Talk:Top Ten Wikipedias

From Meta, a Wikimedia project coordination wiki

What problem do we want to solve[edit]

I must admit I am at loss to what problem we want to solve with this discussion. Why do we want to show the language editions at all, they do not exist on the standard wikipedia logo, and to what purpose do we want to promote the biggest language editions? And what is special about the number 10? We have one super edition and two very big and threafter a bunch. Perhaps the whole idea has grown obsolete when it is full go everywhere? Anders Wennersten 20:37, 23 March 2008 (UTC)[reply]

In my understanding, it is on the main page where we've got the link to 10 wikipedia... Zil 21:32, 23 March 2008 (UTC) (from fr)[reply]
(moved from below)
I might have missed something, but why ten articles? Wikipedia exists on over 200 different languages, so why should just the ten "best" (largest/best quality/largest language or whatever) be viewable? Shouldn't it be better to have a list with all languages more easily accessable, maybee in alphabetic order, so it becomes easier to find one's own language? //Moralist 13:18, 24 March 2008 (UTC)[reply]
Well the idea here is to provide the community with an opportunity to decide upon a set of criteria to rule the inclusion of the current design (since many have opposed the "biggest" argument, which has never been discussed anyways -- even though it might get endorsed from this discussion), not proposing new designs. If we go that way, the discussion would be endless. Waldir 01:42, 24 March 2008 (UTC)[reply]
Perhaps the perceived problem is that one of the prestigious top 10 positions is held by the Swedish Wikipedia, such a small language (10 million speakers) and a Wikipedia with so many stubs. Even if nothing is changed in the definition, this problem will solve itself soon enough as position 11 is held by Russian and 12 by Chinese, both of which are growing faster and have fewer stubs. The Swedish Wikipedia has fallen from a 6th position to the current 10th and is expected to continue to fall behind Turkish and some more languages. A change of definition that would make sense is to change from counting the number of articles to the number of bytes in current versions (which can be estimated from looking at the database dumps called "pages-articles.xml"). This would serve Wikipedias with longer articles. (I personally don't think the UTF-8 encoding matters much for byte count.) I don't believe in basing the ranking on the number of speakers of the language. It makes the understanding of the ranking so much more complex and it would give a high rank to the Arabic Wikipedia, which has 422 million speakers, but having a rather small Wikipedia.
The table below show the 25 largest Wikipedias and Arabic. Arabic is the 31st Wikipedia, not the 26th. Click the little boxes to sort on different columns. --LA2 02:29, 24 March 2008 (UTC)[reply]
Language Article count Size in megabytes
of database dump
en 2 300 098 3500
de 727 311 1100
fr 638 258 783
pl 483 070 355
ja 479 155 674
it 429 824 478
nl 422 907 315
pt 368 050 263
es 345 327 465
sv 280 440 171
ru 252 090 310
zh 170 139 216
no 158 393 120
fi 156 002 149
vo 115 015 12
ca 109 400 93
ro 106 209 63
tr 104 064 80
uk 99 058 69
eo 96 071 58
cs 92 338 110
hu 90 425 126
sk 87 902 56
da 82 820 64
id 79 024 57
ar 56 489 47
But is this ultimately a question of prestige? There has to be a 10-best list at the beginning (like gold, silver and bronze medals in the Olympics) so that the communities who have done the best jobs can feel better? Or is it a question of usefulness -- let's have the list of 10 most useful projects (criteria to be discussed -- e.g. amount of trafic of actual users?) so that users can find the one they need faster? These two possibilities wouldn't yield the same list of languages, or perhaps even the same format for displaying them. --Smeira 17:21, 26 March 2008 (UTC)[reply]
Good list I'd say and good criteria as well. Is it possible to publish it somewhere on regular basis? Mashiah Davidson 11:19, 7 July 2008 (UTC)[reply]
I think you'd better ask that to LA2 directly :) He might not be watching this thread. --Waldir 14:37, 7 July 2008 (UTC)[reply]

Page layout[edit]

(moved from main page)

I suggest to move the serach box on the side of the logo; this would allow to bring link to other languages in a more visible part of the page. Making the discussion about the criteria for top ten much less important. -- 22:19, 30 March 2008 (UTC)[reply]

If you don't want to participate in the top ten discussion, then don't. But please don't try to add more open issues to this discussion, or to change the original one, as this will only make the discussion longer, less readable and above all less productive. You are free to start a separate discussion for commenting a redesign of the page, and I'd actually support the idea -- in a proper place. Waldir 01:47, 31 March 2008 (UTC)[reply]

Necessity of Top 10 vs. full table listing[edit]

(moved from main page)

And do we really need to have the Top 10 section? What for? No criteria will change the fact it's just showing off and no criteria will remove the envy or claim of the right to be there of those which didn't make it in. If we'll apply any of those methods above, we'll have to rearrange the rest of items regarding the same criteria as well.

Why don't we simply make sortable table (pretty much copy from the List of Wikipedias) with columns fitting the (or most of the) criteria above so people can sort in order they really want/need to instead of having pre-served order which complies with certain group of people but other groups not? Going further we could use cookies to remember user's settings of sorting order for next visit. Going even further we can have show/hide columns with criteria (also stored in cookies) as well. Cookies can also store preferred and/or last used language to highlight/put on the top such language mutation. Etc. etc...

Besides the full table is the only equitable solution.

Danny B. 03:40, 6 July 2008 (UTC)[reply]

