This discussion revealed a problematic nature of Wikipedia quality standards. Not that I've something against these, on the contrary. But it became evident that sometimes there is a technical problem to stick to them. Israel article is a featured article, meaning that it meet at least most of the quality standards of Wikipedia. However, that's what get it stuck. It appear that it take a lot of time to many editors to download it. This way, the article is practically less available for reading. Reducing the number of reference templates in it could reduce its ranking. That's a paradox which I'm sure Wikipedia is not in favor of. I suggest to set maximum allowable number of citation per statement, maximum number of templates or at least, reference templates and further ruling to avoid these cases. The article on Israel is not the only article suffering from this problem, I understood that USA article and G.W Bush article also suffering the same problem. Also, I suggest this ruling will be made with special consideration of FA articles (which are also prone the most prone to this kind of problems), such that their status wouldn't be affected without additional reason.--Gilisa (talk) 16:35, 10 May 2010 (UTC)

(copying my comment from the other discussion) Putting an arbitrary cap on the number of citations an article can have isn't really the best approach. For one, it will just mean that fewer sources are used or that less detail is given in sources in order to combine them. But the loading speed is also dependent on other factors like infoboxes, nav templates, and number of links, images, and categories. A better idea would be to reduce the overall size of slow-loading articles - The Israel article is more than 9,000 words and the PDF version not including the references, bibliography, or external links is 20 pages. Or, alternately, create less-complex versions of citation templates. I've proposed this several times but no one ever seems interested. Simpler versions would have only the commonly used fields in the templates and not use the overly-complex {{Citation}} or {{Citation/core}}. {{cite journal}}, for instance offers (at least) 39 parameter options, but the majority of uses probably use less than half of those. Mr.Z-man 17:11, 10 May 2010 (UTC)

I think Z-man's right. The citation templates are the most used. A good article can easily have over a hundred. We need to do a review of the main citation templates and see what percentage the various parameters are used and then determine a long term plan for those parameters. Or, a less popular suggestion might be to create a bot that finds a fully populated (and by fully populated, I do not mean one that uses every parameter but one that has all appropriate data for that source and hasn't been edited in awhile) and converts it to a regular, text citation. I use the citation templates because they are there and I have no desire to memorize how to format various citations.—NMajdan•talk 17:28, 10 May 2010 (UTC)

I agree with you both, and I really liked the Bot idea of NMajdan. I think that the Bot should also consider the time it take the server to upload the article. How can we promote this idea further?--Gilisa (talk) 17:31, 10 May 2010 (UTC)

It appears that the speed also depends on the parameters used. I currently have 400 different very simple instances of the citation template (name, title, year) in my technical sandbox as a test. The time spent on these by the server is absolutely negligible. Therefore I think that what we need is probably not simpler versions of the templates but an analysis of them to find out which parameters cause the problem, so that we can avoid them or they can be fixed.

I think much of this discussion is premature. Let's find out all the facts before doing anything drastic. Hans Adler 17:38, 10 May 2010 (UTC)

There were some problems with my test. I am no longer sure that it depends on the parameters so dramatically. Hans Adler 19:48, 10 May 2010 (UTC)

Quite. Where's the evidence that the citation templates are the cause of some pages being slow to download? All the tests I've seen appear to indicate that they have very little impact indeed, fractions of a second at worst. Malleus Fatuorum 17:41, 10 May 2010 (UTC)

Take a look at the original discussion at WP:VPT (link at the very beginning of the topic). An editor took the content of Israel and put it in their user space. The page took 45 seconds to load. Then, they broke all citation templates and the page then took less than 10 seconds to load.—NMajdan•talk 18:19, 10 May 2010 (UTC)

I did basically the same test as in the VPT thread. I made 2 copies of the Israel article in my userspace (minus categories and a couple templates that added categories). In one copy, I removed the {{reflist}} and the bibliography section (which consists entirely of citation templates). By removing the reflist, the references are still in the article text, but they never get run through the parser (I verified this by reading the code for the Cite extension). In 5 tests on each, with references took on average (43.631 ± 3.776) seconds, while the version without references took (7.701±0.712) seconds. Only once did the version with references take less than 40 seconds and the version without references never took more than 10 seconds. See also bug 22134. Mr.Z-man 18:50, 10 May 2010 (UTC)

