number of users. Stubs would no longer be encouraged, and it would generally be difficult to trick this index because of its diversity of criteria. (Additionally I would limit the tag cloud to the top 150 Wikipedias (by the averaged index), so that the others could then only be found in the "Other languages" link (already now the 39 Wikipedias with less than 100 articles can only be found there)).

If however we stay with the current list and the top ten, we should at least use the averaged index for the top ten. Marcos 03:10, 22 March 2008 (UTC)

(Additionally I would use "number of active speakers" instead of "number of native speakers" for planned languages, as the number of native speakers is quite an irrelevant figure in their case.) Marcos 03:22, 22 March 2008 (UTC)

Ok, I moved the table to the main namespace: Top Ten Wikipedias. I'll spam the village pumps of all the 100+ wikipedias to gather a wide set of opinions. Please make further comments there. Waldir 18:53, 23 March 2008 (UTC)

Update: I left a message in all 100K+ village pumps, except zh, tr and fi (couldn't figure out where/how to write -- and indication of a need to make those pages more international-friendly?) Waldir 19:40, 23 March 2008 (UTC)

Impressive work :) I myself would incline towards usability criteria. The mainpage is not used to pump egos, but to efficiently serve the users. Thus, combining the criteria of (a) number of articles (perhaps as a threshold) and (b) number of active users speaking the given language could be considerable. Also, with IP maps being pretty well developed, one could think of delivering different versions of the mainpage to different IP ranges (we could also take into account the language used in the operating system)... Pundit 01:55, 24 March 2008 (UTC)

You might want to approach the developers with that idea, but it definitely won't work the way we have things set up right now (just one version of the portal). Actually, the various national portals could alleviate this issue by limiting the number of choices. (For instance, Switzerland, Austria, France, the Netherlands, and Germany.) If we got enough of these set up, the main portal could simply link to these portals. Unfortunately, the concept of national portals is currently tied to national Wikimedia chapters, and there are only so many of those. Also, many of the country-specific domains simply direct users to one Wikipedia edition (Hungary, Poland, Argentina, Israel, and Australia). – Minh Nguyễn (talk, contribs) 19:37, 6 April 2008 (UTC)

Of course, you're right. Even if the local domain opens .wikipedia.org content (like here), we need a universal approach. I'd personally go for usability understood as displaying encyclopedias basing on the minimal number of articles (100,000? 200,000? 300,000?), and as a second criterion, on the number of native speakers of a given language, probably nothing else... Pundit 20:30, 6 April 2008 (UTC)

I moved a message posted here by User:DonaldDuck to Top Ten Wikipedias --Waldir 17:39, 6 July 2008 (UTC)

[edit] Sorting by kind of alphabet

This was posted earlier today on the discussion page about the top ten articles.
If you visit the frontpage and you wish to know if your language is there, do you know how much articles have been written in that language? No! You have to search the page if you speak li for example. All those strange languages through each other don't help much. What do you know about a language? That are the characters of some kind, we have the Cyrillic alphabet, Latin alphabet, Japanese writing system, and others. A lot of languages originate from the same root, to make it easier to find the right language, I would suggest grouping bij alphabet/writing system would help user most which look for their language. Romaine 18:25, 5 July 2008 (UTC)

I second this idea. Especially if the current poll happens to change the criteria in which the top ten are ordered, there will be less sense in organizing the wikipedias by size (afterall, it doesn't make it any easier for readers to find their edition. --Waldir 18:45, 5 July 2008 (UTC)

(sigh) We used to use a hybrid approach: first by number of articles, then by writing system. See /Language sorting for the lengthy discussion. – Minh Nguyễn (talk, contribs) 08:44, 7 July 2008 (UTC)

[edit] Italian Wikipedia

The Italian Wikipedia (language code: "it") should be the 6th , why it appears to be the 10th? --137.204.148.73 10:44, 10 October 2008 (UTC)

That's because we sort it by how popular it is (how many people/times it is accessed), not by number of articles. See Top Ten Wikipedias & the poll. Cbrown1023 talk 20:54, 10 October 2008 (UTC)

[edit] Translation of Search input

Hello. User:Albamhandae translated "search" into Korean. but he removed today. I'd like to know whether "search" can be translated every language or specific language which have 0.1 million article at least in their project.--Kwj2772 08:48, 14 October 2008 (UTC)

Korean should not be included in the search box yet. According to ko:특수기능:통계, the project only has 76,299 articles. (It has over 100,000 pages, but not all pages are articles.) – Minh Nguyễn (talk, contribs) 01:49, 18 October 2008 (UTC)