Drop-out compared

2011 April main page tutorial first second page last page
nl 7 steps 6616063 22809 3655 746
de 7 steps 36578909 10295 3217 1041
en 9 steps 146254073 26483 13219 2572
sv 9 steps 3627675 505 899 258

These numbers compare the tutorials in different Wikipedias. For example the tutorial in Dutch Wikipedia has nine steps; its first page is linked via the left side bar and this causes quite a lot of clicks. From the first to the second page the Dutch tutorial looses a lot of viewers, and in the end from 22,809 only 746 endure.

Other tutorials are not linked via the left side bar, and have considerably less views (in comparison to the main page). But it seems that they keep much more readers from the first to the last page. In Swedish (sv) Wikipedia the second page has much more views than the first one, possibly because other pages in that Wikipedia link to that second page (and others). The loss from first to second page seems to be the biggest in Dutch Wikipedia.

Lessons to learn:

  • Have on your main page and left side bar a link to your tutorial
  • Take good care of your first tutorial page, if it is confusing you immediately lose a lot of people
  • Whether you have seven or nine tutorial pages seems not to be too important

English language: Thanks, but no thanks

So this is Wikipedia NOT about.

‘English is like a wonderful woman’, a friend told me, ‘you can’t live with her and you can’t live without her’. By the way, he is a teacher of English in Germany.

In the Wikimedia world, English is the central and relay language. Important texts are originally written in English or reach most of the users via an English translation. This gives phrases in English an enormous influence, even where they should be rephrased or reconsidered.

For example, there is that phrase by Richard Stallman: ‘Think of free as in speech, not free as in beer.’By this, he wanted to characterize the essence of free software / free knowledge. I never really understood that phrase until Chuck Smith explained to me that this relates to a speciality of the English language. In English, ‘free’ is understood usually as ‘gratis’. In other languages such as German you can simply say ‘frei’ or ‘gratis, kostenlos’. Referring to freedom of speech and free beer in German is only confusing.

In the recent Editor survey of Wikimedia, the German translation mentions a “lokales Chapter”. This is of course the 1:1 “translation” of local chapter. But in German the English word chapter is hardly known, and the German word “lokal” relates not to countries but to cities.

German Wikipedians tend to talk about “Referenzen”, “reputable Quellen” or “Primärquellen und Sekundärquellen”. These are unreflected take-overs from English, without regard to the traditional German expressions: “Verweise”, “einschlägige Literatur”, “Quellen und Literatur”. Contributing to this is certainly the fact that many Wikipedians have a techie background.

Especially confusing are the take-overs to new people.  Wikipedia jargon in German is terrible enough, English words or pseudo translations make it worse. When I speak English I try to speak English without German influence (I know that I don’t always succeed), when I speak German I try to speak German without English influence.

Brockhaus back in Business?

Leipzig, gravestones of members of the Brockhaus family

The 30 volume Brockhaus Enzyklopädie, the largest print encyclopedia in German language, is history. In 2005/06 the last edition appeared, and in 2008 a short-lived online experiment failed. A year later, Bertelsmann bought Brockhaus, and even the German anti trust agency said: Yeah, it’s a monopoly, but the market for print encyclopedias has shrunk so much that the Bertelsmann monopoly would be the least problem to a newcomer.

History? Last edition? Bertelsmann now surprised the literary world with the announcement of a new edition, to be expected for 2014 or 2015. The 22nd edition will be available for tablet pc and other mobile instruments, and the buyer can order excerpts from it by print-on-demand.

We’ll see.

(According to buchreport.de, thanks to Mathias Schindler for the link.)

Wiki-Watch: watching the watchers

The last weeks, ‘Wiki-Watch’ caused some trouble in German language Wikipedia.  That ‘Arbeitsstelle’ (department?) of the university in Frankfurt (Oder) is determined to tell the world the truth about Wikipedia.

According to the website, Wikipedia has a ‘global monopoly over knowledge’, while ‘we users’ (readers) ‘know practically nothing’ about Wikipedia. A bunch of ‘admins’, an ‘elite of leaders’, is cutting itself off. So here comes the hero, Wiki-Watch, to make Wikipedia ‘more transparent’.

