Should Edited MTL Be Allowed?

[Agnaa]

I think we're still lacking a consensus.

 

[Antvasima]

Can somebody provide a summary for which staff members that think what here then?

 

[Sir_Ovens]

I think we could set a context based standard for this?

Obviously translating poetry or cultural based nuances would be hell for MTL, but I think straightforward snippets such as info blurbs or text meant to be taken as literal can be MTL'd and then double checked by a human translator.

The main boon of MTL is convenience. I believe if it saves our human translators some time and effort it would greatly encourage them to translate the more complex texts.

 

[Agnaa]

Can somebody provide a summary for which staff members that think what here then?

Not in the mood/with the spare time to do that myself.

I hereby give permission to the first non-staff member who steps up to respond with a summary of which staff members think what.

I think we could set a context based standard for this?

Obviously translating poetry or cultural based nuances would be hell for MTL, but I think straightforward snippets such as info blurbs or text meant to be taken as literal can be MTL'd and then double checked by a human translator.

The main boon of MTL is convenience. I believe if it saves our human translators some time and effort it would greatly encourage them to translate the more complex texts.

This doesn't seem to grapple with the issue at hand, as I outlined it in posts such as this one.

 

[DontTalkDT]

I would make a quick correction that Dylan Dog and other Bonelli comics have some official translations, though a lot not available on the web. There's also some fan translations, though once again not many. I've prioritized using either whenever possible (Except for the Batman crossover, since those came out after my work on the profiles, but if it's necessary I can go and swap them).

Either way I don't see an issue with this, cherry-picking something that only a certain few users would be familiar with can be done with a lot more than the language barrier, obviously we should have standards but at the same time when it comes to very obscure verses you've just gotta trust whoever's making the profiles at some point. Say there's a game with no comprehensive Youtube let's play, and someone makes a profile with screenshots they took on their own, there's an actual monetary barrier to verifying the scans on the profile. I'd consider that a bigger hurdle than language, you can just try to MTL and get the gist with something like Dylan Dog.

^I agree with that as well, to throw my vote in on another of the issues.
Again with exceptions for super high tier, high hax or controversal verses, because of greater abuse potential and evaluation being generally vastly more difficult.

 

[Shadowslash125]

Can somebody provide a summary for which staff members that think what here then?

I hereby give permission to the first non-staff member who steps up to respond with a summary of which staff members think what.

Guess it's time for me to step up onto this biz. I've been lurking here for some time, so it's only fair I contribute.
Seven staff members (including Agnaa, eight if we add Ant, but he hasn't said much to summarize) have made their points known, and I'll summarize to the best of my abilities.

Agnaa (including the OP)

• E-MTL is in the same rein as MTL (aka not allowed and unreliable)
  • Bad human translations fall into this pitfall as well
• If both MTL and human translations are used, source material with human translations will only be used.
• Verse with untranslated scans and stuffs will be completely affected by this thread regardless of the results
• Concerned about scans being taken out of context
• Disallowing translation sources once originally relevant parts are found to be unreliable results in official translations occasionally having small yet relevant battleboarding slip-ups
• If there's a bad fan translation and an official translation, the official translation will only be used. And if the series only has bad translations, then it could be nuked from the wiki
• Concerns on other situations based on translations. Call me lazy, but Agnaa has said it the best and I can't do any justice trying to summarize it

DontTalkDT

• MTL/E-MTL (?) shouldn't be banned, especially for simple verses, unless it becomes a major problem
• Partial MTL (MTL some first chapters, Human translate the rest as an example) shouldn't be allowed, since it doesn't accurately portray the verse and/or character
• Supports human translators completely replacing MTL, when that time comes

DarkDragonMedeus

• Human translations/translation helpers are integral for the more nuanced and complex translations that can bring big implications to verses
• MTL can work for simple stuff as long as its usage is clarified

Armorchompy

• Disapproves of out-of-context concerns, calling it unfounded paranoia
• It's hard to check up on really obscure verses, and it's better to trust that guy crunching hours into such verse at some point

Theglassman12

• Unless there are problems with translation, good translations shouldn't be deleted
• Whoever does the translation should be checked on a case-by-case basis, not the untranslated guidebooks
• Official Translations should be used, especially if the only other forms of translations (fans) are bad
• Any errors in used translations should be fact-checked by translation helpers
• Genuine and Approved translations can be used, despite not having official localizations
• A series has with no official localization utilising MTL for the majority of the scans should be removed.

Executor_N0

• Forbidding MTL isn't a good idea, but at the same time, it shouldn't only be used as a be-all-end-all for translating entire verses

Sir_Ovens

• Proposes a context-based standard for MTL
  • Mainly encourages MTL for simple info blurbs and snippets which can then be furthered checked by human translators
• The convenience brought by MTL can save human translators' some slack and encourage them to deal with more necessary and nuanced translations

EDIT: Antvasima

• Agrees with Agnaa

I'll ask for a staff for permission to speak my mind here as well at a later time. I apologize in advance if I got things wrong. Do note that I will need perms to reply to the corrections for my summary too.