Once again I ask: we are not reinventing the wheel here. The discussion was always from the beginning about finding a better criteria for the top 10 than the number of articles. If you want you can launch a discussion about a redesign of the main wikipedia portal, but I recommend you to read carefully template and its archives before, as this has been attempted many times before. It is very hard to do such major changes. If we can't make it perfect, let's at least make it better :) Have you voted already? ;) Waldir 17:13, 6 July 2008 (UTC)[reply]
We are not reinventing the wheel? And what do you think we are doing now? There will be no better criteria, because all criteria described in proposals are always relative or manipulatable thus never can be fair to all involved wikis. And since they never can't, we are _exactly_ reinventing the wheel since we'll get again to the situation when somebody will get unsatisfied with the criteria and will call for another poll. We have a proverb for that: "From mud to puddle." Vicious circle. Unless objective, absolute and measurable methods will be used.
I don't know if I am going to support the poll which isn't set fair. The question 1 option B description is manipulative and not correct and neutral. It's apparently preferred result thus it's being pushed towards to voters. I also don't see mentioned anywhere the exact method of how it's going to be measured, so I can easy doubt about it's going to be measured as much correctly as possible. You also haven't noticed all 100k+ wikis about the discussion, so lot of people couldn't tell their opinion even though their wikis are involved. So lot of things why I think this poll isn't set good and fair.
Btw: I also don't feel comfortable about your move of this section away from main page somewhere to the dump where it's hard to be found and easy to be missed. I've put it to the main page because it is linked from the poll directly so it was easier to compare the opinions.
Danny B. 02:33, 7 July 2008 (UTC)[reply]
First of all, I understand that you're not comfortable with the moving of the thread. But it needed to be done -- as you can see, many people have added similar thoughts before in various places and for the sake of consistence I kept joining them in one place. Your comment isn't the first one I move from the main page to the talk; If I haven't done so the discussion would have degenerated and become untrackable. So, sorry, but I feel it is necessary to perform some of these moves from time to time.
Secondly, yes you're right, the changes proposed are not perfect -- but let me disagree and say that they are indeed better. The arguments are all over the discussion, and to say that these don't count and that we're virtually staying in the same place is diminishing and offending to all those who lost their time discussing these issues. Sure, better solutions could be engineered. But from the analysis of the historic discussions at template I felt they propably wouln't be feasible, and chose to start with a small move first. I believe that one degree at a time makes us progress, perhaps slowly, but then someone else can take the next step. If we try to make too big a change most probably the community will simply reject it and all effort will be lost (and no degrees of progress will be climbed -- surely a very small amount of progress is preferred to nothing, right?). The criteria can be manipulated as you say, but that's not the question here: the current one can too, with bots for example. The improvement is that unmanipulated, they are better, for the reasons explained all over this discussion, which I won't repeat here.
Third, you might be right about the unfairness/bias of the poll options (in any case, voters cannot be pushed into voting for a given criterion -- we are all smart people, with free will - if you weren't influenced/convinced by the option's descriptions, why would others?). But this was left in discussion for a long time and I expressed below that I wanted more feedback about the proposed vote before actually starting it, and only did so after several people told me to.
Why only now you are commenting on it, I don't know, but please don't say that it was because I didn't warn all 100K+ wikipedias about the discussion: I clearly expressed in template#rethinking the top ten that I posted a message to the village pumps of all but two of these, leaving them out simply because I couldn't figure out where to write due to lack of instructions. I even pointed what language editions they were so other people could take the initiative to post the message themselves. So let me say that I also don't feel comfortable with you saying that I didn't warn all wikipedias, making it look like you're suggesting I didn't make the effort to (or perhaps even worse, that I didn't want to). --Waldir 10:09, 7 July 2008 (UTC)[reply]

I never said you didn't warn all Wikipedias. Please don't put in my mouth such words I didn't say and don't say I'm suggesting something when I'm not. I said you didn't notice all involved Wikipedias. That means those which are over 100k or little less assuming they'll gain it soon. Top Ten Wikipedias has started back in January. Cs Wikipedia has reached 100k in the middle of June. But it was obvious during the earlier time, it's going to reach it pretty soon, such as esperanto and some other. You have noticed cs Wikipedia on July, 5th after the poll has started. So there was no chance from cs users to join the discussion they didn't know about (please keep in mind, nobody is required to follow Meta's recent changes).
Re the unfairness - I would consider it fair if there was only the short description of method without stuff like "portal as navigational tool" etc. - this is pushing of the opinion to voters and can influence them (well, we all good know, that people are mostly lazy to read long discussions and they take pre-compiled summaries).
Also marking the number of visitors as "usefulness" is pretty strange. The bigger wiki the more bots operate on it the bigger number of visitors it they create. Also (consider very simplified data for this case) English is spoken by say 1 billion of people. English Wikipedia is being visited by say 100 millions of them, that's 10 %. Now eg. Czech has say 16 M of speakers. Czech Wikipedia is being visited by 3 M of them, that's +- 18 %. But just because we'll never gain 100 M of visitors (simply because of not having such number of speakers) does that make us less useful? Usefulness is in how many people of those who can use it really use it and not in absolute number of visitors. English Wikipedia is not useful for anybody who does not speak English and so it works for other languages. That's why you can't compare absolute number of visitors. (Please omit volapük as the strange exception which should be dealt specifically.)
By the way, what difference is going to be considered as enough between options to pronounce one of them as elected?
Danny B. 15:00, 7 July 2008 (UTC)[reply]

1. First of all, let's clear up this issue:
  • You said: "You also haven't noticed all 100k+ wikis about the discussion"
  • I replied: "please don't say that it was because I didn't warn all 100K+ wikipedias about the discussion: I clearly expressed in template#rethinking the top ten that I posted a message to the village pumps of all but two of these, leaving them out simply because I couldn't figure out where to write due to lack of instructions"
  • Now you said: "I never said you didn't warn all Wikipedias. Please don't put in my mouth such words I didn't say and don't say I'm suggesting something when I'm not. I said you didn't notice all involved Wikipedias. That means those which are over 100k or little less assuming they'll gain it soon."
I think reading carefully the lines above shows that you're the one twisting what you said. But nevertheless, you're right, wikis that were about to turn to 100K should have been warned -- and I haven't done so. I didn't think of that, but yes, it would have been a good idea.
2. As for the "people are mostly lazy to read long discussions and they take pre-compiled summaries", well, sure, I understand, but if you read the discussions you'd have to agree that even though biased, those descriptions are indeed a fairly good summary of the discussion occurred here, since the discussion itself always leaned towards option B. It had many arguments in favor of it, and few against, and yes that was reflected in the summary for people who don't care to read the full discussion (they'd get the same basic perception anyway).
3. You make a good point. This also applies to the "most spoken languages" option: some editions simply don't have enough speakers to ever be able to show up there. But I think of this as the democracy rule: the most good for most people possible. Imagine, to simplify the reasoning, that we had a only one position available and two languages "competing" for it, English and French. If an hypothetical 90% of the people who visit the portal speaks English, we wouldn't want to replace it with the French edition because the 10% French visitors represent 50% of the French population and the 90% English visitors represent only 5% of the English population.. we'd still be benefiting 10% of the portal visitors instead of 90%. It all ends up to considering what's best for the visitors, not for the wikipedias (besides, relative measures always bring up strange results such as the Volapük you referred ;), which would have to be trimmed off by hand, in a very unfair manner).
4. About the consensus cutoff: well in the poll I only explicitly stated a simple majority rule, and nobody contested it (except you :)) but of course if two options are too close (say, 45%-55%) I believe no decision would be made. But the steward who closes the vote will surely have experience with such issues and as such this isn't something I'd consider we'd have to be worried about :)
Waldir 22:23, 7 July 2008 (UTC)[reply]