Wiki-Watch operates usually under that very name. The ‘Impressum’ explains that professors Wolfgang Stock and Johannes Weberling lead the ‘Arbeitsstelle’; probably the main work is done by their assistent Maximilian Kall. Stock belongs to a right wing Christian organisation (Christlicher Medienverband) and to the advisory board of ‘German Institute for Youth and Society‘ that helps people to overcome their homosexuality.

Hostility met by hostility

Among Wikipedians, the self-declared heroes met a certain degree of resistance. Of course, the name ‘Wiki-Watch’ sounds similar to ‘Human Rights Watch’ or ‘watch list’, and the whole attitude gives the impression that Wikipedia is something very negative. When Wiki-Watch wrote e-mails to German language admins for a survey, some reactions were quite hostile. The questions were actually harmless, but made it easy to reveal the identity of the single respondent.

News sites today came with results of the survey. 56 (out of 281) admins answered; the average admin is for about 40 years old and most commonly left liberal with a green touch. 38 % are unhappy with the rude manners in Wikipedia.

Can Wiki-Watch judge the reliability of articles?

Now Wiki-Watch, the website, entered a ‘beta status’ and presented its tool that evaluates the ‘reliability’ of Wikipedia articles. The whole thing is suspiciously similar to Wikibu.ch, a Swiss tool from 2009. You can enter an article’s name and Wiki-Watch tells you whether it has a feature status, how many ‘authors’ contributed etc.

The main difference to Wikibu.ch: The Swiss (among them Nando Stöcklin, a well-known Wikipedian) explain you that the tool can only give a hint about reliability. A road sign ‘deer crossing’, they say, does not mean that there actually will be a deer crossing, and the lack of such a sign does not mean that there certainly will be no deer crossing.

On the contrary, Wiki-Watch very bluntly judges: ‘reliable source!’ or ‘no reliable source!’ or something in between (five stars maximum). For example, [[:de:Gaius Cornelius Minicianus]] is ‘no reliable source’ because there were only few authors, few edits, few links to that article, and no footnotes. But: This very short article has no controversial content, and under ‘Literatur’ a renown work of reference is mentioned.

The article [[:de:Konrad Adenauer]] is very dangerous, because during last month there have been 4 reverts! But if you do check these reverts you see that they have been simply the deletion of vandalism. There is no reason to detest the reliability of the article due to this. By the way, the makers of Wiki-Watch did not notice that it is quite a difference whether an article has become a ‘Good article’ in 2006 or 2010.

Useful background information?

Wiki-Watch makes Wikipedia ‘more transparent’ also by providing background information. By now, this information is rather short, dull and strange. It lists up the Wikipedia article of internet jargon words, but forgets the much more useful [[:de:Hilfe:Glossar]].

Under ‘exclusive insight’, there are some interesting statistics, still, obviously the data comes from Wikimedia and other sources already known to Wiki researchers. Any reference about where the data come from – nope.

We Wikipedians love Wiki research, we do. But it should be executed with more care, and without such an attitude.

A text-book for Wikipedia – how came?

Some people say that Wikipedia or the internet in general make printed books dispensable. Maybe it is true, unless you write a text-book about Wikipedia writing. Recently appeared my ‘Wikipedia. Wie Sie zur freien Enzyklopädie beitragen’ at Open Source Press, Munich.

In 1887 Ludwik Zamenhof published the first work on his new language, Esperanto. It was a basic grammar and vocabulary in 40 pages. The first Esperanto speakers deemed this and short introductions in newspapers enough to learn the language because… Esperanto is so simple and easy to learn. It took ten years until someone came up with the first decent text-book about the language, which already had evolved within these years.

The same you have with Wikipedia. In 2003, a teacher wrote that his students wrote “great articles” without any preparations because Wikipedia editing is so easy and self explaining. Nowadays it takes a very optimistic (or naive) attitude to maintain such an opinion, look how complicated editing became since then.

Follow the flag

When I talk to Wikipedians a lot of them (a large minority) laughs about the idea of using a text-book to learn editing. If you are such an idiot that you need that, you should not be allowed to edit at all. But in 2009 Frank Schulenburg came to the Mentor pages of Wikipedia in German and asked about teaching aids. Then I said, in front of my fellow Mentors: If five men (or was it more?) will follow me, I’ll take up the flag. They followed, and the result was a text-book at Wikibooks. Some called it simply ‘the Wikipedia text-book’, but I insisted on a title of its own, ‘Eigene Beiträge’ because there can and should be many different text-books about Wikipedia editing.