 

[Armorchompy]

I had a bit of a discussion with Agnaa off-site and my thoughts are:

• MTL shouldn't be allowed unless vetted by an on-site translator- it's just unreliable as is, regardless of intent.
• Out-of-context concerns aren't entirely unfounded but they're going too far: Manipulating context is something that can be done in many ways (and checking on something can be very costly if no online footage exists, as well as outright very difficult if it's niche media- Trust me that it's very hard to get some old games to even work on a modern PC), we should not veto all hand-translated verses.
  • Part of my reasoning for this is that it's honestly not that common for people to outright lie about things like these and the punishment for getting caught is pretty harsh: I'd rather risk someone sneaking fake evidence through every once in a while than prevent lots more people from having fun indexing verses they enjoy (provided the profiles are well-made): facilitating the hobby of powerscaling is ultimately the purpose of the site.
  • That said if something is translated by the user it'd be good to have the translations get confirmed. I always try to get some Italians to approve my threads for Bonelli comics, since they're the only ones that can confirm that my translations aren't bogus.
• Suspicion can be cast on something that's obviously questionable: a guidebook or a novel that most can't even confirm exists, for example, is prime material for fabrication, especially if it just so happens to contain material for huge upgrades. But I wouldn't automatically put heavy burden of proof on anyone who does their own translations. (Keep in mind that faking a manga or a comic is harder than a novel given typesetting is a deceptively difficult skill, and faking something like a show or a movie is nigh-impossible)

 

[DontTalkDT]

DontTalkDT

• MTL/E-MTL (?) shouldn't be banned, especially for simple verses, unless it becomes a major problem
• Partial MTL (MTL some first chapters, Human translate the rest as an example) shouldn't be allowed, since it doesn't accurately portray the verse and/or character
• Supports human translators completely replacing MTL, when that time comes

Other way around on the red marked thing. Partial translations should be allowed. What we shouldn't do is to ignore 50 MTL chapters of a 1000 chapter (otherwise human translated) novel because they are MTLed, because that just absolutely shoots ourselves in the leg in our aim for accuracy.

 

[Agnaa]

I've just stumbled across a new weird case, that feels like it should be tackled with this.

I've found a niche indie RPG. It was originally written in Japanese, but the devs used Google Translate to provide an English release; they've done similar stuff with the lore they display on their website. So, officially-promoted MTL.

To add to these complications, they communicate with their English-speaking fans in the same way. And for those cases, we have zero chance of knowing what the original Japanese they used was before it was MTL'd into English. And these comments are occasionally of relevance to indexing.

I don't have any good solution for this. At this point my brain is broken and I don't know a good way forward.

 

[Armorchompy]

I've just stumbled across a new weird case, that feels like it should be tackled with this.

I've found a niche indie RPG. It was originally written in Japanese, but the devs used Google Translate to provide an English release; they've done similar stuff with the lore they display on their website. So, officially-promoted MTL.

To add to these complications, they communicate with their English-speaking fans in the same way. And for those cases, we have zero chance of knowing what the original Japanese they used was before it was MTL'd into English. And these comments are occasionally of relevance to indexing.

I don't have any good solution for this. At this point my brain is broken and I don't know a good way forward.

That's uh. My approach would be take the game MTL as semi-canon, try to use the JP translation as much as you can, and kinda ignore the WoG as much as you can. But honestly this isn't really relevant to the greater picture.

 

[Shadowslash125]

With perm biz out of the way...

I think Edited MTL should be allowed; at least for the small and self-evident scans. Obviously, scans that present controversial matters such as tier 2 stuff onwards and the likes deserve scrutiny from the staff. One example I can make out is these spec scans that Kamen Riders tend to have in their official pages. The more outdated riders need their specs scans to be revised so that their P&A abilities can be revised (I'm having one that is in limbo due to needing a long time for staff to go through).

And yes, while it's the best course of action to send those kinds of scans away to the translation thread regardless, all for the sake of accurate documentation, realistically speaking, I don't think the translation staff are able to translate them to the point that said scans become relatively available to be utilised in a sufficient amount of time. An example is when I've had one post that has been in limbo for a month now, and no translator has made an effort to touch a few statements that basically describe stuff like "this guy in spandex has a spring in his legs that allow him to jump higher than the average athlete". My time that could be spent on actually making/improving on the crt is wasted by the waiting time, and the translator's work is possibly never ever done if they have other similar posts such as this.

I propose Edited MTL because it can mutually benefit both the members trying to conduct a CRT and the translators that have mountains of scans to translate, since members don't have their time wasted waiting on and on while the translators can have more time to spare for their real life matters. Just because Slime Isekai made the mistake of abusing MTL and whatnot, it doesn't mean all non-English verses have to bear said verse's sins.

If Edited MTL cannot be accepted no matter what, procedures and stuff have to change drastically so that people can be more encouraged to sending scans into the translation service. With the services the translation team has been currently providing, niche verses don't get their scans translated at an appropriate time, and what scans are translated usually gets cherry-picked. If you had an appointment with a doctor, and the number on your ticket is shown, but the receptionist just hands you another ticket and asks you to wait for your new number to be shown, wouldn't you be miffed? This is the translation thread for niche verses, but the queue is weekly. Something has to be done, no?

 

[FinePoint]

I've found a niche indie RPG. It was originally written in Japanese, but the devs used Google Translate to provide an English release; they've done similar stuff with the lore they display on their website. So, officially-promoted MTL.

