The Wikipedia Signpost is an independent publication about the Wikipedia community. |
This article is not an official statement of the Wikimedia Foundation. |
A potential solution for Wikipedia's overloaded servers appeared last week, as news broke that the Wikimedia Foundation was in talks with Google about providing hosting services. Preliminary indications were promising, but no final agreement has been reached and is probably still several weeks away at the earliest.
The initial hints appeared 1 February, as an agenda item for the 7 February Wikimedia Board of Trustees meeting referred to "Hosting by Google". After the board meeting was held last Monday, the summary of the agenda reflected more clearly that this was not simply a theoretical discussion, but indicated that an actual contract proposal was being worked on and a private meeting was scheduled on the issue for 2 March. However, the information did not yet indicate whether there was any discussion on Google's part, or whether the Wikimedia Foundation was simply hoping to interest Google in the idea.
Participants in the meeting alluded to the proposal in discussions on the Foundation mailing list during the week, prompting additional questions about the plan. Wikimedia CFO Daniel Mayer said Wednesday, "The plan for Google hosting will greatly reduce the amount of money we need to spend as will other hosting offers." (Mayer later said he was not speaking in an official capacity.) In response to the queries that followed, Trustee Angela Beesley confirmed that Google had made a proposal to host Wikimedia content, but said she couldn't elaborate.
The Wikipedia community's other elected Trustee, Anthere, reported on Thursday that Jimmy Wales had met with Google founders Sergey Brin and Larry Page during a trip to Stanford, where he was a guest lecturer last Wednesday in a class taught by Howard Rheingold. She indicated that Brin and Page were enthusiastic about the Wikipedia project.
The first appearance of this information outside of Wikipedia discussions seems to have been posted on a blog Archived 2005-02-13 at the Wayback Machine Thursday at dirson.com, one of a number of blogs dedicated to Google-related issues. The Dirson blog provided no real information beyond what had been stated on the mailing list — in fact, an update to the blog copied information verbatim from mailing list posts, complete with misspellings like Zeon in reference to the Xeon servers Google might use as a hosting provider. (Zeon is apparently, among other things, a planet in the Star Trek universe; perhaps next we might see rumors that the Federation will be providing additional bandwidth in exchange for Wikimedia abandoning its plans to relocate the Klingon Wikipedia to Wikicities.)
Dirson's post, combined with the existence of a Meta article about Google hosting, meant that the information was now published both internally and externally. This was apparently good enough for other people to start repeating the story, with the Meta page serving as "official" confirmation. The story was picked up by Slashdot the same day, prompting an extended discussion on that site.
As a reflection of the combined name recognition of Google and Wikipedia, or the large audience Slashdot enjoys, the news proliferated rapidly to many different internet sites on Friday. The "Google hosting" page on Meta had to be protected due to vandalism after the news hit Slashdot, a common problem for Wikipedia articles that get linked there. Meanwhile, Brion Vibber reported that many of the people he spoke with while manning a Wikipedia booth at the Southern California Linux Expo wanted to know about the Google deal.
The coverage brought out a limited response from Google, confirming the basic fact that talks were underway but not elaborating on the nature of the discussions. As reported by Susan B. Shor of TechNewsWorld, Google spokesman Nate Tyler stated, "While we don't have anything specific to announce today, Google and the Wikimedia Foundation are collaboratively evaluating creative ways to support Wikipedia.org and its community."
Details of the proposal remain confidential, although Mayer did say that as he understood the plan, there would be "no strings attached". The Foundation did not offer an official statement, and the fact that outside sources treated information on Meta as official prompted the addition of a disclaimer to the page. According to Beesley, any official statements would be placed on the Wikimedia Foundation website.
A poll to redesign the Wikipedia portal closed last Friday, but the old design remained in place as the proponents of the two most popular new designs worked to combine elements from both proposals.
The Wikipedia portal page at www.wikipedia.org replaced the earlier redirect to the English Wikipedia several weeks ago (see archived story), prompting surprise and criticism from users of the English Wikipedia. The change appears to have been accepted now, but people have been looking for a more appealing design for the portal, as initial attempts at the layout were considered bland.
