I was a bit flummoxed at being the recipient of an email with the subject "Samoan Aurora 18 sumitted to Mozilla - again" which had a BZ reference (https://bugzilla.mozilla.org/show_bug.cgi?id=800550#7) After bouncing the query around the bit on Mozilla's #l10n I was directed to #pootle  on Freenode. The original sender of the email was mozilla.locamotion.org and, that is a service maintained by the folks at translate.org.za

This obviously led to another step of puzzlement - neither +Runa Bhattacharjee nor myself have ever done any Firefox l10n work off that Pootle instance. Having my name as a recipient was a bit off key. And, even though Dwayne Bailey was able to figure out how that may have happened, I remain a bit skeptical. The mozilla.locamotion.org instance is an "unofficial" one ie. it isn't the Mozilla blessed translation content management system. And yet the Mozilla L10n drivers allow commits from the system into the translation repositories. For me this raises the question of how is the sanctity of the translations maintained - do the Mozilla L10n system and the mozilla.locamotion.org system share any authentication mechanism thereby allowing only a certain set of translators/reviewers with commit rights ? Or, does the mozilla.locamotion.org instance inadvertently allow a parallel set of translators and translation communities ?

I don't recall reading anything from the Mozilla team about this arrangement and, during the recent meeting with Rakesh 'Arky' he was a bit ambivalent about how the workflow is. +Runa Bhattacharjee has written up notes from the meeting here http://arrbee.wordpress.com/2012/11/17/mozilla-l10n-discussion-notes/

Note, Mozilla already does things differently by looking at Products at the top-level. So, it is not unusual for the same locale to have to completely disjoint efforts doing Firefox and SUMO for example. Adding the additional layer of complexity of mozilla.locamotion.org (and, all the other tools) is a bit awkward.

As a tool Pootle is a great enabler of translation communities. As a tool that is outside the Mozilla infrastructure, the Firefox L10n Pootle instance is something that should probably receive much more clarity than is available now. And, I'd say that similar clarity should be put in place for the workflow. The underlying notion of "providing tools translators are comfortable with" is a fine one. Unfortunately, a plethora of tools increases the entropy and risk in the system. Firefox is a browser that language communities take great pride in making available in their own languages. Sadly, over the years Mozilla has generally managed to not put resources (internally or, crowdsourced) into fixing any issues. The usual "l20n will solve this" sentiment has been prevailing for far too long. And, with increasing number of language camps that often push in un-reviewed translations into the repositories (or, at least increase the chances of doing that), Mozilla L10n drivers should ask themselves about they can prevent the unraveling of a lot of good work of extraordinary people.

The problem as I see it is two-fold - that of governance and, an obvious lack of respect for the language communities. The diverse number of available methods to translate actually do harm and hinder communities than facilitate consistently high quality of translations. And, they also encourage the "seagull translation system" ie. appear-translate-disappear which leads to a lot of cruft that requires to be reviewed and scrubbed. When products like Firefox, FirefoxOS and Boot2Gecko are so high profile, not putting in effort to come up with a standard solution could be more costly in the longer run than it is beneficial in the shorter.
Shared publiclyView activity