Re 1: Yes, let's clear up: You said: "So let me say that I also don't feel comfortable with you saying that I didn't warn all wikipedias, making it look like you're suggesting I didn't make the effort to (or perhaps even worse, that I didn't want to)." - that was what I was reacting on. You were not assuming a good faith, especially in the second part of the sentence. So come on, who is twisting and not reading carefully? ;-) Anyway, let's omit this item...
Re 2: It should have been done equal - that means either same type of "summaries" and descriptions for all items or none. Voting must have equal opportunity and conditions for all possibilities regardless on previous talk otherwise it isn't regular voting but manipulated. I am seriously disappointed that this important poll hasn't been set in strictly neutral way but it's very apparent what is preferred result. :-/
Re 3: I am not going to talk about hypothetical situations. We have situation of ten positions so I'm solving ten positions, not one (which is really extreme, thus incomparable). Now, I'm still missing the clear and exact definition of criteria to be used for option B. And I think the voting shouldn't have started unless all criteria had described the exact and only way how they'll be measured. People are signing carte blanche now. Speaking about what's best for visitors - one of the best approach is to serve the content regarding on accept language which would in this case mean probably having such language on designed position (I'd suggest eleventh emphasized position in the middle underneath the logo) and having such language in search box pre-selected, maybe some other tweaks as well. That's what I call best for visitors. Because it's best for bigger majority than simple counting on majority based on number of visitors or speakers. Even better approach would be using cookies to remember emphasized and pre-selected language and the best of the best would be allowing customization of top 10 positions, because different people have different charts of top 10 languages they use. (Eg. Czech users would most probably use major world languages - en, fr, de, ru, es, but also slavic languages since they're close to each other, especially sk is very close to cs so pretty much every cs user uses sk. But sk doesn't pretty much have a chance to ever make it to top 10 according to proposed criteria, nor will other slavic languages do. Same case would be for other language families / related languages.). Make people choose their own top 10 - that will be the best for visitors. Re Volapük - that's just an extreme exemption, which shouldn't ever be considered in random counts anyway.
Re 4: Well, I think this is pretty important and major issue on very exposed page thus I'd expect there should be very major consensus reached for the change. Absolute minimum of 2/3 but rather 3/4 of votes. So yes, I am strongly oppose to simple majority because it wouldn't show the consensus, and the result could be very fragile (simple majority is also n : n+1).
Danny B. 04:18, 8 July 2008 (UTC)[reply]

1. Your're right! even though I clearly referred the 100K+ wikipedias in my reply, in the last paragraph I did use "all wikipedias". I admit I was a bit driven by emotions there: I was somewhat replying to what I felt was a somewhat aggressive tone from your message. "Twisting" also was not a good choice of words (from my part, of course)... well as you say let's put this one behind :)
2. I know, and from the beginning I admitted that the descriptions weren't exactly neutral... but as I said, I waited a long time for some feedback in the voting proposal I made and I only had a few comments in the first days (I did warn most people involved in the discussion), but afterwards only requests to start it or inquiries about that. I've been trying to keep this discussion alive and readable since January (even though I am clearly not the only one interested, nor I am trying to make other people support my own opinions, since the very option B wasn't suggested by me -- I changed my mind from the concepts I had initially and adopted that measure because I felt it was the best possible for the current design of the portal), and you will agree that without help (nor criticisms) I wouldn't be tempted to work much harder in the proposal I made previously, before starting the vote. I don't know how ethic will it be to change descriptions after it has started, though. But you can make an attempt, and if nobody objects, neither will I (unless I feel your words are biased the other way, that is :))
3. Well, that situation is not that hypothetical: consider the tenth position, and two different languages (Say Chinese and Dutch, to be more realistic, though I don't know whether the numbers would confirm the example). With the percentages I gave above, I still maintain that it would be best to be of use (i.e. to have the link for the visitor's preferred language edition of wikipedia displayed more prominently) to the most number of visitors possible, not to the visitors of the language that happens to have more people visiting wikipedia. we'd be kind of moving the "merit" criterion on which the current measure is based, from the wikipedias to the languages. Instead, the traffic proposal's idea is to eliminate merit altogether from that "top ten", using a pure measure of usefulness for the most people. Your alternative proposals are all good ideas (except the eleventh emphasized position, which would require a redesign, something that I've seen many failed attempts to do -- in fact, the "eleventh" position was already proposed before, even though it was intended to follow the same criteria than the other ten). Unfortunately, though, they cannot be added after the vote has started :( I had to remove an addition made right the day after the vote has started -- it sucks, but that's the way things work most of the time: when there's only a "mere" discussion, few people care to participate; but when it's gotten more serious, with a vote, the attention of more people will be dragged and consequently there will be new suggestions and ideas that didn't occur to the people who participated in the discussion. The vote cant be stopped, and a new discussion made, as that would seriously damage the process by making most people lose interest, adn thus we'd have less representativity in a second (perhaps third?) vote. A change proposal will never be perfect. But fortunately everything can be changed again some time from now. Many people have expressed their unhappiness with the current criteria, and if the same happens with the new ones (in case the vote actually does change anything), there will be grounds for requesting a new change. You can count with my participation on that :)
4. Yes, sure, as I said I didn't even think much of it and thus haven't added a proper definition of what majority we should expect. I'll specify a 2/3 ratio, and if that's not enough, I'll let the steward closing the discussion decide. Cheers, Waldir 09:43, 8 July 2008 (UTC)[reply]


Could you include a caption? For example, what does 1-stub ratio mean? AFAIK, de-wp does not rate articles as stubs. -- 21:20, 23 March 2008 (UTC)[reply]

Indeed the captions might be unclear. But in my opinion the table would become too cluttered if the complete descriptions were given. I'll try adding tooltips. Waldir 01:38, 24 March 2008 (UTC)[reply]

following comments moved from below

Could we get a legend for the table? It doesn't look very intuitive...

hi Waldir! asked on de_wiki for response... in my humble opinion it is not clearly understandable, what '+100k' means. this topic is shown 4 times and this is not very understandable for users. You should declare this more properly, otherwise it ends up in hopeless desinformation... not meant as affront! --ulli purwin 13:59, 24 March 2008 (UTC)[reply]
There are tooltips now, just hover the mouse over the captions. Waldir 13:06, 24 March 2008 (UTC)[reply]

The last two criteria in the table ("speakers/article" and "users/article") are either named wrong or listed in the wrong order ... I think, it should be "articles/speaker" and "articles/user". -- PE 21:20, 15 July 2008 (UTC)[reply]

Do we have stats on traffic ?[edit]

It might be useful to show in the table the user traffic. Zil (from fr)

There are two links regarding traffic. Since the other statistics used (Erik Zachte's ones) were somewhat oudated in some cases, the traffic figures weren't included too, but this could be done. The links for traffic statistics are right after the relevant title ("most visited wikipedias"). If you want, you can add values to the table. Perhaps we need to figure out another way to make it so it can be expanded more easily... Waldir 01:47, 24 March 2008 (UTC)[reply]

What does most visited or most internet traffic mean ? Individual visit ? Number of pages viewed ? Kbps ? with or without images ? This is not really clear for me... In my point of view, one good approach could be the number of single page in the main name space viewed by a single user per 24h (excluding bots). Zil 22:48, 25 March 2008 (UTC)[reply]

What do you mean, per 24h? I suppose you mean the average daily views, right? Also, when you're referring 'users' and 'bots'... are you excluding unregistered users? I mean, non-editors, or readers. Anyway, thanks for bringing up this issue. We do need to be clear in how we would measure each criteria. Waldir 01:29, 26 March 2008 (UTC)[reply]
Sorry, I was not clear at all 24h mean 24 hour. User mean registered (without bot flag) or IP. Every time someone (again not a bot) see a page on an average day on a specific wikipedia (page viewed per day...) Zil 15:04, 26 March 2008 (UTC)[reply]

