post #2088998 post #3058893
Both of these are bad translations. I've written comments in which I've gone to the effort of explaining why, but to briefly summarize: both of them invent dialogue that isn't actually there.
Theoretically, what would stop a bad actor from combing the entirety of my translation history and then uploading translation edits that are only vaguely related to the actual text? (Picture, for example, post #2622609 with the text "I will now proceed to pleasure myself with these fish!")
As far as I can tell, as long as your cleaning/typesetting doesn't actively ruin the image, you might be able to get away with a lot. Putting aside, say, meme edits that are done for the sake of humor, how would we best handle these edits which present themselves as providing proper translations? Is there some process by which we might be able to flag them for deletion? Is a tag like bad_translation_edit (see my BUR on replacing hard_translated) in order? Either way, there should definitely be a strong requirement to prove that a post actually warrants deleting the post or applying a bad translation tag.

Edit: The latter example previously linked here (now deleted) was an edit of post #3058871 where all of the smaller bits of text were rendered as "yes" and "yeah", and the colored speech bubble was translated as something to the extent of "you fuckin raped my zangoose pussy"

I would be remiss to not bring up pool #20544 in regards to this, too (pool #18826 being the original Japanese version). The English version is an extremely sloppy translation, including gems like 'What that? It doesn't work such that magic word.'. Aside from the horrible grammar though, I can't say the translation is wrong, per se... I only translated the first couple of pages of the Japanese version here, and aside from the bad grammar, the general idea seems close enough. Though I don't know enough about what the later pages say to know if it gets worse. But like in your first example, the original creator knows of and "approves" the English version that was translated by someone else, but likely doesn't know English well enough to know how badly written it is (or maybe does know, and just appreciates that someone took the effort even if the result isn't great, who knows).

Well, Let me say beforehand ： I do Not really feel that I deserve the right to say what I am about to say.

Yes, those two ""translations"" are bad, Even for me.　Of course, the same goes for your example sentence ^{(( "I will ... these fish!" ))}, if it ever existed.
However, we aren't supposed to be able to stop people from posting itself - The same can be said for edits in general, not just translation.

What I am thinking is Bad translation-edits should be considered "bad-edits" as well and removed.
General "bad-edits" are removed by approvers already.　Would it be strange to wish that the same process could be applied to that?

In reality, it should require prove/proof like you said.　And, Only a few translators who can do it right.
But, once a bad user's recorded (( Or if more happened )), There should be a chance it will work, I think.

... With that in mind, I did a search [ delreason:*translation* ] and There really was a case [ posts/1221919 ] of deletion, apparently.

～～
That's why, I am Not a fan of "bad_translation_edit" tag, since I don't see the need to do it with Tag, at all.
To translate or To consider it ""bad"", One needs to study both English and Japanese, and it is harder more than mindlessly editing tags.
... Is it appropriate to treat such things with tags?　I'd rather throw it into invalid-tag if it ever really gets made.

In the first place, If users can find bad translation-edit, they can report it with proof, for removal ； If it's just Notes, then they of course can replace it without bothering to use tags, Can't they. :P
Imagine e621 with such a tag introduced - The crowd always only complains about anything would go around and just add it everywhere.　Without even providing a shred of evidence.

I wouldn't paint post #2088998 and post #3058893 with the same brush. I agree that the latter is objectively awful, with regards to how the editing process altered the image visually and to the actual "translation" as well. I admit that the former does not perfectly convey the original meaning either. However, translation is not an exact science, and we should take that into account when considering a proposal like yours. In my opinion, as long as the general "spirit" of the text is preserved, small changes can be forgiven.

I do believe that it would be a good idea to avoid translation edits as much as possible, since notes are far more adequate for the same purposes.

kurogi_foxsiv said:
In reality, it should require prove/proof like you said.　And, Only a few translators who can do it right.
But, once a bad user's recorded (( Or if more happened )), There should be a chance it will work, I think.

As long as you can break down the process to a sufficient degree, I don't think it should be too hard to make a case against an extremely bad translation. In the case of the Zangoose image, that looks something like this:

Raw text: 中に出すなって言っただろ!!!
中 - inside
に - particle for connecting a place to a verb
出すな - don't send it out/don't put it out/don't emit it! (negative imperative, base verb form is 出す, implicitly "don't cum".)
って - quoting particle
言った - said (past tense, base verb form is 言う)
だろ - sentence-ender expressing strong certainty (abbreviated from だろう)
Literal meaning: I said "don't cum inside"!!! (extremely certain)

This could get a bit more cumbersome for text that's longer than a few sentences, but as long as some significant part of it can be proven to be radically and obviously wrong, I don't think that's an issue?