Translation Teams: A Request for Help

[Grzegorz Adam Hankiewicz]

On 2005-05-17, "C. Michael Pilato" <cmpilato at red-bean.com> wrote:
> I am in dire need of some insight into how it is that you actually
> *do* get stuff done.

Divide work in unbelievably easy to do work units which nobody can
find excuses not to do. A paragraph a day keeps the coordinator away.

> What makes your Subversion book translation actually work?

To make a translation work, or just any kind of free software
project, you need a single person which is going to take all the
flak. I mean, this person knows that he is going to be alone for
all the responsability, work, and tough decissions.

If you find a person which under the worst conditions still agrees
to do the work, the translation will progress.

> How are the teams built?  How do they coordinate?

As said, you only need one person in charge, though it has to be
a tough one.  The rest, doesn't matter much. The coordinator will
dispatch work to other people. I divide work in files, each volunteer
goes for a single file, and does work as he wants. I dislike people
working on the same file because styles are mixed. It's better to
get the translation first and then argue about homogenisation.

If you manage to convince people to work a paragraph a day, you have
won a steady contributor. The reason for this is twofold. Translating
is one of the worst volunteer jobs you can do. Not paid, not sexy,
difficult, knowing two languages is just the basic you need, you
will end up having to dig lots of dictionaries of both languages
to figure out how to translate.

If you have collaborators, you will need to discuss translation
terms, where each subjective opinion is equally valid, and thus
somebody has to "dictate" what to choose. Finally, you have readers,
who will easily criticise your work not constructively, making a big
fuzz about how you are stupid and how they could easily outperform
you... in their dreams.

Second, most people who volunteer for translation have never done
this before, or it was paid.  This usually shows off after a few
weeks. If by this time they have not get used to the minimum work
contribution unit, it's too difficult to bring them back, and you
can only leave them alone and hope for them to return some day.

Translating a paragraph a day seems ludicrous. It is, however,
the best thing a translator can do. You only need 5 minutes a day
to do the work. Translating has the benefit that you don't need
hours of concentration to get into the correct mood for solving
that complex algorithm which nearly works yet fails with a very
strange input value and you don't know why.

Psychology plays a very important role here. If you get used to
small work units, you won't find excuses to avoid your work. If
you publish your small work units, people will see work is being
done. Slowly, but that only means they need more help, so more
people will volunteer (and hopefully some remain). The worst you
can do is have one month without repository activity.

Coordination is done through a mailing list by the coordinator
who started the translation.

> How should we deal with lone rangers who email saying they'd like
> to start a translation into some language?

You are looking for lone rangers who are steady workers and
experienced. 99% of the time you won't get any of this, mostly for
the reasons above. I would remove your rule of having two people
in charge, they both can be lazy and forget their commitment the
next day.

I would give the person in charge a three month trial period to
translate one chapter of the book. You give them full repository
access. If the time is over, you remove their access and ban them
from being a coordinator. Harsh, but otherwise everybody is going
to volunteer, and you don't want that.

Apart from the one chapter requisite, you also need them to translate
the web page. This includes providing a "hidden" version of the
page were they say in their language that nobody is working on the
translation and that people can volunteer emailing to blah blah
blah. You keep this version of the page hidden somewere in the
repository, so when the time expires you can replace the web with
the "sorry, nothing is being done" version.

You actually don't need a full web page, just the few paragraphs
of text you are going to put in the place of the current page.

If the volunteer manages to get the web page and the chapter,
they get to be coordinators. From then on, they are responsible
for all decisions regarding who to give access, how to coordinate,
whatever. And you blame them if something goes wrong.

> Are you tracking the latest changes to the English book, or are
> you working against some older tag of that book (you /are /doing
> one or the other, right?).

Translating and keeping updated is the worst you can do. It is
distracting.  You don't want to have each contributor translate
towards a dancing target. They will spend 5 minutes figuring what
to merge from the new version and... hey! My free time is over,
I'll leave it for tomorrow, have to go to work now, and drop the
kids at school, and...

> I think recent list traffic is a clear indication that there's
> *got* to be an easier way to get new translation teams on board,
> to keep translations from growing stale, etc.

You don't need a person in charge of managing translations and
volunteers. You need a set of instructions written in a file,
and a note on the books' web saying if you want to translate,
please go read the instructions. These instructions should be
as discouraging as possible, with those rules I mentioned above,
or others you may come with.

If somebody volunteers you ask:

- Have you read the translation instructions?
- No
- Then please read them to know the conditions.
or
- Yes
- Then here you have the write access and time up to X. Good luck.