(following comment moved from main page) The best way to evaluate wikipedias is through the traffic they generate. We can use for traffic details. For example, today 26 March 2008 is answering at the question "Where people go on" as follow:

  • - 53%
  • - 17%
  • - 5%
  • - 3%
  • - 3%
  • - 3%
  • - 2%
  • - 2%
  • - 1%
  • - 1%
  • - 1%
  • - 1%
  • - 1%
  • - 1%
  • - 1%
  • - 1%
  • Other websites - 4%

--MariusM 08:27, 26 March 2008 (UTC) (end of moved comment)[reply]

Not sure... French will generate more traffic than English to say the same thing (All the language have a different structure, length of an average word, even some are using Unicode) . And should you promote the one with bigger images ? Zil 15:04, 26 March 2008 (UTC)[reply]
I don't understand how French is generating more traffic than English for the same thing. AFAIK, alexa is measuring the number of visitors for a certain site. Please correct me if I'm wrong. A visitor of French Wikipedia is counted same as a visitor of English Wikipedia. The lenght of the average word in French or English don't influence the alexa's results.--MariusM 10:52, 28 March 2008 (UTC)[reply]

Traffic and use[edit]

contents moved to the main article

speakers / article and users / article[edit]

This is kind of confusing. Ich understand that as number of speakers (users) divided by number of articles. This should give a very low result for Volapük. Why is it on top? I would expect to see there big languages with rather small wikipedias. -- 23:21, 23 March 2008 (UTC)[reply]

I hope the tooltips I'm about to add can clarify this issue. Waldir 01:49, 24 March 2008 (UTC)[reply]
6 years and it was still wrong. Fixed. Volapük has a high ratio articles/speakers and a low ratio speakers/article (and this is a Top Ten, not bottom). Same for articles per user. --Mfb (talk) 14:54, 12 June 2014 (UTC)[reply]

legend for table[edit]

contents moved to the correct section


contents moved to the correct section (please read the previous comments before expressing you opinion to avoid redundancy)

Usefulness vs. Prestige/Reward?[edit]

I've mentioned this under the first heading above (what problem are we trying to solve), but I thought it might be worth mentioning this topic in a separate heading here. There are two possibilities here:

a. We're trying to figure out which layout would be the most useful to the highest number of users looking for information on Wikipedia. This would seem to me to be simply a practical problem: if we define usefulness -- say, the Wikipedias that are more frequently used/consulted by actual users -- then the problem is mostly solved. The more useful a Wikipedia is (e.g., the more frequently it is actually used/consulted by users in need of information), the more prominently it is displayed on the main page. (Languages with many speakers that for some reason don't use their Wikipedia very often would of course get a lower position; but I suspect this is mostly due to lack of good internet connections and familiarity with computers and/or the internet; as these factors improve, so would Wikipedia use and the position on the Main Page would become better.)
b. We want to include a reward for hard work and good thinking about how to develop a Wikipedia. More prominence on the main page is then given to projects which had more "hard work" done on them, and which led to "better results" (more actual information, a broader range of topics with good treatment, good quality of presentation -- texts, pictures, etc. --, good organization -- categories, help pages, navigation templates etc.).
c. Some compromise between a. and b., since we value both things.

Personally, I'd favor a. above. As has been said on the content page, b. is basically about rewarding the work of Wikipedians, which can and should be done, but perhaps elsewhere -- as long as the goal and raison d'être of Wikipedia is, ultimately, the user, not the Wikipedian. (Is it? Do we all agree on that?) We could, say, reserve the Main Page for the most useful Wikipedias, and create another page for the "best" or "more developed" Wikipedias -- a page in which people could actually also discuss how to measure that, and what is the 'ideal Wikipedia' that all projects should (or should not?) be striving to achieve. (I think a good part of the discussion on the content page here would be more appropriate on such a page -- since I favor usefulness as the criteria for organizing the Main Page layout). A compromise case -- c. -- might seem more palatable to more people, but I'd feel we'd be losing some usefulness here. --Smeira 17:36, 26 March 2008 (UTC)[reply]

Position of English wikipedia[edit]

(moved from main article)

Dynamic landing page[edit]

Why can't the landing page be dynamic and detect the user's default language settings in their browser/operating system, and then offer the Wikipedia for that language in a very prominent spot on the page? I imagine a page where there's a large main header at the top, offering the Wikipedia for the user's default language. Then every other language which has a wikipedia is listed in alphabetical order below. Most users are going to be looking for the wikipedia in the same language as their operating system and browser, and if they aren't, they can easily find the one they're looking for below. And then there's no politics of which wikipedia is the most prominently listed. Michael Tyznik 21:03, 2 April 2008 (UTC)[reply]

This is a really good idea, I am quite surprised noone has thought of that before! I have a few thoughts to add to it:
  1. Another variable we can use to determine the location of a user could be the IP address, many sites do that already. In fact, using all these criteria would make the concept stronger: for example, if the browser is in english, the operating system is in french and the IP says the user is accessing the site from russia, that's probably a sign that whe shouldn't give too much a prominence for any of these language editions of wikipedia. But if all three agree, maybe we could even redirect the user straight to the wikipedia on their language!
  2. The problem is we are not discussing a redesign of the page, here. Making it dynamic while still keeping the current design would mean what, putting the "suggested" wikipedia on the top-left position of the globe? (supposedly the most "prestigious" position). On the other hand, perhaps a redesign is really necessary. I'd probably be positive to dropping the puzzle globe design in favour of (one or more) dynamic top suggestion, using your idea, and then a tag cloud in alphabetical order, with the font sizes relative to whichever criteria ends up gaining the consensus in this discussion.
  3. That being said, I still wouln't support changing this discussion into a redesign one (as was suggested more than once, actually -- see the first section of this talk page), since it is already hard to reach consensus to make such a little change in the design as a slight reordering due to using different criteria. Trying to make bigger changes could render this discussion useless as many others have became before. As such, I'd suggest a new discussion about a redesign to be launched, after this one is finished.
  4. Finally, this discussion on the static design has to be conclusive even if we decide to go on with that dynamic idea, since the static version would have to be available for people without javascript for browser, OS and IP address detection.
Waldir 02:34, 3 April 2008 (UTC)[reply]
    1. In that case, why not put three big links to the Russian, English, and French Wikipediae at the top of the list?
    2. My ideal scenario doesn't have to be implemented in order to use the main idea I'm proposing. It would be rather simple to have a little banner at the top with the dynamic link as an addition to the current page rather than redesigning the whole page.
    3. Using this method would I think make consensus easier to reach. This would serve the purpose of making the home page the most convenient for the most people, so that argument can be removed from the discussion of the ordering of the links on the landing page.
    4. Javascript is not necessary for browser, OS, and IP address detection. These are all transmitted by the browser itself in the User-agent part of the HTTP request. All the processing would be done by the server, meaning that every person who accesses the home page through any browser could have the recommended link. The only people who might have a problem would be people who are using a proxy to appear as though they are located in a different country.
Michael Tyznik 13:35, 7 April 2008 (UTC)[reply]
Thanks for the clarifications. It was nice to know that server-side processing can be used. I have to question the top banner idea, though. You see, we already have some redundancy in reproducing the top links twice (around the globe, and in the 100,000+ section below it -- which is a concern that people have tried to address, but without any result so far), should we really be making it worse by adding more duplicate links? In another thought, we could remove the one, two or three wikipedias linked in the top banner from around the globe, but that would implicate a lot of processing to put the remaining wikipedias in order not only around the globe, but also in the sections, so that there is no redundancy. Of course, we could just ignore this redundancy issue, and just add the banner. But another question arises, which is: if the banner is meant to be even more prominent than the globe, then wouldn't the page be too much of a bunch of intrusive ads making the choice harder to people, instead of easier? I mean, there would be, potentially, too many things requesting their attention. --Waldir 14:14, 7 April 2008 (UTC)[reply]
I do think that simply adding this would make the landing page more cluttered and perhaps redundant, but I think that's because the page is designed poorly in the first place. Which is why I suggested a redesign above. But I do think that having a link to the probable destination of the user in a prominent place will make the site easier to use, even if it makes it look a little more cluttered. Michael Tyznik 19:10, 7 April 2008 (UTC)[reply]

