Archives/2007-03-25

From LangCom

This page is no longer maintained and may be outdated: moved to m:Special projects subcommittees/Languages/Archives/2007-03-25.
Archived discussion
This is an email discussion about a private blog post by a member of the subcommittee. No objection was raised.



1.

Berto 'd Sera
25 March 2007 00:50


Hoi!


I’m thinking to post a blog on en.planet about these interface localization thing. I’ll post the text here before publishing, anyway. Although this will be just a private post of mine it’s important that we don’t start giving out a schizophrenic message. My impression is that apart from publishing policies we also need to explain things in human readable terms.


Let me know if anyone opposes, because in that case I won’t spend my time in writing on the subject.


Bèrto ‘d Sèra

2.

Jesse Martin (Pathoschild)
25 March 2007 00:57


Hello,

That's fine by me, particularly since it's not posted as a message by the subcommittee. All our discussions (except GerardM's messages) are publicly archived, so we have nothing to hide. :)

Yours cordially,
Jesse Martin

3.

Berto 'd Sera
25 March 2007 04:32


Hoi !


That’s the text I’ll publish on my blog, if no objection is made (the « other post » quoted is the one I mentioned yesterday)

You don’t let a baby out of the incubator until you are sure it can survive


In a simple, perfect world it wouldn’t take much to have a new wiki: say we would have a form and an automated script. Users X wishing to open a wiki written in the ZXX linguistic entity would simply choose the ZXX code form a list and click a Create wiki button. After all, setting up a wiki is but an administrative job, the sort of things that any host does by automated scripting directly from a cpanel. So why are wikies THIS darn primitive? Why do we need to make requests, have incubators, translate interfaces… is it just a way to keep people from having a wiki or what?


A first easy answer is that a wiki is not a private home page. Webhosting is easy because you pay your money and then the host basically wants to hear from you just one more time: on renewal. So they do whatever they can to get rid of everyone asap without needing to pay for human personnel. Sometimes it’s frustrating (we all had a problem that lacked qualified assistance in out life), but it’s quite logical.


WMF is not selling anything, though. We are providing a free service. This means that WMF first receives money from donors to produce a certain service (an encyclopedia) and then chooses on how these funds will be spent. So opening a wiki is an act that involves responsibility before donors, who (together with our volunteers) can be pretty much described as wikipedia’s “taxpayers”. A simple automated procedure cannot be used, because there is a need for a project to be evaluated before saying that it’s okay to invest resources on it.


Once we need to make a scrutiny, we are immediately confronted with the same old NPOV problem that marks the whole wikipedia life. We need a decision making procedure that will be fair to everyone and not just based on personal impressions/moods/etc. In the past we had a procedure made of a so called “public vote”. This worked quite well until only major linguistic entities were involved, but as soon as minor entities started to appear the number of pure political conflicts just skyrocketed.


Everyone who’s been reading meta remembers a good number of cases in which crazy mobs insulted each other for months. Socket puppetry and bogus voting were the most common events. It soon became self-evident that even the most NPOV wikipedian can go nuts when he deals with linguistic problems that are in some way connected with his/her own direct political environment. The system was obviously not working; we had to find a way that would keep politics off the process. LangCom was the answer.


Nowadays many people lament their requests being held for “mysterious reasons”. There is no mystery, though. LangCom publishes its discussions at http://langcom.wikimedia.org/wiki/Category:Archive . The idea behind this post is that (as always when a transition in methodology happens) there is no such thing as an excess of information. Maybe someone will disagree with what we say in our proceedings, but it’s in anyway better to have a clear discussion on a given point than having people dream of weird conspiracy theories.


So, the current situation is:

  1. All requests made on an older standard have been closed (which does NOT mean they were denied, we simply require another standard, i.e. they should be resubmitted asap.
  2. Once a given request has met the minimum levels you get an incubator in your language (voting/flaming is NOT a level, we simply do NOT care about how many votes you get, oppositions are simply ignored unless they have a firm technical base, no political argument can be considered, “dialect” argument included, you do need a minimum number of native speakers, though)
  3. In this incubator you form a community and translate your interface, this is what really gets evaluated, that is, what REAL results your project can give to wmf.


So, the main things you should know are:

  1. Flaming won’t help a new project (and it won’t even damage it, it will simply be ignored as “irrelevant”);
  2. Community building is a BASIC step of the incubator process: there is no time limit, nobody is asked to deliver within X days or die, on the opposite, people are requested to grow to a decent organization level before being granted an independent domain, if it takes long, then be it. The incubator is what the name says, if we meant it to be a limbo, we would call it limbo.wikimedia.org, but we call it incubator, and that’s what it is;
  3. A full translation of the user interface IS a strict prerequisite from many Committee members (including me), so you can expect NO as an answer any time a project requires a domain before having produced a localization file.


This latter point probably needs some explanation. Many people believe that “you can just make the translation later on”. This was definitely true as far as major languages were concerned, but how many major languages still miss a wiki? You are right, the answer is “none”.


Now put this element together with the fact that right in these day the “Byelorussian syndrome” finally reached a conclusion and the whole matter originated by allowing people to produce “just any interface”. You are probably not familiar with the “Byelorussian syndrome”, though, so I’ll take a minute to explain.