In an attempt to solve this, CatherineMunro developed a new design featuring the Wikipedia logo in the center, surrounded by a group of the largest Wikipedia languages. Below this would be a multilingual search box for these languages, followed by all remaining languages having at least 100 articles, clustered into three groups. At the bottom would be links to Wikipedia's sister projects, as in the present layout.
This effort received a very enthusiastic initial reaction on the Village pump, as virtually everyone who saw Catherine's portal design endorsed it to replace the current, basically all-text page.
However, problems cropped up with rendering the page properly in older versions of Konqueror, particularly version 3.1. Screenshots of the page in these versions showed that instead of grouping the largest languages symmetrically around the Wikipedia logo, they ended up clumped together and overlapping badly with the logo image. AlanBarrett made some efforts to improve the format to work around these problems.
Eventually, a couple of other proposals were brought up by Node ue and Forseti. Node's version was a variation on Catherine's design, but smaller and without the search box. It did not display properly and Node admitted that "it isn't exactly functional".
Forseti's design was closer to the current format, but brought the Wikipedia logo into the layout and added a border. It added some text from the Wikipedia article in the background behind the logo and offered different versions to resemble Wikipedia's MonoBook and Cologne Blue skins.
The poll on which design to use, which was held on the Meta site where the editable portal page is located, started 28 January and closed on 11 February. With the support generated at its reception, Catherine's version received the most votes, 75 in all. Forseti's version, a late addition to the poll, still managed to pick up support from 30 users over the last week of the poll. The only support for the existing version, however, came from two voters who weren't even logged in.
Given the overwhelming unpopularity of the current design, requests to implement a change started coming in even before the poll closed. No action was taken, however, and the portal wiki page itself remained protected due to an earlier edit war.
Meanwhile, Catherine, Forseti, and others discussed how their proposals might be merged to reasonably satisfy supporters of both designs. As part of this effort, Forseti posted three reworked versions that retained the MonoBook design, but moved the Wikipedia logo to the center and arranged the largest languages in various patterns around the logo. Based on this, it looks like another round of discussion will be needed to pick the final design — perhaps even another poll.
After leaving Wikipedia two months ago due to outside musical commitments, UninvitedCompany returned last week to start up a project to expand Wikipedia's collection of recorded music. The project soon developed to include recordings of performances by Wikipedians along with an effort to import suitable recordings from outside sources.
A competent amateur pianist and organist, UninvitedCompany suggested that audio clips of pieces would be beneficial to illustrate well-known pieces, and offered to record his own playing of some classical keyboard works. He uploaded his first recording, the prelude from the Short Prelude and Fugue in G Minor (BWV 558), attributed to J.S. Bach, on Tuesday.
This was soon followed by Dysprosia, who provided a recording of an Intermezzo, Op. 76 No. 7, by Johannes Brahms. Not quite satisfied with the performance, she characterized it as a "rough first recording" and promised a better one later.
The proposal also prompted Raul654 to write a script, with help from Ævar Arnfjörð Bjarmason and other Wikipedians, to upload more sound files. After some minor technical issues, the Raulbot began converting a series of recordings into Ogg files for use on Wikimedia Commons. However, the large size of some of the files caused some problems, and Raul654 said he hoped the limit on file sizes for uploads could be raised to accommodate this effort.
Initially, Raulbot started out by uploading a number of files from FreeMusic Archived 2005-02-10 at the Wayback Machine, a site containing classical music recordings, many by the MIT Concert Choir. Raul654 said Friday that he would next work on recordings of oral arguments before the United States Supreme Court, taken from the Oyez project. Files from both sites are licensed under the Creative Commons Attribution-ShareAlike license (cc-by-sa), and therefore eligible for inclusion in the Wikimedia Commons. This license was also used by UninvitedCompany for his recording, pointing out that the GNU Free Documentation License is difficult to apply to audio due to the lack of a transparent format under the terms of the GFDL.
Prior to this effort, only a very small handful of full recordings existed on Wikipedia, along with a few more excerpts claiming fair use (some recordings have also been used to illustrate word pronunciation). In addition to his work uploading audio clips, Raul654 started a catalogue of Wikipedia's full-length recordings of songs, which can be found at Wikipedia:Sound/list.
