[Foundation-l] Where do our readers come from? Q&A

Tue Jan 19 07:48:28 UTC 2010

Details on how to measure it are relatively complex. We can make a guess
because of data collected from sources available for Catalan. My mail was
just to explain the phenomena.

Figures results from: a) Surveys. Last one answered by 400 Catalan Wikipedia
readers. We use results from answer to question about other language
versions frequently used. [1]  b) Most viewed pages in Spanish, French and
English not yet existing in Catalan.[2] c) % of visitors to web pages
exclusively in Catalan using web browser configured in other languages [3].
D) Own experiments with common searches in Google configuring the browser in
Catalan, French, Spanish, and English, and some final cooking. Result is
very approximate but gives us an idea about what is happening.

The bilingual factor is not negative. It apparently reduces hits to Catalan
pages but really it increases hits to non Catalan pages.

The factor due to inexistent or not well developed articles has to be
improved by growing the project.

The more frustrating one is the Google Factor, You can Google “Integral”
even with a Catalan configured navigator and you will get  the English
version first, then the Spanish one (witch is a translation from an old
Catalan version) both in first page but not find the Catalan one witch is
the larger of all before page 10. This article is a very special case due to
specific factors.

A technical solution would be great. And perhaps it is not of high
difficulty. We could guess languages from IP address and highlight interwiki
links to those languages.

[1]
http://ca.wikipedia.org/wiki/Viquip%C3%A8dia:Segon_sondeig_dels_usuaris/4._Utilitzeu_amb_freq%C3%BC%C3%A8ncia_alguna_altra_edici%C3%B3_de_la_Viquip%C3%A8dia%3F

[2]
http://ca.wikipedia.org/wiki/Usuari:Meldor/Top_visites_2009#Mes_visitats_a_can_.28castell.C3.A0.29_que_no_tenen_link_al_catal.C3.A0

[3] http://www.eines.cat/?p=804

> From: Marcus Buck <me at marcusbuck.org>
>
> Joan Goma hett schreven:
> > There are 3 phenomena acting simultaneously against the number of visits
> to
> > small projects: The bilingual effect, the size effect, and the Google
> > effect. For Catalan case we estimate a penalization factor of 8.3 (that
> > means that visits are 8.3 times less that what they should be). It comes
> > from: 1.2 bilingual factor (visits lost because people also understand
> other
> > languages, even if they have the opportunity to read the article in their
> > mother tongue, they also read it in others). 2.5 size factor (visits to
> > other projects because readers don?t find what they were looking for in
> > their mother tongue). And 2,77 Google factor. (Visits lost because Google
> > directs people to other tongues projects). The only positive factor is
> the
> > bilingual one. We are working hard to correct the others. For other
> projects
> > those factors can be very different but the concept can be there.
> >
> Interesting. What's the math behind that numbers? Or the source?
>
> Marcus Buck
>
>
>
> Date: Mon, 18 Jan 2010 11:58:20 -0800
> From: William Pietri <william at scissor.com>
>
> On 01/18/2010 09:29 AM, Joan Goma wrote:
> > There are 3 phenomena acting simultaneously against the number of visits
> to
> > small projects: The bilingual effect, the size effect, and the Google
> > effect. For Catalan case we estimate a penalization factor of 8.3 (that
> > means that visits are 8.3 times less that what they should be).
> >
>
> In the long term, it seems like we could compensate for all of these
> effects in software.
>
> I'm imagining a user experience where we make it easy for multilingual
> users to switch back and forth. That would include passive detection of
> multilingual users, hinting when good content is available in other
> languages, and making it easy for multilingual users to help translate
> content. It might also be worth looking at URL schemes that are not 100%
> language-specific, to focus the Google effect more usefully.
>
> That would require a lot of technical work, and would raise a number of
> non-technical issues, but I don't see any insurmountable barriers to a
> more fluid experience for multilingual users.
>
> William
>
>
>