I see. You have a point, but I'd like to let this whole discussion mature until we reach some consensus on the criteria. I mean, a great worth of ideas have been produced and shared here, and I think it is worth seeing where all this will lead. In any case, after this we might move on to a redesign stage :) I totally support that idea, so stay tuned! :D --Waldir 00:27, 12 April 2008 (UTC)[reply]

  • Very interesting. Why not, based upon the IP address, just blink the language link or put it, say, in reverse video, with a short welcome message alongside ? It'll be a way to draw attention without really change places around the globe. - Al Lemos 18:40, 17 April 2008 (UTC)[reply]
I don't think blinking is something acceptable in terms purely of good taste :P but what do you mean by "reverse video"? --Waldir 21:00, 17 April 2008 (UTC)[reply]
Something like this (or just see reverse video ;) - Al Lemos 16:17, 18 April 2008 (UTC)[reply]

I like the idea of dynamic landing pages by IP the best as the most user friendly. I'm sure the page layout issues can be sorted out.

This idea would also probably help some of our smaller projects. Hindi is the world's 3rd most common language yet the Hindi Wikipedia is our 56th largest Wikipedia. 4 other South Asian languages are in the top 20 on the list of languages by number of native speakers and none have Wikipedias in the top 30. Yet I'd guess 10 to 20% of en.wikipedia's edits come from either South Asia or the South Asian diaspora. Many of these editors may not even know of these other projects despite the "languages" list in the left column of each article page. The page for an Indian IP would most prominently display English (an official language) and Wikipedia versions in languages from the list of Indian languages by number of native speakers. German, Russian, etc would be elsewhere further down the page. Likewise, an Algerian IP might see Arabic, Berber and French most prominently displayed.--A. B. (talk) 20:13, 19 April 2008 (UTC)[reply]

Why pick one Top Ten metric?[edit]

Why not maintain multiple metrics (and generated page versions based on them)? Then cycle between them every so often. That may possibly finesse the "one metric is best" argument! ++Lar: t/c 02:55, 5 April 2008 (UTC)[reply]

Relevant thread --Waldir 17:18, 6 July 2008 (UTC)[reply]

This particular idea was raised briefly at template, but I don't think it has gotten enough weight/discussion yet. I think that this is the best most practical approach, and also the best to foster contributions. Incidentally, what is the rationale behind the current set-up at ? Cirt 06:29, 11 April 2008 (UTC)[reply]

Waldir 17:20, 6 July 2008 (UTC)[reply]

wrapping up?[edit]

It is approaching one month since this discussion has been open, and a few good points have been noted. I suggest that on the 30 day mark, April 23, we do some wrap up of the first part of the discussion. Which method do you propose for evaluating the chosen criterion? Personally, I reckon it wasn't a vote, but still can't ignore that the number of thoughtful comments (as well as the ratio of pros/cons for each one) do show some level of community support. Based on that, I would propose a cross-project vote, held here on meta, featuring a first question for choosing among the following options:

  • number of articles (the current criterion)
  • most visited wikipedias (the one with more comments, and most support: 7 for, 1 against)
  • size of compressed database (an adaptation of the current measure, but reducing impact of bot-created articles)

I also considered the following:

  • number of articles with a given minimum size

since it would be a simpler adaptation of the "number of articles" criterion, and would (arguably) filter out bot-created articles and stubs, but decided not to, since the non-stub size would have to be set manually, and as it was pointed out elsewhere, "meaning of stub could change from one wiki to an other".

Then, a second question would be whether mixed criteria should be used, as it was suggested more than once, and how much of them. If that passed, the two (or more) most voted criteria would be used either as a combination to yield results that would benefit both criteria, or having the page cycle between them (even though this would probably be confusing).

Finally, there would be a third question, regarding whether a visual signaling should be used to highlight the language version determined by the language of the browser, OS, and IP of the visitor.

This is evidently just a proposal, and options could be added or removed from it, as well as it could be totally dismissed. If it is approved, though, I'd suggest the vote to start on 23 April 2008, 00:00 UTC (that's 5 days from now). What do you guys think? --Waldir 21:38, 17 April 2008 (UTC)[reply]

How long would you suggest that the vote/discussion last? Seems reasonable to start soon to move things along, yes. ++Lar: t/c 23:32, 17 April 2008 (UTC)[reply]
Thanks for your wrap-up! I agree with your summary and love to add your forth point. It will reduce stub impact in general, as information not so much useful, not only massive bot creation.
I am afraid 23 April may be too a rush. How about setting a tentative start date as April 27? We need to determine the vote length as Lar pointed out as well as its suffrage (who can vote). I recommend the same criteria of steward election, and for length, at least 3 weeks. --Aphaia 23:42, 17 April 2008 (UTC)[reply]
...what is it that should be evaluated? En_wiki is much larger than de_wiki - but because of what? because of more articles?? depends on the quality... should always keep that in mind... otherwise Wikipedia comes out to be 'the most notified garbage of the world'... diggit? -- 01:04, 18 April 2008 (UTC)[reply]
Where did that third question (about highlighting) come from? Are you trying to through in new questions just before the deadline? The user's browser or OS might be locked down by the employer or library that owns the computer, and the IP location will always discriminate against any minority language in favour of the official majority language. For example, all Kurds would see the Turkish language Wikipedia highlighted and all users in Florida would see English highlighted, even though 16.5 percent of them prefer to speak Spanish. --LA2 01:53, 18 April 2008 (UTC)[reply]
LA2, we can reliably associate IPs with countries. Within a country, geolocating an IP to a particular location gets trickier. I see sites with built-in geolocation put me in locations up to 1300 km apart. So in the case of your Florida reader, they'd probably get a U.S.-customized page that would include (as the most prominent choices), English, Spanish, Chinese, French, German, Tagalog, Vietnamese, Italian, Korean, and Russian (based on the U.S. Census ranking of languages spoken at home). Based on the data at en:List of languages of Russia, a Russian IP might see Russian, Tatar, Ukrainian, Bashkir, Chechen, Chuvash, Kazakh, Buryat, Tuvin, and Yakut most prominently displayed. I' expect there'd always be a generic landing page for satellite broadband users and countries for which we did not yet have landing pages; it might take a while to implement landing pages for all 150+ countries and territories. This is just one example and there are many other ways to to handle IP landing pages. --A. B. (talk) 17:34, 20 April 2008 (UTC)[reply]
The third question came from the section #Dynamic landing page above. In fact, it was the most reasonable (IMO) option of the "dynamic portals" that have been suggested several (3, to be exact) times in the discussion both in the main and talk page. But as I said, this is just my personal opinion about what we should put on vote, and both additions and removals are welcome to be suggested.
As for the date, I agree that it might be a rush on 23. April 27 is ok for me. --Waldir 02:25, 18 April 2008 (UTC)[reply]
I agree those three parameters. And I'm thinking: if the soft detects which languages are more used in a zone (ranks of IPs), it could use that data to modify the order of the languages.--Eloy 02:54, 18 April 2008 (UTC)[reply]
Top ten languages