Its caseload significantly reduced over the past few weeks, the Arbitration Committee saw a number of new requests last week, but only two were ultimately accepted. A third one was nearly accepted until the arbitrators reconsidered. No cases were closed last week.
At the start of the week, the Arbitration Committee had only two open cases, one involving Robert the Bruce and the second case dealing with Lyndon LaRouche articles. Deliberation and voting on proposed rulings continued through the week, but neither case reached a point where closing the matter was considered.
With their light caseload, the arbitrators nearly accepted a request for arbitration submitted last Monday by Bblackmoor, who had been involved in a dispute with Axon over the Open gaming article. The dispute was obscure enough that an earlier attempt to conduct a survey on the underlying issues drew little involvement from the community, and a request for mediation received no response due to the Wikipedia:Mediation Committee's recent organizational problems.
Based on the arbitration policy, a case is accepted when four arbitrators have voted to do so. Accordingly, Bblackmoor’s request might have been considered accepted when Grunt changed his vote at 17:59 (UTC) on Tuesday. However, after Sannse and Theresa knott urged more use of other dispute resolution methods, several arbitrators reconsidered and Bblackmoor ultimately withdrew his request.
Continued problems surrounding CheeseDreams prompted the arbitrators to open a second case regarding her behavior, at the request of Ta bu shi da yu. The issues raised included harassment, including allegations of making frivolous requests for arbitration herself. The case was further complicated by the use of a large number of sockpuppet accounts, and the difficulty of blocking CheeseDreams due to her use of a dialup with changing IP addresses.
Another new request was brought on Saturday by ChrisO against WikiUser, claiming a lengthy history of personal attacks, legal threats, and other disruptive behavior. The problem had spread far enough that three arbitrators had to recuse themselves from the case for being too closely involved. Grunt and Neutrality urged their colleagues to accept the case, however, and the necessary votes were present by Sunday.
Although the coverage of the 2004 Indian Ocean earthquake has generally brought a lot of favorable publicity for Wikipedia, an incident last week illustrated instead its potential for spreading misinformation and highlighted the need to cite sources when writing articles.
The furor began with a story in the 31 December 2004 issue of the Washington Post. In that day's issue, reporter Jose Antonio Vargas wrote a piece entitled "Seeking the Hand Of God in the Waters". Near its conclusion, the article contained the following statement: "Following the devastation in Lisbon in 1755, priests roamed the streets, hanging those they believed had incurred God's wrath."
Vargas' article was part of the media coverage following the 2004 Indian Ocean earthquake, and drawing analogies to a very similar event that hit Lisbon in 1755 fit in quite naturally. It apparently struck a chord, and this particular statement was picked up and repeated in some form by many different sources in the following days.
However, the statement raised the ire of Theresa Carpinelli, host of a Catholic radio show in Canton, Ohio. Questioning this allegation directed against Catholic priests, she tried to track down a source for it and wrote about her efforts in a two-part article (Part One, Part Two) for the portal site Catholic Exchange. Carpinelli concluded that the allegation was unfounded and probably originated with Wikipedia.
The story was picked up last Thursday by John Hinderaker of the Power Line blog. In a post called "Not Even Voltaire Believed This One Archived 2005-08-29 at the Wayback Machine", Hinderaker commented on the disputed sentence, "It appears that it may have originated in an unsourced, wholly imaginary Wikipedia entry."
In trying to find the source of Vargas' statement, Carpinelli reported that she conducted a Google search for "1755 Lisbon priests roamed city hanging people". She found a number of sites that had relied on Vargas, along with Wikipedia (in all likelihood, she probably found a mirror, which is still the first hit besides Carpinelli's own article, rather than Wikipedia itself).
Carpinelli then went to Wikipedia to investigate it further as a possible source (she did not identify the actual article, but reconstructing her actions makes it clear that it was the 1755 Lisbon earthquake article). She speculated that Vargas came across the article by a Google search himself; as of now, the Wikipedia article is the second hit for "Lisbon earthquake".