In late 2009 I attended the OpenRheinRuhr conference in Bottrop/Germany, and met a representative of Open Source Press. We talked about the desirability of a Wikipedia text-book, and suddenly she told me to write a mail to the publisher. And so it came.

Historian: Wikipedia falls down on core task

Historian Peter Haber, PhD, is a Wikipedia-fan. To him, reading Wikipedia is even natural as earlier reading Brockhaus encyclopedia. Many colleagues are skeptical, but they never talk about it. So he and his students in Vienna examined some articles closely. His conclusion: Wikipedia is not so suitable for its core task – giving comprehensive guidance to get started about a subject.

What? Wikipedia is not the best first step to learn about a new subject, historian Peter Haber says.

Haber criticizes that the English language article about Engelbert Dollfuss mentions at large the shortness of this Austro-Fascist chancellor. This happens hardly in the article in German.

Another example for different national views, according to him, are illustrations chosen for the articles about the “Cold war”. The article in German has a sober map, the article in English shows Reagan and Gorbachev (the conflict subsiding), the Russian one a list of the block partners (with a long pro-Soviet list).

Factual errors are not the problem, according to Haber, there are some but you would find them also elsewhere. He finds it more important that Wikipedia has its flaws where everybody believes it is strong: Wikipedia is not so good in providing a first overview about a complex subject:

“It is a most sophisticated task to briefly introduce to a historical subject. Such contributions are not suited for collaborative creation.” (Zeit interview, July 8th, 2010)

Because, when several authors come and add something, the texts grow. You don’t need to be a trained historian to collect facts, but have to be skilled to present something complex.

The seminar in Vienna has checked for about 20 articles on history. The quality of the articles is divers, Haber says. But the Wikipedian who has a look at the student’s presentations has the same impression. One of Haber’s students is clearly a Che Guevara fan, I wonder what his presentation sounded like.

Articles based on works in only one language? Wikipedia finds it okay, Haber doesn't.

And the notes about national differences are based on single cases. Is it really more than incidental that the article in English mocks about the shortness of Dollfuss? Are all of the Russian Wikipedians really living in pro-Soviet nostalgia? Is it English Wikipedia that tries to suggest a harmonic end of the cold war, and not a single author who chose one of many possible illustrations? Wikipedians and non Wikipedians will have to wait until the end of the year for the complete survey, to see whether such observations have a substantial basis.

Wikipedia articles refer usually only to literature in the same language, Haber criticizes. According to him this is problematic from a scientific point of view. Actually most Wikipedia guidelines say that one should primarily refer to works in the reader’s own language. Here we come to the main questions: What is Wikipedia, what does it claims to be, and how readers use it?

Haber’s sound remarks repeat to a certain degree what Roy Rosenzweig wrote in 2006 about Wikipedia: not a lack of accuracy, but verbose speech is the problem. Like Rosenzweig Haber warns that historians should care more about Wikipedia:

“It is here where the popular knowledge on history is produced, not in expensive, voluminous essays.”

Germans don’t talk, quantitative analysis says

Felipe Ortega obtained his PhD with a thesis about “A quantitative analysis” of Wikipedia (Madrid, 2009). There are many graphs and numbers in it, and I do not pretend to understand all the math behind, but there was one assertion that struck me:

“Finally, if we turn now to talk pages, we find that […] the French version stores more than 150K more pages than the German one. In this case, French Wikipedians seem to support more active discussion about article content than German ones.” (p. 74)

According to him, Wikipedia in French had 630.000 articles and 367.000 talk pages, Wikipedia in German 700.000 articles and only 220.000 talk pages.

The more talk pages, the more “active discussion about article content”? Knowing my talkative fellows from Germany, Austria and Switzerland, I couldn’t believe Felipe Ortega’s finding and had a closer look.

A geographical article in French language Wikipedia such as “Chauvigny” (a small municipality in central France) has indeed a talk page (Discussion Chauvigny). But the only entry on that talk page is a template that the article belongs to a WikiProject about French municipalities. No discussion took place on this and many other talk pages.

In contrast, Wikipedia in German rarely uses this kind of templates (on talk pages). That’s why many talk pages have not been created. So the difference between de.WP and fr.WP has nothing to do with national differences in talk culture, just with different quality management systems.

Do you now understand why my trust in quantitive analysis has its limits?