This is actually extremely common, and the source of most of the indie games on Steam with dialogue that seems broken or nonsensical.
Paying for translators is expensive, but machine translating a game to also sell it for an English audience is not (you especially see it in the adult category).

Ultimately, I don't think it's our job to question if what they publish and what they think actually align.
That is to say: if they say something only in English, and post it officially, and it doesn't blatantly contradict anything, then it should be usable as-is.

If they say something they don't mean it's their responsibility to correct it, not ours.

 

[DarkDragonMedeus]

This is actually extremely common, and the source of most of the indie games on Steam with dialogue that seems broken or nonsensical.
Paying for translators is expensive, but machine translating a game to also sell it for an English audience is not (you especially see it in the adult category).

Ultimately, I don't think it's our job to question if what they publish and what they think actually align.
That is to say: if they say something only in English, and post it officially, and it doesn't blatantly contradict anything, then it should be usable as-is.

If they say something they don't mean it's their responsibility to correct it, not ours.

While it doesn't hurt to try translating things, and be used for placeholders prior to human translators looking into it. It is ultimately best to get human translators to look at Raws and attempt translations.

 

[FinePoint]

While it doesn't hurt to try translating things, and be used for placeholders prior to human translators looking into it. It is ultimately best to get human translators to look at Raws and attempt translations.

I believe you may be slightly confused, friend.
I am talking about the instance Agnaa mentioned of a developer making WOG statements via machine translation and not providing the original Japanese.

In terms of the overall thread and MTL in general, I essentially agree with Armorchompy as well as your statement just now.

 

[Shadowslash125]

With that out of the way,

I still need these to be addressed in this thread.

I think Edited MTL should be allowed; at least for the small and self-evident scans. Obviously, scans that present controversial matters such as tier 2 stuff onwards and the likes deserve scrutiny from the staff. One example I can make out is these spec scans that Kamen Riders tend to have in their official pages. The more outdated riders need their specs scans to be revised so that their P&A abilities can be revised (I'm having one that is in limbo due to needing a long time for staff to go through).

And yes, while it's the best course of action to send those kinds of scans away to the translation thread regardless, all for the sake of accurate documentation, realistically speaking, I don't think the translation staff are able to translate them to the point that said scans become relatively available to be utilised in a sufficient amount of time. An example is when I've had one post that has been in limbo for a month now, and no translator has made an effort to touch a few statements that basically describe stuff like "this guy in spandex has a spring in his legs that allow him to jump higher than the average athlete". My time that could be spent on actually making/improving on the crt is wasted by the waiting time, and the translator's work is possibly never ever done if they have other similar posts such as this.

I propose Edited MTL because it can mutually benefit both the members trying to conduct a CRT and the translators that have mountains of scans to translate, since members don't have their time wasted waiting on and on while the translators can have more time to spare for their real life matters. Just because Slime Isekai made the mistake of abusing MTL and whatnot, it doesn't mean all non-English verses have to bear said verse's sins.

If Edited MTL cannot be accepted no matter what, procedures and stuff have to change drastically so that people can be more encouraged to sending scans into the translation service. With the services the translation team has been currently providing, niche verses don't get their scans translated at an appropriate time, and what scans are translated usually gets cherry-picked. If you had an appointment with a doctor, and the number on your ticket is shown, but the receptionist just hands you another ticket and asks you to wait for your new number to be shown, wouldn't you be miffed? This is the translation thread for niche verses, but the queue is weekly. Something has to be done, no?

Specifically:

If Edited MTL cannot be accepted no matter what, procedures and stuff have to change drastically so that people can be more encouraged to sending scans into the translation service. With the services the translation team has been currently providing, niche verses don't get their scans translated at an appropriate time, and what scans are translated usually gets cherry-picked. If you had an appointment with a doctor, and the number on your ticket is shown, but the receptionist just hands you another ticket and asks you to wait for your new number to be shown, wouldn't you be miffed? This is the translation thread for niche verses, but the queue is weekly. Something has to be done, no?

Currently, the number of translation requests completely outweigh the number of translator staff to go through, leading to large amounts of time spent waiting for said translations for CRTs. We've now lost one certain member of the translation staff, and I imagine the already turtle-like pace the translation thread has would be further slowed. If MTL truly is unacceptable with no exceptions, the services for translation have to be stepped up from whatever this is to give a better option. Right now, the MTL ban is just strong-arming people and inducing the fear of being banned to push the translation thread all in the name of accuracy (which is a fair reason). People flock to using MTL because it's more convenient that what we have here, and with the strict crackdown on it, most non-English verses would be affected, especially with the other rules that would further slow down time meant to be spent on CRTs. The bigger verses might not have too much of an issue with it, but the rather niche ones would just suffer.

It's like what Gabe said about piracy: "The easiest way to stop piracy is not by putting anti-piracy technology to work. It's by giving people the service that's better than what they're receiving from the pirates". Something should be done to make the translation thread better and more appealing (besides absolute accuracy), so that people would be more encouraged to utilise the thread ON THEIR OWN instead of being scared away to it.

 

[Agnaa]

I think we ultimately have our hands tied.

MTL still makes frequent errors.

We care about very very specific details for a lot of these things.