In the version that Vargas might have looked at, the article contained a sentence reading, "In the following days, priests roamed the city hanging people suspected of heresy on sight, blaming them for the disaster." The language that "priests roamed", Carpinelli determined, had been in the article since its creation by Viajero in October 2003. Carpinelli said she "requested of Wikipedia that a source be cited for this allegation." This request was initially made 6 January, but placed directly into the article, rather than on the talk page, immediately following the disputed statement.
Upon returning 17 January to discover that no source had been provided, Carpinelli this time found her way to the talk page and raised her objection there. Muriel Gottrop responded, "Dear anon user, if you disagree with the article in some way, you are invited to change it."
However, instead of eliminating the statement from the article, Carpinelli acted on this invitation by expanding considerably on the objections she had earlier placed in the article. She said, "I am simply going to leave it for all to read, that the writer cited no source, because no source exists except within the writer's mind." Ultimately, the allegation was edited out by Sandover on 20 January, who also went back to the article last week and began adding outside references that could support its remaining content.
Both Hinderaker and Carpinelli commented on the failure of the Washington Post to run a correction regarding the statement in its article. This shows, one might say, one of the benefits of the wiki system by comparison; Wikipedia has at least managed to correct its information, albeit more slowly than it might have if Carpinelli had just removed the material from the article herself.
Incidentally, the 1755 Lisbon earthquake article was nominated as a featured article candidate on Sunday, although the nomination has already picked up objections from several people.
As part of the Wikimedia Foundation's continuing effort to become more self-sufficient and keep up with the demands of Wikipedia traffic, it is eliminating its financial dependence on Jimmy Wales' independent company, Bomis, Wales disclosed last week.
In a post to the wikitech mailing list last Wednesday, Wales indicated that he planned to change the contract with the colocation centre so that the Wikimedia Foundation would be paying for hosting directly, rather than paying Bomis. Under the new contract, Wikimedia will provide bandwidth for Bomis and charge the company according to the amount it uses.
Wales originally used Bomis to help fund the Nupedia project from which Wikipedia developed. Wikipedia was placed under the non-profit Wikimedia Foundation in 2003, but Bomis continued to pay for basic expenses. As traffic increased and more servers became necessary, Wikimedia fundraisers helped purchase the needed equipment while Bomis remained responsible for the cost of hosting and bandwidth.
Since September 2004, Wikimedia has been paying for half of the total bandwidth used by both organizations, somewhat less than a proportional share. The change reflects Wales' repeated statements that he intends to keep his commercial enterprises, which also include another company called Wikia, separate from Wikimedia. It also marks a major change in that the Foundation, while obviously still heavily dependent on donations and volunteer effort, is directly handling its own expenses.
In related activity, the Wikimedia servers were being changed to a new network switch, and the slower existing switch was being returned to Bomis, its owner. Wales indicated that this would allow Bomis' bandwidth charges to be allocated easily. The change caused some intermittent delays in using Wikipedia on Saturday.
Also, a set of recently ordered new servers (see archived story) arrived last week, which should help Wikipedia site performance. However, they are still being set up and configured for use, and have not been added to the active server lineup yet.
More hardware will likely be ordered in the coming months, as the Wikimedia Foundation approved a $130,000 budget last week for the first quarter of 2005. The budget allocated at least $75,000 to servers and other hardware, as both Wikipedia traffic and the associated costs are still growing at an exponential rate. As part of the budget plan, the Foundation plans to start a fundraising drive on Friday, 18 February, with a goal of raising $75,000 over three weeks.
At the instigation of Ausir, the first Wikiportal in the English Wikipedia has opened, and last week the new concept prompted the creation of a number of new portal pages.
The idea of Wikiportals, a page functioning as a portal for a particular subject area, was first introduced in the German Wikipedia in 2003. It was later incorporated in the French, Polish, Dutch, Japanese and Hebrew Wikipedias, but was not adopted in English until the concept was brought here from the Polish Wikipedia by Ausir.
Ausir opened the Biology portal on 5 February. This soon spurred the creation of portals for Art, Sports, and numerous other subjects over the past week.
