voting on a string

[Kenneth Nielsen]

Hallo Jeroen

2009/1/28 Jeroen Vermeulen <jtv at canonical.com>:
> henning at ubuntu.com wrote:
>> Hi all,
>> currently there are no plans to implement something like this but it
>> might be worth explaining the idea in a blueprint so that it won't get lost.
>
> Frankly I'm a bit worried that this would be spreading the translation
> effort too thin.  The current process is: translator (any logged-in
> user) submits a suggestion, after which a member of the translation team
> checks whether it should go in

The problem is, then what? What if it doesn't go in? Then you are left
with nothing but bad options.

1) You can leave the string untranslated, in which case the time you
have spent proofreading that string will be lost. And furthermore the
translator will not get any better[1]
2) You simply correct the string, in which case you end up using way
to much time on it and the advantage of sharing the workload decrease
and the translator will not get any better[1]
3) You give the translator proofreading feedback and let him make a
better suggestion. This removes all the disadvantages in [1]. BUT if
the translation have 5+ contributers that means keeping track of that
many sets of feedback and sending that many sets of e-mails afterwards
of which you may never get a reply because that may be way more effort
than this translator ever wanted to contribute. This is clearly not a
workable solution.

This situation can be solved in two ways.
1) Provide a feedback feature in rosetta, a feature that I know is
planned for, but asaiu not in the near future. This will allow us to
give feedback, to learn the translator to give better suggestions, but
will remove the possibility of doing so in vain.
2) Allow a translator to sign off on a translation[2]. This will allow
for feedback and since this will require the translator to learn about
the workflow he is probably also more likely to be commited enough to
respond on the feedback. As a side note, when it comes to your
database size concerns, this could be implemented on a per user basis,
so that any given user could only sign off on translations in a finite
number of packages.

Regards Kenneth Nielsen

[1] Meaning that if the reason you reject it is a matter of
convention, then this translator will never learn that convention and
you can keep rejecting strings from this particular translator.
[2] Note that my proposal is different. I don't want a voting system,
but merely a way for a volunteer to say "I have now made suggestions
for all the remaining strings in this translation, either by making my
own or by signing off on somebody elses, and I am now ready for
proofreading feedback.

> .  If you expect more people to sign off,
> that opens the door to lots of people working on the same fraction of
> strings without achieving as much as they would as reviewers.
>
> That's assuming those voters are aware enough of applicable terminology
> and style standards etc. to be reviewers.  But if most aren't, then the
> voting is really only helpful for the cases that a good reviewer
> instantly recognizes as correct anyway.
>
> Finally, there's a technical issue with a cost trade-off: we have
> millions and millions of translated messages for every Ubuntu release,
> with large numbers of people involved.  If this feature sees serious
> use, we'd have to store millions and millions of votes: "person X
> already voted for translation Y."  That will slow down the site--maybe
> so little that you wouldn't notice, or maybe so much that it annoys lots
> of people.
>
>
> Jeroen
>
> --
> ubuntu-translators mailing list
> ubuntu-translators at lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-translators
>