I have experimented a bit more. This was complicated by the strange fact that as soon as I copied Barack Obama into my user space, poof!, the problem was gone. Or rather, it wasn't as easily measurable as I expected. When I saved the page I had to wait as long as expected, but the HTML comment of the page served indicated an unrealistically short time for generating it. When I reloaded the page it was again very fast. What I have to do to get the realistic times:

• pick out a version other than the latest one from the page history and load that, or
• append "?action=purge" to the URL.

Once I knew how to test, I tested the following:

• 200 simple citations using the "citation" template: 15 seconds.
• 400 simple citations using the "citation" template: 35–40 seconds.
• 500 simple citations using the "citation" template: 35–50 seconds.
• 800 simple citations using the "citation" template: 55 seconds.

With 900 citations I get a timeout and can't measure the time. With 1000 citations the same happens, and when I look at the page later the last templates are not properly handled. They are displayed as a link to the article citation. Hans Adler 19:48, 10 May 2010 (UTC)

For context, looking at some of the longest articles, 2009_flu_pandemic_timeline has 611 unique citations (only some of them are templated); so, I think very few articles run into a template run time problem in practice (and many of them probably have other stylistic problems, e.g. WP:SIZE violations. --Cybercobra (talk) 01:09, 14 May 2010 (UTC)

List of allied military operations of the Vietnam War has been under discussion. It uses no citation templates, one named reference used 1043 times and 581 single use references. I can open it only when logged out and it takes over a minute; logged in, it times out. Apparently being logged out causes a fresh version to be served, but I never did understand the details there. So— I don't think it is just the citation templates, but possibly the Cite parser. There are several open bugs that are related to the way that cite.php handles data; see T22707 and T18330. ---— Gadget850 (Ed) ^talk 00:09, 14 May 2010 (UTC)

In that case, it probably has less to do with the references and more to do with the sheer size of the article. The wikitext size of Israel is "only" 168 kB, that list article is more than twice that, 438 kB. It is currently the largest article on the site. It results in a 1.4 MB HTML page. It also has links to over 2000 pages that need to be checked for existence. Those 2 bugs are about nesting refs, which isn't really relevant to performance. Mr.Z-man 15:26, 16 May 2010 (UTC)

Hello,

I am currently brainstorming an idea for an automated linking bot:

Such a bot should take care of the following things: - see that any related and nontrivial articles are linked - not link any unrelated articles

In order to estimate triviality ( a.k.a. is an article "common knowledge" ) the bot would make a word frequency analysis of the English wikipedia, possibly matching similar sounding words with a SOUNDEX or similar algorithm in order to eradicate differences between British and English spelling. In order to establish whether two articles are related, the bot would look at a) how many categories do the articles share b) how big are those shared categories ( The category "living persons" is huge and therefore doesn't really imply relatedness, whereas the category "Fundamental Particles" is quite small therefore implying a lot of relatedness c) how close together are the articles, a.k.a. how many clicks do you need to get from one article to another d) how similar are the word frequency distributions in both articles, what "rare" words do the articles share e) possibly more

The bot has an offline snapshot of the Wikipedia for word analysis/relationship analysis etc. For every page in the Article namespace the bot does the following:

1.) Fetch page 2.) Eliminate all words that are not the name of a Wikipedia article 3.) Eliminate all words that superficially appear to be unrelated and are quite frequent ( set a threshold ) or on a list of unrelated subjects ( that shouldn't be auto-linked) and that ARE NOT IN A LINK IN THE CURRENT PAGE 4.) For each of the remaining words see if the current page and the article on the word are related. Words that appear in a link ( e.g. if there is one link already in the article) are automatically related. 5.a ) If not process next page 5.b ) If they are, search the first appearance of the word in every section, with longer section in every paragraph maybe , and place a link to the word's page on that word.