We have almost no-one willing to translate these materials.

I don't think we can make any meaningful incremental improvements there.

I don't see any options besides losing accuracy and inciting no-knowledge wars over minor wording differences they convoke from the aether (by allowing MTL), or creating delays so massive that many non-English verses practically won't be indexable (by not allowing MTL). ofc it's not as black and white as allow/disallow, we can ultimately set up a tradeoff between those two things to some extent, but I'd lean more towards not indexing over indexing incorrectly.

 

[Armorchompy]

May I ask that we could try to reach a consensus, at least regarding the situation that I was concerned with? I hope this doesn't come off as too selfish but I was planning on revising the Dylan Dog pages, so I'd like to know that they won't be deleted before I work on that.

... Anyways I hate to be a broken record, but can we reach some sort of agreement regarding human translations here.

 

[FinePoint]

I think a practical option is to allow MTL only if translation help from the Translation Helpers was requested first and no progress made after a considerable time, and that of course any issues with the translation which affects the verse can always be brought up in a CRT if someone notices it.

That way it becomes a problem which can be solved by having active and helpful staff, but if not doesn't unfairly bar countless verses from ever being represented. It would encourage us to fix what is the real problem in the room, whereas otherwise we end up punishing random users for our own shortcomings.

 

[Antvasima]

I think we ultimately have our hands tied.

MTL still makes frequent errors.

We care about very very specific details for a lot of these things.

We have almost no-one willing to translate these materials.

I don't think we can make any meaningful incremental improvements there.

I don't see any options besides losing accuracy and inciting no-knowledge wars over minor wording differences they convoke from the aether (by allowing MTL), or creating delays so massive that many non-English verses practically won't be indexable (by not allowing MTL). ofc it's not as black and white as allow/disallow, we can ultimately set up a tradeoff between those two things to some extent, but I'd lean more towards not indexing over indexing incorrectly.

I am personally leaning towards this regrettable solution as well.

@AKM sama @DontTalkDT @Mr. Bambu @DarkDragonMedeus @Just_a_Random_Butler @Celestial_Pegasus @Ultima_Reality @Elizhaa @Qawsedf234 @Damage3245 @Starter_Pack @Abstractions @Colonel_Krukov @SamanPatou @GyroNutz @Firestorm808 @Everything12 @Maverick_Zero_X @Crabwhale @GrathOfLux @Dereck03 @Planck69 @LephyrTheRevanchist

What do you all think? This is important.

 

[Mr. Bambu]

I would prefer that path as well, if those are truly the only options. Knowingly making bad pages is worse than not having the pages.

 

[DontTalkDT]

I am personally leaning towards this regrettable solution as well.

@AKM sama @DontTalkDT @Mr. Bambu @DarkDragonMedeus @Just_a_Random_Butler @Celestial_Pegasus @Ultima_Reality @Elizhaa @Qawsedf234 @Damage3245 @Starter_Pack @Abstractions @Colonel_Krukov @SamanPatou @GyroNutz @Firestorm808 @Everything12 @Maverick_Zero_X @Crabwhale @GrathOfLux @Dereck03 @Planck69 @LephyrTheRevanchist

What do you all think? This is important.

I'm still strongly of my initial opinion.

I would not want to ban them, unless actual trouble regarding stuff starts popping up. Like, if your fiction debates weird metaphysical stuff where exact wording is majorly important it makes sense to say that doesn't work. For a simple verse where it never becomes a problem, I think it's fine. Should be replaced, of course, if it ever gets a better translation. But if it's all we ever get in the west and it's somewhat workable, then it may as well be the 'official' version for us. It's what people will talk about.

But I'm especially not in favor of the partial translation thing. If you have a novel where the first 300 chapters are machine translated and the last 1000 are human translated (yes, cases like that exist) it would be majorly weird to have to cut out information from those first 300 chapters. It would simply result in a profile that doesn't accurately reflect the character, completely shooting us in the leg in terms of the goals we want to achieve.
If a reasonably paced effort is ongoing that to replaced MTL with human translation one could say to wait until that catches up, but otherwise not considering the whole thing is not favourable for us.

When we do translations, I'm fine with us not doing machine translations, but if there somewhere is a community with a following doing edited machine translations of stuff, I think that's ok until actual trouble happens.

Quite frankly, westernized translations also have a tendency to mess up the weird metaphysical details we sometimes care about, but we simply fix them when we notice. Wikis are in a constant state of flux, being updated with new information. It would be sad to ban the debate of works of fictions just because we're for some reason incredibly scared that we could potentially be wrong about something due to this one of many potential sources of error. We are wrong about stuff 500 times a week, the mere potential of being wrong in some rare undetectable cases shouldn't be an issue for us.
Heck, we allow plenty of verses where we don't even have access to all material. That's really not better.

And let's not even talk about the trouble of figuring out whether fan translation #376 uses MTL or not. That's drama and speculation in its own right.



In short, I'm in favor of an incident based approach. If and when a verse becomes a problem due to bad translation, we can decide to delete it (this goes for machine and human translation equally). But if no issue ever happens, there is no need to act.

 

[Agnaa]

When we do translations, I'm fine with us not doing machine translations, but if there somewhere is a community with a following doing edited machine translations of stuff, I think that's ok until actual trouble happens.