What about Top ten languages ? I think this is the best option. Cirt 03:22, 18 April 2008 (UTC)[reply]

I am still wondering if we are not making something simple complicated. The bottom line of these exercises will be that the Russian version takes the place of the Swedish one, and we Swedes have since long anticipating this change and will welcome Russia (who will also soon pass us in number of articles, not least becuase we are more or less deleting (redirecting) articles in our quality initiative). The possible change after this one would be to replace the Italian and/or Dutch one with the Chinese one, and I think that would be non-natural as both of these version have more than 400 000 articles. Anders Wennersten 04:58, 18 April 2008 (UTC)[reply]
I'm not clearly understand "not least becuase we are more or less deleting" part, but Russian Wikipedia use pretty strict rules for speedy (and normal) deletion of short acticles. F.e. (6 of 10 random pages selected): sv:Ningbo, sv:Cirkumfix, sv:Alv (jord), sv:Kamakura (stad), sv:Kyle Massey, sv:Richard Berg - can be labeled as {{empty}} or {{db-dict}} and speedy deleted). (Other sample - in ru-wiki two my fogotten stubs about streets has beed deleted (even with foto) due small size not looking to my status (administrator)). Other sample: ru:Обратная проводка marked for deletion only due size. #!George Shuklin 05:51, 18 April 2008 (UTC)[reply]
Well personally I just think that organizing it by Top ten languages would be simple and also most useful to the widest amount of people. It also sends a message that we are a global community of contributors. Cirt 06:48, 18 April 2008 (UTC)[reply]
George Shuklin → I think Anders means that currently is in a phase where we tend to set a higher standard with regards to quality and length of articles, wich hopefully will mean that growth of (in no. of articles) will (if not stop, at least) decrease in the near future. With regards to the more general discussion, I agree with the summary above, and I too think that dynamic landing page is an option to explore further. --MiCkEdb 09:42, 18 April 2008 (UTC)[reply]
Cirt → Top ten languages = Top ten languages, Top ten languages, Top ten languages, or Top ten languages ? :) Well, no really valid sources, and note that the german one doesn't exist... Bourrichon 12:02, 18 April 2008 (UTC)[reply]
A good point Bourrichon - and that is certainly something that could be hashed out, but I still think that it is a better metric to use than individual project statistics. Cirt 13:01, 18 April 2008 (UTC)[reply]
Anders Wennersten, As we see, several sites have traffic close to 1 per cent of the total. In this case, I would suggest that we use other metrics in conjunction. The design of the wikipedia portal is an important PR issue and we should not bind ourselves to strict numbers. However, I do not share your view that it is unnatural for Chinese wikipedia (or anything else) to replace a bigger brother if it turns out to be genuinely more useful, as can be measured by traffic. Wikipedia is after all a tool for the readers. I find your view subjective and would like to see you justify it.Hillgentleman 04:39, 19 April 2008 (UTC)[reply]

Since there are arguments for the top ten wikipedias by (non bot) articles and for the top ten wikipedias by language, why don't combine these lists somehow? If that would mean there'd been 16 instead of 10 projects around the puzzle globe, then just do it. There shouldn't be a fixed numbre of projects around that globe, only the range should be limited (min and max numbre of projects arount that globe ). --32X 20:24, 18 April 2008 (UTC)[reply]

The effect of adding ru and zh around the logo on the main page (a suggestion in Nov, 2006)
Why are we now discussing "top 10" instead of other number? It is because "top 10" seems to be a consensus and difficult to change. When ru and zh had their 100k articles many months ago, I suggested ru and zh are to be included as well, and changed "top 10" to "top 12". However, I got no support at that time. (The discussion was here.) -- Kevinhksouth 02:42, 19 April 2008 (UTC)[reply]
Actually Kevinhksouth - I think that the page looks nice with those additional 2. Cirt 08:21, 19 April 2008 (UTC)[reply]

No, no, guys, let's not make this fall apart at this point. Many points presented here are driving the discussion away from the central point and this will make eventually it split into several branches rendering it hard to read and impossible to produce a consistent outcome. I will address some points raised here:

  • Cirt: please state your comments for using the top languages in the proper section. If you want this as an option for the vote, fine -- just say that.
  • Eloy: yes, modifying the order is a possible way to implement the visual clue referred in the (proposed) third question. Of course in the actual vote this question will be properly written to allow people to have an idea what exactly would be used.
  • Anders: there is already an option in the proposed vote for keeping the current criterion. There's no need for campaigning for or against the options in the discussion; if you want to do it, use their respective sections in the main page.
  • 32X: the second option proposed above already mentions combined criteria. Please read the discussions before you post.
  • Kevinhksouth: please read the section #What problem do we want to solve above. A redesign of the portal (either removing the globe or adding more items around it) has been discussed at great length before, without any results. We do *not* want to go that way here. This discussion is merely about finding the best criteria for the order of the items in the *current layout*. Changing the layout is another discussion (I took the liberty to convert the image into a link to keep this discussion cleaner). The best way to go forward is to deal with one issue at a time.