This ensures that wikipedia isn't spammed with links, but interesting articles still get their links.

I would implement and possibly run such a bot... Masterfreek64 (talk) 19:33, 10 May 2010 (UTC)

An interesting idea (I think the concept is to automatically add wikilinks to articles). Isn't there a strong possibility of overlinking? Is there any demonstrated need to add links automatically? Hmmm. WP:OVERLINK does not say what I thought it did (I agree with several editors I have noticed who have reverted dubious link additions with an edit summary pointing to WP:OVERLINK, but OVERLINK is actually very weak atm). Johnuniq (talk) 08:39, 11 May 2010 (UTC)

First, there was User:Nickj/Link Suggester and User:Nickj/Can We Link It, so the principle of suggesting links is well established, and if you can improve on existing algorithms, great. Two problems with the idea of automatically adding links though. One, it's going to be very hard to get a balance between overlinking and underlinking; it would require vast amounts of fine-tuning and in the process probably piss lots of people off. Two, linking from an article is both a relatively low priority task, and a rather easy one which is good for newbies to get started with. Far better to put the effort into the third suggestion I made in the section above at Wikipedia:Village_pump_(proposals)#Alternatives_to_TFA_unprotection, namely creating a mainpage box showing visitors/new editors what they can do, with wikilinking appropriately a basic and easy starter task. That said, orphaned articles are a problem, and if you could come up with a really good algorithm for suggesting inbound links, that would help. The suggestion should be left on the talk page like User:WildBot does though, not implemented automatically. Rd232 ^talk 15:48, 11 May 2010 (UTC)

I think one way to improve the suggestions is to exclude linking a single word, because of high FPs, unless the word is capitalized. Sole Soul (talk) 19:31, 11 May 2010 (UTC)

Maybe we could have the bot err extremely on the side of underlinking( approach the optimum from below) and possibly add a human voters system. The way this would work is that the bot would keep a special page where even anonymous users would be presented with a paragraph and individual words would be highlighted and they could decide if links are supposed to be added. If there were a random rotation, each link would be presented to a user once, the change put in according to this users decision according to WP:BOLD and later on one or two other random users would verify the decision independently.

A possibly even better way, which might be perfected with changes to mediawiki proper, would be to have links "on probation". The software would count every link clicked ( through referers, javascript, id parameter?) and determine if a link is used enough, possibly one could also add a button to disagree with certain links. Through the use of JavaScript, the voting logic could be kept seperate(why am I thinking of a fast, hash-table based FastCGI script here?, with a hidden iframe to send AJAX to that script whenever a user clicks a link, with a connection reset every so often for privacy). A page could either be marked on probation(count links) or as regular, and a page would be placed on probation for 1 week or 1 million views after the bot edited it. Obviously the bots editing rate would have to be limited to one re-visit per month or per 6 months unless drastic changes occured. Of course wikilinking is a task for new editors, but it is one of the tasks most likely to scare them away.Masterfreek64 (talk) 22:48, 11 May 2010 (UTC)

"one of the tasks most likely to scare them away" - why? compared to what other editing tasks? Rd232 ^talk 23:25, 11 May 2010 (UTC)

Is there some evidence that Wikipedia has a significant problem with underlinking? This really seems like an overcomplicated solution to a minor problem. Though if you want to work on an algorithm, nobody can really stop you (though running some script and recording something in a hashtable every time someone clicks a link is probably a non-starter, even a few minutes of collection would result in massive amounts of data). Mr.Z-man 03:00, 13 May 2010 (UTC)

Mr. Z-man, if he does no harm and can prove that the task is useful, I say more power to him! I am confused about how recording link-clicks in a hashtable got into this discussion. Tim1357 ^talk 20:52, 13 May 2010 (UTC)

That was part of his most recent idea. I'm not saying its harmful, but its a massive and difficult automated solution to a problem that's easy and simple to fix by hand. As Rd232 noted, there are similar problems that are much more important and difficult to fix by hand, such as orphaned articles. Mr.Z-man 14:48, 16 May 2010 (UTC)