Quite frankly, westernized translations also have a tendency to mess up the weird metaphysical details we sometimes care about, but we simply fix them when we notice.

In short, I'm in favor of an incident based approach. If and when a verse becomes a problem due to bad translation, we can decide to delete it (this goes for machine and human translation equally). But if no issue ever happens, there is no need to act.

I might be able to jive with thus, but what sorts of things would be considered an incident?

I'm currently in a thread where a series is being argued to have the highest non-+ High 1-A on the site, as well as a Tier 0. Some questions about the exact wording are coming up, but the series is entirely MTL, and we don't have people knowledgeable in the language to help out, so we're each MTL'ing explanations we look up to support our arguments. Is that an incident?

Heck, would any substantive question about translation particulars be considered an incident?

To go back to earlier examples from this thread, would grammar issues (i.e. a character's pronouns changing mid-sentence for no reason due to the MTL getting confused) be considered an incident?

More on the fan TL end, would a translation saying they cut out some boring pointless conversation be considered an incident, disqualifying use of that translation?

 

[DontTalkDT]

I might be able to jive with thus, but what sorts of things would be considered an incident?

I'm currently in a thread where a series is being argued to have the highest non-+ High 1-A on the site, as well as a Tier 0. Some questions about the exact wording are coming up, but the series is entirely MTL, and we don't have people knowledgeable in the language to help out, so we're each MTL'ing explanations we look up to support our arguments. Is that an incident?

Sure, in my book, if you have one of those metaphysical specific verses and the wording is in question, nuke it.

Heck, would any substantive question about translation particulars be considered an incident?

Within reason.
If something is only about a single character, maybe delete just that character.
If something isn't important to the pages (e.g. purely supportive evidence, which is not essential for approval), it's probably not an issue.

And obviously the suspicion that the translation is wrong must be somewhat reasonable. Like, if the translation says "120m fireball" and all evidence points to that being the translation, just going for no reason "but I want a human translation of that" is not a reasonable suspicion.
The same goes for human translation btw.
Pokémon has several instances of the English translation not saying what the Japanese original says. But if I went to a random Pokémon page, picked out a random feat and said, "ok, I need a human translator to confirm that this piece of translation is done in the most literal way", it would not be a reasonable suspicion against the feat either.

To go back to earlier examples from this thread, would grammar issues (i.e. a character's pronouns changing mid-sentence for no reason due to the MTL getting confused) be considered an incident?

Only if it puts doubt on who the sentence is talking about.
If it's clear to everyone what is meant, then it's not an issue.
Heck, there are some human translations out there from people who just are not that good at English.
Double heck, there are some English novels out there written by people who need a better grammar checker.

More on the fan TL end, would a translation saying they cut out some boring pointless conversation be considered an incident, disqualifying use of that translation?

Yes, yes, that would be an incident. If a translator blatantly isn't trying, it's obviously a problem.

 

[Agnaa]

Yeah I think that works tbh.

I do feel like we should have similar standards for human translations, but probably be a fair bit more lenient, although I'm not sure exactly how to do that. Maybe require more than a suspicion/question, a demonstrated failure in translation, before we disallow a human TL?

 

[DontTalkDT]

Yeah I think that works tbh.

I do feel like we should have similar standards for human translations, but probably be a fair bit more lenient, although I'm not sure exactly how to do that. Maybe require more than a suspicion/question, a demonstrated failure in translation, before we disallow a human TL?

That sounds like a decent approach.

 

[Agnaa]

As a bit of a test case for that sort of thing, what would we end up doing with Medaka Box?

The original fan manga translation had a lot of issues, including ones that have caused frequent debates over tiering like this, although we ultimately didn't accept those.

There's been a more recent fan translation, but it only got through ~100 of the 192 chapters.

Other additional content largely has good translations.

So what action would be needed on that front? Would all statements from where we only have the old TL need to be double-checked by translators? If we can't get those, what would be done to the profiles?

 

[DontTalkDT]

As a bit of a test case for that sort of thing, what would we end up doing with Medaka Box?

The original fan manga translation had a lot of issues, including ones that have caused frequent debates over tiering like this, although we ultimately didn't accept those.

There's been a more recent fan translation, but it only got through ~100 of the 192 chapters.

Other additional content largely has good translations.

So what action would be needed on that front? Would all statements from where we only have the old TL need to be double-checked by translators? If we can't get those, what would be done to the profiles?

My approach be to double check the old translation with translators only where it seems like it might matter, not every single thing in the manga.
And if we can't do that, take the text with an extra grain of salt, I suppose. I don't recall what happened in threads, but I think a "surpasses dimensions" statement completely out of context should honestly be too vague for 1-A, regardless, if that's what you mean with that scan.

 

[Agnaa]

Ultimately there are a lot of things that reasonably might matter. Really, all ability explanations that aren't blindingly clear by the visuals themselves.

Given how, for that series specifically, we've struggled to get translators to provide checks for the recently-found guidebook stuff, I have doubts that they'd be willing to explain the proper meanings of all the pages like this.

And, importantly, what happens if we can't get get translator vetting for that?

 

[DontTalkDT]

Ultimately there are a lot of things that reasonably might matter. Really, all ability explanations that aren't blindingly clear by the visuals themselves.