I beg the next commenters to have the above in consideration and collaborate accordingly, so that we can reach useful results. --Waldir 15:25, 19 April 2008 (UTC)[reply]

Is it just me, or are we going in circles? OhanaUnitedTalk page 17:59, 19 April 2008 (UTC)[reply]
Precisely! that's why I tried a wrap-up and a vote, as it seems all pertinent arguments have been expressed and people are now starting to repeat each other... I suppose April 27th is a reasonable deadline to start the vote as Aphaia suggested. We still need to decide about what options to include in the vote, and the voting procedure (see Aphaia's post above for a suggestion) --Waldir 23:59, 19 April 2008 (UTC)[reply]
Waldir -- Sorry if I was not clear - I had already brought up my comment in the proper section - I brought it up again here to supplement my thoughts on why we shouldn't wrap this up. Of course if the community feels we are coming/have come to a consensus, I defer to that. Cirt 09:35, 20 April 2008 (UTC)[reply]
Thanks for the clarification, but I still don't get why we shouldn't try to wrap the discussion and make a decision. It seems to me that the most pertinent arguments have been exposed, and we need to move forward, otherwise we'll be walking in circles as OhanaUnited said... Having the top languages criterion as an option in the vote isn't enough, in your opinion? --Waldir 14:17, 20 April 2008 (UTC)[reply]
So where is the !vote? OhanaUnitedTalk page 18:37, 20 April 2008 (UTC)[reply]
Re: to Waldir - No, that would be fine, you are right, let's move forward. Cirt 20:31, 20 April 2008 (UTC)[reply]
What causes the stall in the vote? It seems like everyone just quiet down and not doing anything. OhanaUnitedTalk page 16:23, 24 April 2008 (UTC)[reply]

Vote proposal[edit]

So, let's see: does anyone object the following proposal?

1. Which criterion should be used in the main wikipedia portal ( for sorting the various language editions of wikipedia?
  • 1.a. Number of articles (the current criterion)
  • 1.b. Most visited wikipedias (usability for the visitor instead of praise to the editors; portal as navigational tool. Discussion here)
  • 1.c. Size of compressed database (an adaptation of the current measure, but reducing impact of bot-created articles — see explanation)
  • 1.d. Number of articles with a given minimum size (a simpler adaptation; filters out stubs and bot-created articles)
  • 1.e. Most spoken languages (like 1.b, intended to be useful for the most people possible. Discussion here)
2. Should only one of the above criteria be used, or a combination of them?[1]
  • 2.a. One — 1 (the most voted one, on question 1.)
  • 2.b. Two — 2 (the two most voted)
  • 2.c. Three — 3 (the three most voted)
3. Should a visual clue be used to highlight the language version determined by the language of the browser, OS, and IP of the visitor?
  • 3.a. No
  • 3.b. Reverse video (like this)
  • 3.c. Reordering of the ranking (dynamic portal)
  • 3.c.I Do noting if the "local" language version is not in the top ten[2]
  • 3.c.II Replace the editions (remove the last one of the top ten and add the "local" version in its place)
  • 3.c.III Replace the editions (remove the last one of the top ten and add the "local" version in the first place, moving all the others down one position)
  • 3.d. Different font color (e.g. like this)

Every voter could choose only one of the options for each question. The vote would start April 27, lasting for 3 weeks, as suggested above. The same criteria for the stewards election would determine who can vote. Namely, voters should (1) Have a user account created 3 months before the start of the vote (i.e. 27 January 2008) on any other Foundation project (e.g. wikipedia, wikibooks, wikisource, commons); (2) Have a user account on Meta. (3) Place links on their meta and other project(s) userpage(s) linking to each other. I further propose that the accounts, on top of being 3 months old, should have a minimum 200 edits on any project.

  1. By multiplication of the rankings — harder to cheat, but less comprehensive to visitors
  2. If it is, it will move to the top position and the others which were above it will drop one position each

I have a question: should we add criteria that has been in use in other listings, even though they got no or little mentioning in this discussion? namely "Depth" (which has been used for some time in List of Wikipedias) and # of internal links (which has been used in

In another issue: I didn't put blinking as an option for Q3, does anyone feel it should be included? --Waldir 00:47, 21 April 2008 (UTC)[reply]

I don't think it's a big loss not to include it but maybe for completeness? I would not vote for it, though. I also would like to see the "rotate among different criterion" on offer, but again, not a biggie. Given that we need to move forward, I think these questions are thorough enough and are what should be voted on. ++Lar: t/c 13:10, 24 April 2008 (UTC)[reply]
@Lar: I didn't include that as an option because I agree with the following reasoning (quote from this talk page, by User:Mxn):
"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."
I think he's right: a portal must remain consistent to its users, as that conveys stability and reliability (and usability above all).
Note that the dynamic portal proposals do not fall under this rule, since a user would always see the same version, adapted to his region. --Waldir 17:15, 24 April 2008 (UTC)[reply]
We making any further progress with this? Cirt 17:23, 2 May 2008 (UTC)[reply]
Even though I would like to have had more feedback on the vote proposal options, the voting will of course move forward (with a different start date, evidently). I haven't started it yet because I've been a little busy lately, but feel free to be bold and implement the vote with the parameters as was discussed here. I'll surely help, as time allows me. --Waldir 18:15, 2 May 2008 (UTC)[reply]
I strongly encourage you for begin the votation and then inform about the process in all "villages pumps", "cafés", etc. — The preceding unsigned comment was added by (talk) 17:38, 1 July 2008 (UTC)[reply]

Ok, I just lauched the vote, in Metapub#Top 10 Wikipedias (poll). Please help spread the word about it. And, of course, vote! --Waldir 12:22, 5 July 2008 (UTC)[reply]

Top ten languages based on page views as determined by country of origin[edit]

As I noted on the Top Ten Wikipedias page in the most spoken languages section, it seems many users are somewhat diglossic in their Wikipedia usage. 10 to 20% of en.wikipedia edits probably come from South Asia or members of the South Asian diaspora. By number of speakers, Hindi is the world's 3rd largest language and Bengali is the 8th, yet our Hindi and Bengali projects are ranked 56th and 57th. Many South Asians editing en.wikipedia may not even know we have those projects.

It should be possible to obtain approximate page views by language based on our server logs:

  1. The logs have page views by IP
  2. The IPs can (in most cases) be identified with a particular country and a language assigned to the IP. Multilingual countries (Canada, Nigeria, India, etc.) can be dealt with one of several ways:
    1. Some ISPs are confined to certain regions, so you could assign any associated IPs to that region's language. Cons: It sounds like a lot of work.
    2. You could use one of the IP geolocation services. Cons: There might be a cost to this. Last time I checked, they weren't very accurate.
    3. Allocate a country's number page views to different languages based on percentage of speakers of each language. Pros: simple to set up initially. Cons: Could be controversial for some countries. May be had to nail down the percentages; see en:Canada#Language for example. On the other hand, it should tend to average out over many countries (the exact French-English split in Canada would hardly affect the overall global rankings of these two languages).

I prefer separate landing pages by IP as noted above, but this is an alternative if the idea of separate landing pages is thrown out. --A. B. (talk) 19:18, 19 April 2008 (UTC)[reply]

Mixed criteria and those left out[edit]

Not all languages can be displayed in the Top Ten (or top twelve, or top sixteen) group, but also all language options have to be easily accesible from the very first page in wp. So I have thought of splitting languages in groups.