Creating wikilinks is most certainly not a tedious task which should require automation to be feasibly executed. Furthermore, wikilink placement is an editorial decision which should be under the control of human users unless in absolutely exceptional circumstances. These two considerations are enough for the proposed bot to be taken as unnecessary and even potentially detrimental (in defence of the proposal, though, it should be pointed out that the technical challenge of implementing it effectively is quite interesting) --Duplode (talk) 04:42, 13 May 2010 (UTC)

The idea is good, but the bot shouldn't actually edit articles without human intervention. It could leave on the talk page a list of links which aren't there but should (or are there but shouldn't), though. A. di M. (talk) 13:08, 16 May 2010 (UTC)

I've created a central area for discussion regarding subtle factual vandalism, a problem I've seen become worse and worse while our general vandalism fighting abilities get better and better. Right now the only purpose is to identify and discuss examples and patterns of subtle vandalism (think date change, statistics change, vital number and formula changes, etc.) and to identify the problem. The next step is to brainstorm ideas, and the ultimate goal is to develop tools to combat these problems.

I'd invite anyone with experience/interest in vandalism patrolling to add their specific expertise and experience to what is sure to be an important challenge as Wikipedia goes forward. Shadowjams (talk) 07:59, 11 May 2010 (UTC)

I found some subtle vandalism in one article by tracking down a user's (few) previous contributions made before his getting flagged for vandalism on another page. I think the best way to catch subtle vandalism is to do some SERIOUS auditing of the previous contributions of any user flagged down for vandalism. You're not apt to randomly find things, but if somebody gets caught trying to mow down pedestrians on the sidewalk, chances are there were some other bad driving incidents in the past. Previous contributions by such users are highly suspect — or at least worthy of increased scrutiny. Carrite (talk) 01:10, 17 May 2010 (UTC)

The new editing interface was one of the big "user friendly" changes added to Wikipedia. I think its great, but a question no one has been able to answer is why the switch to Vector? The new editing interface has been added to all skins not only Vector, so why the need to change the skin? The only identifiable difference is the "look" which was much better with MonoBook. Not only that, but it apparently was composed of less bytes too. I propose we move the default skin back to MonoBook until there are identifiable, significative and noteworthy changes. RaaGgio (talk) 17:54, 16 May 2010 (UTC)

I propose that a higher quality article standard above featured articles be created. This should have much stricter standards and would be to featured articles what featured articles is to good articles. (I edited the proposal because of the responses)—Preceding unsigned comment added by Iankap99 (talk • contribs) 01:15, 9 May 2010 (UTC)

I project that 0 articles would ever qualify for it. --Cybercobra (talk) 01:18, 9 May 2010 (UTC)

Yep, none of our articles is ever "perfect" in the sense that nothing could ever be added or taken away without damaging it. Of course, that's really an argument in favor of this proposal, since it would be quite easy to administrate: every article that qualifies for this status has already been identified. — Gavia immer (talk) 01:24, 9 May 2010 (UTC)

This is almost biblical... "perfect"? God is not a wikipedian :P Choyoołʼįįhí:Seb az86556 ^{> haneʼ} 01:26, 9 May 2010 (UTC)

"But "Featured Article" doesn't have any units. It's an arbitrary scale mapping..." ~ Amory (u • t • c) 01:37, 9 May 2010 (UTC)

It would be easier simply to raise the featured standards a little more, whenever it is felt that the current ones may not be enough. MBelgrano (talk) 01:48, 9 May 2010 (UTC)

What would these "much stricter standards" be, and who would be the judge? Malleus Fatuorum 03:33, 9 May 2010 (UTC)

Maybe have good articles, great articles, and featured articles? Featured articles are already supposed to be "Wikipedia's best work." You can't exceed the best. Tisane (talk) 03:53, 9 May 2010 (UTC)

Yea, something like that, you should propose that --Iankap99 04:33, 9 May 2010 (UTC)

