Thanks for clarifying. But I have to disagree on the use of automatic translators on the readers’ side. First, it’s inefficient for every reader to translate some message. Second, LLM-based translators are far from perfect and only the original author or some other native/dual speaker, who wouldn’t need a translator anyway, has a chance of catching silly but semantically relevant errors.
Think about those slapstick scenes where someone inadvertently insults someone else’s wife or similar.
I’ve also had bad experiences in the past, particularly with Fr->En. I won’t put blind trust in some machine, since I will never call LLMs “Ai”, which is a huge misnomer pushed by marketing.
Don’t use LLM-based translators. I use Google Translate, which has been around for much longer than LLMs, and it works very well. I speak a smattering of French, and I haven’t yet found a problem with French-English translations.
First, I have Google aversion. Second, Google translate is the epitome of LLM, even though it’s been around longer than the notion of LLMs; they were kind of the pioneers.
And, as I said, I’ve had very mixed results with Fr->En in particular. Plus, I am in no position to judge the faithfulness of the translation, beyond dismissing obvious gibberish, so subtle but very relevant translation errors cannot be detected, especially if the original goes beyond every day use of the language, so it’s underrepresented in the training data corpus, for instance. Such translators are very handy for people doing translations, to do the grunt work, but a human, who is fluent in both languages, needs to check and correct the results. I’ve seen crazy funny mishaps with De->En and vice versa. I wouldn’t want our diplomats solely rely on LLM translators, that’s for sure.
Edit: I was also under the impression that the OP was capable of English, since the initial post was in that language, but then they he replied in French, which I thought might have been by accident. If it hadn’t been from him, I’d have probably just ignored the message.
Oh, that’s great news. But from my last experience with MS Office, which does support ODF, I am skeptical, because MS pulled an anti-DR-DOS-like move[1] in reporting a bogus error when opening ODF files. I can already see the headlines, chiding a “premature move to an untested format”, or some such.
It’s also very suspect, because the ministry in question is led by the CDU, which have a habit of embracing and extinguishing progress. They are the ones responsible for years of stagnation in the green energy sector, for they needed to make sure that their ilk weren’t cut out of profiteering. The “C” is for corruption, sadly.
From what I’ve read, this is lip service from Microsoft, and it’s only half-baked support, presumably reluctantly.
Germany has been moving towards open source for a long time. If I remember correctly, several government organisations have already converted to Linux.
Yes, of course, that’s my point. People running MS Office at home, for instance, who interact with some government office, will raise a stink when seeing such “errors”. And vice versa, because I see an awful lot MS Office usage, when I happen to be at one of those in person and catch a glimpse at their screens.
I believe, from my vague memory, that the cause of the “error” was MS insisting on some long deprecated header or it only supports v1.0 etc. It made a difference, if the original was saved in MS Office, as an ODF, or LibreOffice. So a perfectly valid ODF created in LibreOffice “raises alarm bells” when opened in MS Office; who do you think will take the flak for that? These tricks are ancient, yet they don’t seem to go out of style anytime soon. And once such matters are settled in the courts MS will have earned more interest from their ill-gotten gains than any punitive damages will set them back. Plus, businesses going under in the meantime will be gone forever, which is the real strategic win. There was a thriving service sector around the LiMux project, for instance; local businesses with OSS devs on their pay role, to tailor the “Standard Client”, which used Ubuntu as the base, but there was quite some substantial customization involved, including apps written by such businesses, to polish it all off into a neat package.
Well, the last big story was, in fact, LiMux. But I don’t really follow the news lately, for mental hygiene, so I may have missed more recent events. I think there might the odd (dim) lighthouse project, but those are only there to cover up the huge technical debt.
P.S.: It’s also kind of a running gag to come up with tacky “patriotic” project/initiative names, e.g. “E-Mail Made in Germany”, which was and still is their excuse of “E2E”, when in reality it’s just TLS between all providers that tack that label on their service; the obvious goal being to discourage actual E2E mails, “lawful interception” and whatnot. Then there was “De-Mail”, which with a straight face told us that the postal service would sign “legally binding letters” in our name, because that’s how that worked under the hood (all but defunct now); of course there are no rogue employees who would ever abuse that power. When you’re in the mood for a good laugh, search those terms on media.ccc.de. There were some hilarious talks; newer ones often have English dubbing and some may even be in English.
