I think that storing the localized content in separate flat HTML files makes it very difficult to maintain translations, as the article points out.
Ideally, you would have a master document in which some parts are shared across all translations (like the layout, the actual terms in the spec, browser compatibility information, etc), while other parts were localizable (like the field descriptions). Then when content was added/changed, it would be clear which parts of the translations needed updating, and you could either flag them as such for volunteers, or use machine translation as a stop-gap, or both.
With the architecture they described, however, your options are either to make the translations completely independent or to make translations completely machine translated, neither of which is a good solution.
It's quite telling already that they basically admit that mixing markup and content as tightly as that is problematic.
I mean I don't think it's actually that big of a deal as long as they stick to basic HTML markup (paragraphs, bold/italics, etc), but you're already running into problems when you add links.
Not having multi-language support as a first-class citizen is a step backwards IMO. Especially considering the world is still getting connected at an enormous rate.
Multi-language support has the inherent problem that is requires effort linearly in the number of languages. Wearing a hopeful hat this new architecture would only require an effort "linear" to the error rate of the machine translations.
Personally I had some very good experiences with some translators, and it could work even better if Mozilla manages to tune and train them properly. (Like making so that manual fixes improve the algorithm)
I don't doubt that the Github-based CMS will allow for community-provided translations. I _do_ doubt that we'll see anything like the current set of languages covered at anything like the current levels of breadth and quality.
Like I said in another comment, approximately nobody translates technical documentation for fun.
A surprising amount of mediawiki documentation is translated by volunteers (i cant speak to the quality as i dont speak nonenglish, realistically the english source material isnt that great) but volunteers do translate things
Notice that the English page seems obviously translated by a non-native speaker, which has some implications not favorable to Mozilla's behavior here, specifically that ESL people will contribute documentation in English that native English-language speakers won't even provide for themselves.