Call them Awesome Articles. Each would involve sound, video, and a network of subarticles. They would draw the reader in, captivate him, and eventually spit him back out, gasping for breath, an hour later. When finished, they would be reviewed by a team consisting of a genius middle schooler, a high school teacher, and a nobel prize-winner in the field. They would have glossy magazine-layout versions that are printed out, monograph-form, and distributed in the gift bags at NSF and UNESCO meetings. –SJ+ 08:32, 9 May 2010 (UTC)

Well, why not have some higher standard like that to aim for? And different versions for different audiences, especially for more technical subjects, is far from ludicrous. Rd232 ^talk 10:22, 9 May 2010 (UTC)

I'd prefer the category Super Duper Articles. Seriously, I think that as Malleus Fatuorum says, FA are meant to be the best as it is - how do you get better? Raise the bar for FA, by all means, but I don't think a new "FA+" category of pages is needed. -- PhantomSteve/talk|contribs\ 10:26, 9 May 2010 (UTC)

I will support this only if we call them Holy Shit, This Is Friggin' Awesome-class articles.
All kidding aside, I fail to see what the point of this is. If FA isn't good enough, we can up the requirements. Voila, "problem" solved. EVula // talk // ☯ // 15:38, 9 May 2010 (UTC)

The criteria have already risen enough in the last few years, but there are many articles which were featured before that and probably wouldn't be promoted if they were on FAC today. What about systematically nominating articles featured (or last reviewed) the longest time ago for FAR? (Is there a list of FA sorted by the date when they were promoted or last reviewed?) ― ___A._di_M. (formerly Army1987) 16:16, 9 May 2010 (UTC)

Yea that's true, we should do that —Preceding unsigned comment added by Iankap99 (talk • contribs)

How about "Best article in the entire Wikipedia"? Or top 5, or top 10, or something? --Yair rand (talk) 17:05, 9 May 2010 (UTC)

The "stricter standards" I mention would be simply new or more strict requirements at the Manual of Style or content policies or guidelines, and they would be approved or rejected the regular way: a user suggest something, and if gets consensus it's added, if it doesn't, it's not. But in any case, it's not a systematic process (from here to a year in the future, we may develop much more requirements than in the last year, slow down, or even stop creating new ones), so a systematic review is not a good idea. It's better to review the status of featured articles when someone actually finds problems, than merely "just in case". Or perhaps make this periodics reviews at a wikiproject, and bring the concerns to the regular process when an article with problems is found. MBelgrano (talk) 18:31, 9 May 2010 (UTC)

Something like this could easily be achieved by having a list top 20 articles by quality or something like that, and advertising it in a prominent place. It would be updated each time a new article, better than the last one on the list, passes FAC. Unfortunately I am afraid some editors would take it a bit too seriously, try everything to get their article on the list, and cause a lot of disruption. I am not sure that the benefits outweigh this cost. Hans Adler 17:20, 11 May 2010 (UTC)

Personally, I think standards creep is more of a bad thing than a good thing. It means people spend more time improving articles that are already high quality just to meet some arbitrarily higher standard. If anything, I think our quality standards could use relaxing. Mr.Z-man 20:04, 11 May 2010 (UTC)

I wouldn't put it that way! The real issue is that so much more energy tends to go into articles of more narrow interest, whereas many "basic" articles don't get that kind of attention. That may be partly because it's harder to edit those articles, but mainly it's just what excites people. "Space Age Widgetmakers: The Cartoon (episode 587)" tends to get more interest than "Widgets"... ! Rd232 ^talk 21:31, 11 May 2010 (UTC)

I could care less where the effort goes, I'd just prefer it to be spread around a little more. Most people will stop doing work once there's no more incentive. Once an article is "featured", there's much less incentive to continue improving it. But just think how many fewer unsourced articles we might have if people spent their time adding 2 or 3 sources to those articles instead of adding 175 footnotes to Bacteria or 176 references to Wii. Or how much time is spent checking a few articles against obscure and trivial parts of the MoS that could be spent improving the prose in articles in need of a rewrite. Its like we prefer quantity over quality for everything except (perhaps ironically) quality. We'd rather have a few "extremely high" quality articles than more "very good" quality ones. Mr.Z-man 23:39, 11 May 2010 (UTC)