While the larger French and German Wikipedias have more Wikiportals, the Polish ones are generally the most advanced in terms of visual design and layout. Since the Polish Wikipedia was also the most immediate source for the idea, the layout of Polish Portals has been copied for use in the English Portals.
Despite the jump in portal creations, their exact goal and their difference from WikiProjects remains unclear. In some other languages, the two exist side-by-side, often covering the same subject matter, and some of the new English Wikiportals have corresponding WikiProjects as well.
The Polish Wikipedia, however, doesn't seem to have WikiProjects (or at least the WikiProject page has no interlanguage link to Polish), which may be part of the reason why the portal idea was implemented there. But not all Wikipedias are clear about WikiProjects either.
The French page about WikiProjects indicates that they're supposed to be for coordinating issues across different Wikipedias. The German and English pages about WikiProjects say that they're for collaborating in a particular subject area within Wikipedia, with no allusion to other languages.
Portals are almost like mini-versions of the Main Page for a particular subject area, although their visual resemblance to the Main Page is stronger in other languages than English. WikiProjects seem to be more like working pages to decide on policies and conventions. However, a clear distinction between them isn't clearly mentioned in documentation on the matter in any language.
With respect to Wikiportals, there are also inconsistent practices across different languages, particularly involving their location. In the German and Polish Wikipedias, for example, the portals are part of the article namespace; French and now English, on the other hand, have them in the Wikipedia namespace.
On Saturday, Ausir posted a request to the wikipedia mailing list for a separate namespace to be created for Wikiportals. He said they don't really belong either in Wikipedia namespace or in the main article namespace. So far, there has been no response to the request.
For the past month, a small group of users has been closeted away, working hard at writing a faux article to sneak into Wikipedia. The joke was flushed out last Thursday when Ta bu shi da yu nominated it as a featured article candidate.
Those participating in the prank undoubtedly had to wipe the smirks from their faces when this activity was exposed. Upon its discovery, Bishonen, the initiator of the article, commented on being "caught with my pants down again". While the various authors had definitely been focused on satisfying the featured article criteria, they seemed to feel they still had their work cut out for them in that regard.
The article in question, originally titled Antique toilet paper holder, has been hidden privily away in Bishonen's user space since 9 January. The conspirators feared that if it was part of the article namespace, any old John could head straight for Votes for deletion and have it summarily ejected.
In terms of content, the article presents an incredible tissue of stories overflowing with connections to art, literature, religion, and politics. The participants seem to have scoured an impressive list of sources to compile this, all of them real publications, although Taxman expressed skepticism as to whether any of the content was based on these sources. But Giano contended, "A surprisingly large amount of the information in the article is confirmed in the references".
After Ta bu shi da yu's nomination, the article was awash in enthusiastic new contributors. It finally reached the point that the content was actually running over the limits of its original focus, including content about toilet paper holders into the 21st century. To solve this problem, it was renamed to "European toilet paper holder" and some now-extraneous content dealing with Asia was unloaded into a separate article, Asian toilet paper holder. While the project is commodious enough to entertain other sibling articles for the remaining continents, nobody has yet felt the need to relieve the project of its Eurocentric bias.
Bishonen had apparently been plotting to move the original composition into the article namespace on April Fool's Day, and was waiting to nominate it for featured article status until then. While this cunning plan went down the drain, the authors were still discussing how to scrape together enough support to have it featured on the Main Page on that date. However, Raul654, who serves as Featured Articles Director, informed Bishonen that he would not roll over and feature a hoax article, even on 1 April.
Others were more sympathetic, feeling that such creativity deserved better than to be banished to the outhouse of Bad jokes and other deleted nonsense. Mirv called it a brilliant article and said, "Hats off to all who worked on it, and I can't believe I didn't see it until it hit FAC." By the end of the week, the project even had a slogan: Wikipedia is not toilet paper!
Some efforts were taken more seriously, as eight articles were successful last week in their nominations to become featured articles. They included Windows XP, History of Russia, the Book of Kells, Hong Kong's Octopus card, Ugandan rebel group Lord's Resistance Army, the Beatles' song Get Back, Anne Frank, and a cricket article dealing with Test matches in the 19th century (to 1883). There were no new featured pictures last week.