Given how, for that series specifically, we've struggled to get translators to provide checks for the recently-found guidebook stuff, I have doubts that they'd be willing to explain the proper meanings of all the pages like this.

And, importantly, what happens if we can't get get translator vetting for that?

I feel like you are going pretty far into the territory of assuming that just everything written is completely wrong for no specific reason, when you say that all abilities that aren't automatically clear by visuals are problematic. Unless I missed something about Medaka Box' translation when reading it, it's generally coherent (is it even MTL?).
The chance that bigger explanations that have plot ramifications are completely wrong is pretty low.

I'm not sure what exactly in the page you linked is the problem. Like, it's explained to be an ability that lets her turn into another person. IIRC we see her turn into another person and use their abilities. So what do you even need a translator to vet here?

 

[Agnaa]

I feel like you are going pretty far into the territory of assuming that just everything written is completely wrong for no specific reason, when you say that all abilities that aren't automatically clear by visuals are problematic. Unless I missed something about Medaka Box' translation when reading it, it's generally coherent (is it even MTL?).

There's a loooot of weird bits of wording that come from translation mistakes, just compare the original one with the newer translation of the coloured version to see what I mean.

It's not MTL, it was just created by novices.

The chance that bigger explanations that have plot ramifications are completely wrong is pretty low.

Plot ramifications, no, but profile ramifications? Absolutely.

We've banned people over inserting the word "void" into a scan when translating it, resulting in the series getting NEP. But that ultimately doesn't have "big plot ramifications" or anything.

And hey, if those major things are what you're worried about, then even MTL from a decade ago would suffice.

I'm not sure what exactly in the page you linked is the problem. Like, it's explained to be an ability that lets her turn into another person. IIRC we see her turn into another person and use their abilities. So what do you even need a translator to vet here?

Off the top of my head:

• What invocation of "concept" are they using; is it one that could be indexed as Conceptual Manipulation?
• How exactly does the first speech bubble relate to her power? Is she able to replace any word with similar concepts, or just herself? Since the former would allow her to transform enemies into weaker characters.
• What is the exact nature of the comparison to Yuzuhashi's ability? If he got a profile, would he be able to transform into Medaka and replicate her abilities, like Yuzuriha can, or would he only be able to take on her appearance?

Even if you don't buy this one specifically, I would be astonished if you did not agree that there are at least 5 scans in those 80 chapters whose specific wording is likely to impact the profiles. And as I said, we've struggled to even get one translation for other content from this series.

 

[DontTalkDT]

There's a loooot of weird bits of wording that come from translation mistakes, just compare the original one with the newer translation of the coloured version to see what I mean.

It's not MTL, it was just created by novices.

Like, anything specific? 'cause I wouldn't consider bad grammar a problem for this.

Plot ramifications, no, but profile ramifications? Absolutely.

We've banned people over inserting the word "void" into a scan when translating it, resulting in the series getting NEP. But that ultimately doesn't have "big plot ramifications" or anything.

And hey, if those major things are what you're worried about, then even MTL from a decade ago would suffice.

Yeah, faking translations is obviously bannable?
But the chance that "void" would just by chance show up due to a translation error is pretty low. Like, translation issues are of a very specific kind. It's not that sentences are created at random and the sentences that are present have to fit to their surroundings.
My point about plot ramifications was pretty much the opposite btw: If the logic of how a power works matters to the plot, then it's hard to get it wrong in translation without creating a plot hole that would show that. Weird side things that are not as important to the plot are more prone to perhaps slipping through.
But again, part of a reasonable suspicion is that there is something that points to it being wrong. You need to give translations a certain level of trust from a practical standpoint. Believe revision can happen when new facts appear.

Off the top of my head:

• What invocation of "concept" are they using; is it one that could be indexed as Conceptual Manipulation?

What you are doing here is something I personally consider an unfortunate practice to have started to pop up more frequently: Taking the assumption that Japanese is somehow more clear in its wording than English.
The chances that a word is used that means "universal" and then got translated to "concept" are pretty low, unless there is some known Japanese trend I'm unaware of. Grasping at a translators opinion on what a likely just as vague word in Japanese means is not a good basis of discussion, even when we have a good translator. Because the word just likely isn't unambiguous. (caugh looks like 概念 caugh)
Language speculation aside: Work with what the translation provides until you see anything else. Seriously, no official translation would do that part any better. You are starting to have an issue with foreign media in general here, not with bad translations. You won't ever have that level of rigorous disambiguation of all words for any translations.

• How exactly does the first speech bubble relate to her power? Is she able to replace any word with similar concepts, or just herself? Since the former would allow her to transform enemies into weaker characters.

Ambiguity of writing isn't an issue of translation. That speech bubble in isolation simply doesn't explain that, because it explains the nature of the language device, not directly her power. In the context of speech bubble two and what we see, it seems herself centered. Nothing more is really suggested.

• What is the exact nature of the comparison to Yuzuhashi's ability? If he got a profile, would he be able to transform into Medaka and replicate her abilities, like Yuzuriha can, or would he only be able to take on her appearance?