Our planet is filled up with governments and people have mixed opinions about them (I personally cannot stand any of them). As a result a lot of political statements are produced. In this case, there is a group (called NN by the journal they publish) who opposes the current Byelorussian President, Mr. Lukashenko. Actually it’s many such groups (as with any president on earth), but this one has a distinct peculiarity: they work on the “Byelorussian identity” by working on the structure of the Byelorussian language.


While their linguistic operation probably has a number of merits, as a matter of fact the resulting language is NOT what described by a BE ISO code. Moreover, the resulting language is not a living language, it is a construct made with a political aim in mind. Now suppose a number of those who believe that George Bush “stole” the last American elections made up an interface based on Mark Twain’s language like “a body’s gwyne-a-edit th’artkl-hear”. You may probably resort to old Lil’ Abner graphics and make it a very fortunate interface but the fact remains a fact: this is NOT an English language interface.


So how could we deliver a wiki made in a constructed language in the first place? Easy, we simply did not check the language being used. We allowed an “it will happen later anyway” policy and the result was an endless war, plus a wiki in an official state language (Byelorussian) that has lived thrice as much as my native PMS wiki and managed to make ~6.000 articles against our almost 5.000 in less than a year (and we are less than 2 million people with a 2% alphabetization rate). So there ARE reasons why we really want to check that the ISO code a project declares is going to be used as such.


Another reason is more practical, it’s called marketing. For most small minor linguistic entities a wiki is a debut in the IT arena. If you let a crowd of casual translators make up an interface “as time goes by” you’ll end up with the same term being translated in an endless variety of ways. As a matter of fact, building a technical vocabulary from scratch requires tact, co-ordination and some thought (see this other post for a number of reflections on how to build an interface). We all tend to undervalue this, because we all have a fallback language (English). But what if the community has no immediate fallback language?


You don’t need to reach the Amazonas for that, my own linguistic community is made up of two main geographical branches, one of which lives in Argentina in the so-called Pampa Gringa and the other in Piedmont, Western Alps (a lesser third lives in North America, but it’s almost extinct, from a linguistic POV). So what fallback language can we use? None, that’s it. And the further we get from western cultures, the more potential communities will have this problem. A fully translated user interface is not just a political correctness indicator. It’s a usability factor. How many articles would you see on en.wiki if its user interface looked like that of fa.wikipedia.org? None, or almost none…


The user interface is also the one and only thing that can get the press attention on a new wiki. No new wiki is going to come out from the incubator with 100 featured articles, this is just obvious. So what do you expect local media to talk about? Unless they can speak about the way in which you translated “mouse” and “web” they will only say “now we have Zxx.wiki, too. We all hope that someday it will be interesting to read it”. Not what you’d call enthusiasm, but on the other hand what can an honest journalist write?? Your wiki IS empty, isn’t it?


A nice interface and a nice main page have an extremely relevant impact on a project’s capability to attract traffic. It’s no secret that you need all the press you can get to make up a community, and you won’t get this press unless you give the journalists something they can write about. Our picking up common PMS street slang in our interface prompted the publication of the first Piemontese text in 150 years (at national level, we obviously always published for ourselves, but the involved numbers were MUCH smaller).


Just because of a few funny words we had hundred of thousands of people informed about our existence just 2 weeks after opening, when we had less than 100 articles. The tide hasn’t gone down, yet. Articles at national level keep appearing and they ALL talk about user interfaces first (plus invented traffic numbers, journalists cannot accept the fact that there are no real traffic data in wmf, so they make tem up themselves). These articles speak about all Italian minor linguistic entities as a collective category and bring traffic to every one of us, but they ALL center on user interfaces.


In the beginning some people in our wikies considered such behavior as offensive. Journalists sort of smiled about us. So what? Let them smile, as long as they talk about you it’s just added traffic you get and more helping hands for your community. Moreover, there’s nothing wrong in speaking about the language used. Actually it’s the opposite, since the one and only difference among wikies is (or should be) restricted to the language in which they are written. From any other point of view a wiki should simply be an encyclopedia.


I hope it’s now a bit clearer that we do not keep any project in hold. We simply care for projects to be capable to be successful wikies, and that’s it. You don’t let a baby out of the incubator until you are sure it can survive. Because if you do you are a killer.

Bèrto ‘d Sèra

4.

Jesse Martin (Pathoschild)
25 March 2007 18:27


Hello,

I would prefer a disclaimer somewhere noting that this isn't a message by the entire subcommittee, but otherwise it looks fine. Do you have any problem with archiving that, or would you prefer to publish it on your blog first?

Yours cordially, Jesse Martin (Pathoschild)

5.

Berto 'd Sera
26 March 2007 00:36


Hoi!

I added the disclaimer since the very start :) I am used to take my own individual responsibilities in the clearest possible way :)

It's published here: http://eng.i-iter.org/you-don-t-let-baby-out-incubator-until-you-are-sure-he-she-can-survive

It's repeated on en.planet and a number of other blogs. Not that it gets even 15% of the traffic more general topics get, which is in itself a indicator of "quiet state" for the issue. There is no general uprising in Wiki St and we can work for a good resolution that will suit all and everybody.

Yet while waiting for a Board's statement would be formally correct it's going to be risky if we keep people waiting until forever while the Board does not have time for us. Let's request a mailing list and start the jobs.

Bèrto ‘d Sèra

6.

GerardM
26 March 2007 06:42


<this user has not agreed to public archival.>

Navigation