Wikipedia talk:Wikipedia Signpost/2017-08-05/Recent research

From Wikipedia, the free encyclopedia
Jump to: navigation, search
  • Can we trust this academic paper? Did they compare the 9% increase in improved articles to a control bunch of non-edited articles? The authors do not discuss the overall growth rate of tourism in Spanish cities - is it more or less then 9% ? Would Kippelboy know more? Vojtěch Dostál (talk) 07:09, 5 August 2017 (UTC)
Without having read the full paper myself yet: The review already says that there was indeed a control group. That said, keep in mind that this is a preprint and not yet peer-reviewed, as also already pointed out in the review, which furthermore states that "it's not entirely clear to this reviewer how the statistical significance was ascertained".
BTW I took a slightly closer look at the statistical methods of a somewhat similar discussion paper involving two of the same authors (Slivko and Kummer) in this review two issues ago: "How does unemployment affect reading and editing Wikipedia ? The impact of the Great Recession", and also briefly compared them to that of another author in the subsequent issue.
Regards, Tbayer (WMF) (talk) 05:35, 6 August 2017 (UTC)
  • With regards to the paper on the spanish cities, the edits were reverted on Dutch Wikipedia because they were considered 'promotional spam'. Explanation is provided here. An example edit is this, which says (roughly translated back to English), among others: "In Mogán, it is roughly 360 days per year good weather. The village is also known as the village with permanently good weather. Mogán is situated right next to the sea, and because it is always good eather (sic!) in Mogán, it has been called out by the WHO, world health organization, as an ideal spot for rheumatic patients to recover from their disease" and "Mogán is very proud of her traditions, and this is visible in multiple ways. If you like historical traditions, then a visit to Mogán is recommendable. There are many traditional dance activities and traditional festivities.". These are subjective statements and indeed quite promotional. There were also strong suspicions that this was likely translated from copyrighted sources (although they were not identified). Now imagine users posting long pieces of text like this on multiple articles - that indeed triggers deletion and accusations of sockpuppetry. This is not in line with Dutch standards - I'm curious what the quality of the contributions was in other languages. effeietsanders 09:46, 5 August 2017 (UTC)
Very illuminating, thanks! Regards, Tbayer (WMF) (talk) 05:35, 6 August 2017 (UTC)
I was pointed elsewhere to this explanation by the research leader: in this talkpage archive. You can read it, it's in English. Basically they admit that the translator added copyrighted text to the Dutch articles (which implies that they didn't do this on purpose, and that this only happened in Dutch), translated from tourism brochures. This text shows that the research leader was aware of this at least after the fact, and I'm quite surprised this is not disclosed in the publication. effeietsanders 12:46, 6 August 2017 (UTC)
And on the German Wikipedia (edits included [1], [2], [3], [4]), there was some milder pushback regarding e.g. lack of references, travel guide language, and unencyclopedic illustrations, to which one of the researchers responded, stating later that there had been an attempt to address the issues. See de:Benutzer_Diskussion:Oltau/Archiv/2014#Spanischkurs and de:Benutzer_Diskussion:Mefk81#Spanischkurs. Regards, Tbayer (WMF) (talk) 22:06, 6 August 2017 (UTC)
Oh, once you hit a wall on nlwiki, it is typically much harder to recover. I won't say it's one of the more welcoming projects. But the way this was conveniently left out of the article is striking - especially as it could signal impacts on other languages where the same method was used. effeietsanders 14:07, 7 August 2017 (UTC)
Well, the authors didn't in any way imply that it was the Dutch Wikipedia's fault. One of my hypotheses was that their translator to Dutch had done a significantly worse job than the others, and I see this is indeed the case. Thanks! --Nemo 06:44, 15 August 2017 (UTC)