While I have an opinion on that, ultimately, like for the last point, this just doesn't seem like a translation issue to me. If the original doesn't explain something clearly, the translation won't either.
Blindly hoping for hidden additional information isn't a reasonable suspicion. That's the level of doubt where you just get a problem with foreign media in general, not with bad translations.

Even if you don't buy this one specifically, I would be astonished if you did not agree that there are at least 5 scans in those 80 chapters whose specific wording is likely to impact the profiles. And as I said, we've struggled to even get one translation for other content from this series.

You could read the official light novel translation of A Certain Magical Index and would find places where the wording choice of that changes nuances as well.
When working with translations you can't be so critical that you hold a generalized suspicion against every line of text. That's not practical and also not an appropriate way to interact with the sole discussed version of a work.

So the question is, are there reasonable doubts about specific parts of the translation? Not just arbitrary doubts? Because I'm not in favor of deleting profiles because there is some chance one of countless vague wordings might be somewhat more explicit in Japanese. That's just unreasonable. At some point, you just have to work with what you have access to, otherwise we can start by deleting all of Pokémon for the possibility of weird untranslated Pokémon thing #157 having an impact on Pokémon profile #298.

 

[Agnaa]

I can't square this angle you're taking now with your earlier comments.

You've gone from saying "Yeah, if there's some questions about specific metaphysical wording, nuke it" to "It needs to lead to a plot hole, we can't assume that the original language is any more clear."

Whether a certain cosmological construct contains "other spatial dimensions" or "other universes" will almost never lead to plot holes, but such a thing is very important for our tiering.

 

[Executor_N0]

Sometimes it gives the feeling that people expect the translator to have the job of interpreting the work on how it fits with our system, when that is far from the case. We shouldn't expect any given work of fiction to bend itself to fit with our system; it's our job to interpret the work and see how it fits the most with our system. A translation needs to be consistent with the work itself, not with our system. If a valid translation for a word is "void, concept, universal, transcendental", but it doesn't fit with what our pages require for what we consider these works, this isn't a problem of the translation, this is just a matter of our users having to interpret the work and trying to figure out what the work has to say, instead of working backwards from our system and trying to find what is the highest possible usage of any given word.

If people were to just give more thought to what things could mean, a lot of this would be solved. I can't even count the amount of times when someone asked me "What is this word? Is it concept? idea? is it conceptual manipulation type 1?" or going trying to look for some bizarre ideal of "an unbiased native speaker that somehow can not only translate the text to English with perfect accuracy, but do so while fitting with what our system requires", when that on its own is almost impossible, unless the work itself gives content for that (Which for my experience when people ask me that, it's almost never the case).

MTL has a lot of problems, but so does almost every translation if you put them under the requirement of having to fit with our system, when not even English-original works do so (I don't think anyone would blame every time someone uses the word concept and isn't for what our page counts as conceptual manipulation).

Using the whole "transcendence" thing as an example, since I have some experience with it, in Japanese, a word for Transcendence (Surpassing, being above something) is 超える (Koeru), while the word for crossing over, going to the other side, is 越える (Koeru). As you can notice, both not only have the same reading, but even pitch accent. It's very common for even Japanese speakers to confuse the two Kanji, and so usually I see a lot more of 超える than 越える in texts that are specifically about walking, crossing over, traveling, especially in the context of time or dimensional travel.

A lot of lines about "transcending time" or "transcending dimensions" refer to just time traveling or dimensional traveling, but 超える is used nevertheless (And personally, it's not a big deal. With some abstract things like time or dimensions, it could be very easy to understand why engaging with them on any level would deserve the usage of "transcendence"). Would it be a bad translation to translate such lines as "We transcendenced time" instead of "We traveled across time"? Personally, I don't think so.

There are definitely cases in which translations are manipulated with specific wording to get some ability or tier, but the seed for that isn't the translation itself, but the overreliance people have with the idea of "Our system has a specific wording and other works need to follow that wording", when I would say the approach would be people understanding that how you call something matters less than how the thing functions, in which case you stop being dependent on a single word, and becomes dependent on a bunch that have a lot more context that you can work with (It happens a lot that in many franchises, there's no context, but again, that isn't a problem with just non-English works, we just should judge them the same as English works instead of having some bizarre ideal that a native speaker has the ability of figuring out content that isn't in the work that happens to explain how a work fits with our system).

It's just my opinion, but I think that the problem goes much deeper than "Manipulated MTL/Translations in general", in most cases that feels just a side-effect of unrealistic expectations in hopes of getting a higher tier or powerful ability.

 

[Agnaa]

I think there are cases where a translation is demonstrably wrong, in a way that doesn't matter for casual readers, but does for us trying to index it. Take this calculation, which was based on this translation:

Shichika dashed out of the hut, or what was left of it, jumping straight through the blown-out wall rather than the door.

"Shichika!"

By the time Nanami called his name, he was already out of sight, out of the hut, and up the mountainside.

I calculated this as him moving up our minimum height for a mountain (609.6 meters) in 0.66 seconds, for a speed of 923 m/s. This was accepted by a CGM.

When I looked into the actual raw text, it said he entered the forest on the mountain (山林に這入った), which makes the distance far lower, meaning it could be as slow as 15 m/s or so.