  1. One group for big, relevant, mostly used would be composed of a) the ones with the most visits (say 4 languages), b) the most spoken ones not included in the previous group (say another 4), c) languages with more wp articles longer than 1000 (or another amount) bytes not included in groups a or b (say another 4). These languages would be listed alphabetically (what alphabet? I don't know).
  2. All other languages have to be listed in a way that makes them available. Basically I would find embarrasing if Bengali or Hindi speakers didn't use wp because they couldn't find the way in. B25es 20:30, 20 April 2008 (UTC)[reply]
Once more I must say, the idea is valid and a good one; but this discussion is clearly focused since the beginning *not* in redesigning the portal but simply choosing an appropriate criterion for the current design (and it is already being hard enough!). I acknowledge that a redesign might be the next step, but let's take one step at a time, shall we? If we try climbing too much steps at once we might fall :) --Waldir 00:34, 22 April 2008 (UTC)[reply]

Current rationale[edit]

Forgive me if this is stated somewhere, couldn't find it just now - but what is the rationale for the present layout of ? Cirt 19:24, 22 April 2008 (UTC)[reply]

Quoting from template#Russian Wikipedia will have 100K articles soon: "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." – Minh Nguyễn 07:34, 17 August 2006 (UTC)[reply]
--Waldir 04:34, 23 April 2008 (UTC)[reply]
Hrm, so the current rationale is - there is no current rationale?  :) I guess from that above comment by Mxn, originally the arrangement around the logo was just by highest number of articles per language/project? Cirt 18:27, 23 April 2008 (UTC)[reply]
Yes... that actually was the very reason this discussion was started: to have a clearly defined rule for sorting the top ten, as the current "rule" is only somewhat "accidental"... --Waldir 23:33, 23 April 2008 (UTC)[reply]
Well, in the end, I suppose that the most important thing is to have some sort of rule that is uniform and well described. For example if we go by "largest..." something, that then this is held to in the future as the standard. Cirt 23:59, 23 April 2008 (UTC)[reply]

spread the word[edit]

This is the message I will send to various village pumps and mailing lists to spread the word about the poll. Feel free to adapt it, sign it with your name and post it anywhere you feel is relevant. Waldir 12:33, 5 July 2008 (UTC)[reply]

Hi all.

I would like to request your attention to a vote that will start this midnight, regarding a rearrangement of the top ten Wikipedias that are displayed on the main wikipedia portal (

This topic has been wandering around for a long time on template, coming to surface in many occasions, especially on the times around the milestone of 100.000 articles of the Chinese and Russian Wikipedias.

After a tentative wrap-up of all the proposals made in that page throughout the months in template#rethinking the top ten, a discussion was launched in Top Ten Wikipedias, to which all the major Wikipedias have been invited to in their village pump.

A lot of good opinions have been collected and discussed, and a vote proposal has been made and received some feedback. That proposal was now implemented on Metapub. Please head to the poll to vote. I hope to see you there! --Waldir 12:33, 5 July 2008 (UTC)[reply]

Sorting by kind of alphabet[edit]

Ok, it is not much the top ten, but I would like to mention it and suggest a different approach for the front page lower section.
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 14:05, 5 July 2008 (UTC)[reply]

That's a good idea. It kinda falls out of the scope of this proposal (since we are dealing with the top 10 ordering criteria only) but it is a good topic to be debated afterwards. I also believe other similar ideas have been proposed in this discussion and/or in template. Not sure though. --Waldir 17:39, 5 July 2008 (UTC)[reply]
I posted it on the page you mentionted too. Perhaps it is better there indeed. Thanks! But the idea does possible influence the top ten maybe. Greetings - Romaine 18:28, 5 July 2008 (UTC)[reply]

Nv × Na[edit]

I'm sorry if this has been suggested before.

The purpose of the landing page is to be as helpful to as many visitors as possible. That means that both the number of visitors a Wikipedia gets and the number of articles it has to offer are important. I propose multiplying those two numbers as the selection criterion. This would assure that only large Wikipedias with many visitors are allowed to make it to the top ten.

I wouldn't replace the number of articles with the database size, for the following reason. I'd rather have 100 medium-sized articles on 100 different subjects than 50 large-sized articles on just 50 subjects leaving the other 50 with no coverage whatsoever --- when you're thirsty, half a cup of water is much closer to a full cup than it is to no water at all. — AdiJapan  15:19, 5 July 2008 (UTC)

Your point is good but I don't think multiplying them is a good metric here. Instead I'd prefer that we come up with a completeness number for each WP (based on article counts, filtered article counts, or whatever), then group by completeness and order each group by viewership. -- 17:47, 5 July 2008 (UTC)[reply]
See Top Ten Wikipedias#Combined ratios --Waldir 17:27, 6 July 2008 (UTC)[reply]

Poll is bloating Metapub[edit]

Please move it to a subpage and link to it from Metapub.  — Mike.lifeguard | @en.wb 15:53, 5 July 2008 (UTC)[reply]

Done. Thanks for the notice, I didn't notice that at the time of the posting. --Waldir 17:36, 5 July 2008 (UTC)[reply]

Vote maintenance[edit]

I just removed an extra option added by to the poll. His reasoning makes sense, but unfortunately the poll questions cannot be changed after the vote has started. I reproduce the text below for reference:

====Sub-option D.4====
Replace the Wikipedia at the top of the page with a link to the "local" version (preferably chosen by detecting the browser language rather than the location). Add "The Free encyclopedia" in the local language below the Wikipedia.
  1. Filceolaire 11:39, 6 July 2008 (UTC)[reply]

Also, I'd like to ask the voters to pay attention to the requirements. I had to strike out some votes from users that didn't comply with the requirement #2 (having an exixting user page on meta linking to the main project they work on). Waldir 17:04, 6 July 2008 (UTC)[reply]

Localization method[edit]

Judging from current status of the poll it seems like a localized top ten list might win on option 3. This opens up a new question; How do we determine which language is local? As far as I know there are generally two methods; checking the language of the web browser or checking the origin of the IP. I am a Swedish guy living in Spain, so I use a Swedish browser on a Spanish IP. I get localized content all the time, and then mostly in Spanish on pages like Yahoo, Facebook and advertising, while Google for instance give me Swedish content. For Wikipedia, I see three possibilities:

  1. Use browser info
  2. Use IP info
  3. Use them both (move two languages to top ten if needed)

Personally, I'd prefer the third one as I'm checking things up in both Swedish and in Spanish all the time. There is one technical problem I've left out from the equation this far, and it's bilingual areas. For instance Barcelona, where I'm living, is in great part Catalan (and my guess is that cawiki users would be offended if eswiki was the standard) but the really official language is Spanish. It's an everyday political issue over here and quite charged. I don't know how IP localisation works, but I guess the IP is just identified as a country and given that language, which would also give problems in Switzerland with their four official languages. Browser info would almost always take you to the right spot, so it might be better.

Ok, so what's your input? I couldn't find earlier discussions on this topic, but if there is, please hit me with a stick and point me in the right direction. - Axelv 20:14, 22 July 2008 (UTC)[reply]