Back in 2013, when adding translations to the wiki was proposed for the first time, it looked like we did not feel really comfortable with the idea, given the initial proposal to add translations passed with only 60% approval. The main concern always has been the apprehension that translated contents will be outdated very quickly and significantly lagging behind diaspora*s development. We still tried it, and provided support for translations shortly after the proposal.
After more than a year in, it was proposed to disallow translating the installation guides, with the main reasoning being an overall bad adoption of wiki internationalization and issues with maintaining compatibility to the English versions.
Now, almost five years after the initial proposal, we’ve seen a lot of content changes and probably are able to judge the situation better than we have been back then. So, after some reoccurring issues in regards to the state of translations in the wiki, I am hereby proposing to stop translating the wiki and to completely drop all existing translations.
I’m basing this proposal based on the bad state of translated contents. Contrary to our translation teams for diaspora* and the project website, the wiki does not see much love. After five years, here is the state of our multi-language wiki as of right now.
In total, 23 pages are translated into one or more languages. Overall, attempts have been made to translate the pages into 12 locales (
ar, cs, de, el, eo, es, fr, it, ja, nl, pt, pt-BR, ru). Yes, we have two different versions of Portuguese, so actually, we support 13 languages.
There are 87 non-English page versions in the database right now. Distributed evenly over the 23 pages with translations, each page is translated into 3.78 languages on average. So by no means are translations complete, but rather, some pages have been translated into some languages, while other pages have been translated into other languages. The only page on which the language bar does not contain more red links than actual content is the main page.
The average lag between the English versions and the translations, that is, the time time difference between the latest English change and the translations, is 221 days. In a prefect world, the average would be negative, since translations should, in theory, be newer than their English originals. In fact, some articles are indeed more recent than their English counterparts. If we only take the outdated pages into account, the average delay is 399 days. The most outdated article lags a whopping 1520 days behind its English counterpart.
In realty, the lag would probably be higher, but the average got uplifted by a lot since, last summer, @waithamai went through all the translations and replaced any mention of Loomio by Discourse, as not a single translation had caught up by then.
To me, the entire situation feels like we’re trying to give the impression that our wiki is available in more than one language, when the opposite is true. Most pages are not translated, and even if they are, a lot of them contain heavily outdated or incomplete information.
Ultimately, I believe it would be better to go back to a English-only wiki. The wiki is mostly used by technical people and podmins, since a lot of workflow descriptions and architectural details are documented there. For end-user communication, we have our website (which looks far better in regards to translations) and the in-application help. Therefore, I think we would not cause any real negative impact by deciding the wiki is an English channel, just like this Discourse is.
Opinions on this matter would be highly appreciated.
In case you want to reproduce my numbers, the script used to generate the statistics in my proposal has been published on GitHub along with some basic instructions on how to use it. Please note that I emergency-changed some articles, so some numbers may be slightly different, but the overall picture looks the same.