How can we make articles better than FA status? I can't think of how I'd add to most of them.--Patton123 (talk) 14:31, 16 May 2010 (UTC)

• Great idea! Non-existent problem, meet unworkable solution. (Seriously, Iankap, we don't need this, though it's a nice idea :P.) AGK 14:29, 17 May 2010 (UTC)
• I can't really see what's wrong with the current system to be honest, so I'm going to join most others in opposing this. We should be focussing on improving the several million poor articles to an acceptable level rather than improving the 5000 best articles to an even better level. Alzarian16 (talk) 20:57, 17 May 2010 (UTC)

To whom it may concern,

Recently I heard that dark backgrounds for one's screen saves more energy than Light backgrounds, this apparently is because using a white screen requires more energy. This is why there has been a custom search created known as Blackle (http://www.blackle.com/), where the screen is completely black. Now, I know that Wikipedia is a very well known source of information and many people around the world use it. So I was wondering If we could save energy together for our planet's sake and change the main color of Wikipedia to black. It may not be big but every bit counts..and we have come to a point where a change must be done. I thought that to save the environment, I might ask you to please make this change. I know maybe this might make the layout less pleasant But I find it important we do so!

Thank You and please do consider my comment. — Preceding unsigned comment added by 202.129.235.49 (talk)

No, this has been debated many times and the community has categorically decided against this course of action. It is possible for individual users to select that layout in Special:Preferences, but it will not be made the default. MBisanz ^talk 15:12, 13 May 2010 (UTC)

Using a black background only makes a significant difference on CRT monitors, on LCD displays, there's no difference 1. Mr.Z-man 15:18, 13 May 2010 (UTC)

It's also usually hideously ugly. ♫ Melodia Chaconne ♫ (talk) 16:04, 13 May 2010 (UTC)

See our article on Blackle.com which links to references that give proper context to the claims. -- Quiddity (talk) 17:10, 13 May 2010 (UTC)

Exact duplicate threads: Wikipedia:Village_pump_(proposals)/Archive_52#Black_background_to_save_energy, Wikipedia:Village_pump_(proposals)/Archive_39#Public_Black_Background_version_of_Wikipedia. I think this might be WP:PEREN-worthy. --Cybercobra (talk) 21:14, 13 May 2010 (UTC)

That last comment deserves attention, given that exactly the same policy was proposed here in early autumn last year (i.e.

in early autumn 2009) - does any one remember it? ACEOREVIVED (talk) 20:18, 17 May 2010 (UTC)

I always find it annoying to have to go into the article to unwatch it. It would be nice to have a small link to remove an article from your watchlist appear on Special:Watchlist. - ʄɭoʏɗiaɲ ^τ _¢ 15:51, 13 May 2010 (UTC)

What's wrong with the View and edit watchlist link below the Watchlist heading?—Emil J. 15:53, 13 May 2010 (UTC)

That it lists them alphabetically instead of by the last edit, and because it requires an extra click (quite clearly I'm a lazy bastard). - ʄɭoʏɗiaɲ ^τ _¢ 16:28, 13 May 2010 (UTC)

importScript('User:Alex Smotrov/wlunwatch.js');

^^^ Add to your special:Mypage/skin.js. –xeno^talk 15:57, 13 May 2010 (UTC)

Done, it's perfect! Thank you very much :) - ʄɭoʏɗiaɲ ^τ _¢ 16:28, 13 May 2010 (UTC)

Zdroj:https://en.wikipedia.org?pojem=Wikipedia:Village_pump_(proposals)/Archive_62
Text je dostupný za podmienok Creative Commons Attribution/Share-Alike License 3.0 Unported; prípadne za ďalších podmienok. Podrobnejšie informácie nájdete na stránke Podmienky použitia.

---