This doesn't really matter much in the grand scheme of the series. He travels with someone far weaker and slower, so they'd take a while between cities regardless. If he moves more quickly during battles, so do his opponents, and that doesn't change the plot. Plus, he eventually grows stronger and gets other similarly high speed feats.

It's experiences like these which make me want to dig down into the details of how feats are originally phrased.

But more specifically on the idea of not believing that the original language is more clear, my only real concern with a shift in this direction is that, if we don't root ourselves to a knowledgeable speaker saying "This usage of the word is most likely, so let's just go with that", evaluations become a lot more complicated, more subjective, with possible results that are a lot wider. It generally also seems really hard to write guidelines for interpreting evidence in the context of a series as a whole like this.

With us already having a devastating shortage of evaluators, it seems hard to implement.

 

[DontTalkDT]

I can't square this angle you're taking now with your earlier comments.

You've gone from saying "Yeah, if there's some questions about specific metaphysical wording, nuke it" to "It needs to lead to a plot hole, we can't assume that the original language is any more clear."

Whether a certain cosmological construct contains "other spatial dimensions" or "other universes" will almost never lead to plot holes, but such a thing is very important for our tiering.

I said "obviously the suspicion that the translation is wrong must be somewhat reasonable" in that comment you linked.

You can not just delete every translated work in existence just because a random user went ahead and said "ok, but I want a new translation of that one sentence because I suspect it of being wrong for no particular reason whatsoever" and didn't manage to get a translator. That's not a realistic way to handle the many many translated works we deal with.
The burden of proof here can't be "translation is assumed wrong until proven otherwise". The whole investigation process only starts when we have a reason to suspect something.

And no, that does not have to be a plot hole. A plot hole is merely an example of something that could lead to reasonable suspicion. Point is that there must be some justification for the assumption of a translation being wrong before we take some action on it.
If you want more methods, you could, for instance, find the original text and machine translate it. Now, your machine translation will of course not be reliable at all, but when you spot that some word is used and look up in the dictionary, to discover that it could have some other relevantly different meaning that also makes sense in context, that is enough to generate a reasonable doubt. Then one can investigate, maybe clear up the issue one way or another, or decide that we don't have the means to decide on that and take actions like deleting the profile.
But that must come from more of a ground than "I just don't like that sentence" or "this explanation is vague so let's hope the Japanese version isn't for some reason".

Another reasonable suspicion can also be if it has a history of being wrong in big ways a lot already. We already practised that. Fairly sure we have deleted verses like "Demon Bane" and "Problem Children Are Coming from Another World, Aren't They?" in the past for reasons like that, and it's an ok practice to do that when known translation issues just make things unmanagable.

When it comes to translation, you will never have the ability to investigate without restriction, be it MTL, good fan translation, bad fan translation or official translation. We can't ensure to never run into a matter where a translation was wrong and we just couldn't find out. However, it's not practical to be so paranoid about it that we can't work on translated media. Maybe we are wrong sometimes, but then we just correct it once we find out. No harm done.
Not even trying in fear of failure was never a good approach to anything.

To finish this, let me also give an example for what I consider a reasonable suspicion:
In the official translation of the novel "The Executioner and her Way of Life" there is a character called Nono Hoshizaki who wields the pure concept of "Star".
I have a reasonable suspicion that this translation is imprecise. Why?
Because the powers she displays aren't all star-based, but also include things like connecting to the memories of the planet.
The suspicion is that it might be the pure concept of "heavenly body"/"celestial body" in a more literal translation. (i.e. including planets and stars)
The fact that I know that there is the word "hoshi", which is ambiguous in that fashion, aids in supporting this possibility.
Can I confirm that is the case? Not without at least having the raws. So if I made a profile for her, one might need to discuss what to do in that regard.

I think there are cases where a translation is demonstrably wrong, in a way that doesn't matter for casual readers, but does for us trying to index it. Take this calculation, which was based on this translation:

I calculated this as him moving up our minimum height for a mountain (609.6 meters) in 0.66 seconds, for a speed of 923 m/s. This was accepted by a CGM.

When I looked into the actual raw text, it said he entered the forest on the mountain (山林に這入った), which makes the distance far lower, meaning it could be as slow as 15 m/s or so.

That is something that a good translation would easily do as well, though. In fact, it is a quite common opinion that a good translation (in terms of commercially good) shouldn't be all literal.
Every translation can be wrong at that level and there is no feasible way to check everything.

Given that I believe we can all agree that deleting every work not originally written in English is no reasonable approach, there is just no solution to this other than to work with the information you have and correct matters if new information appear that proves things wrong.
We can't raise that high a bar here.

 

[Agnaa]

I said "obviously the suspicion that the translation is wrong must be somewhat reasonable" in that comment you linked.

You can not just delete every translated work in existence just because a random user went ahead and said "ok, but I want a new translation of that one sentence because I suspect it of being wrong for no particular reason whatsoever" and didn't manage to get a translator. That's not a realistic way to handle the many many translated works we deal with.
The burden of proof here can't be "translation is assumed wrong until proven otherwise". The whole investigation process only starts when we have a reason to suspect something.

That's not the basis I provided for this example of Medaka Box.

From my first post, I've said that Medaka Box has an incorrect translation, including ones that are site-relevant like translating a phrase as "a style that surpasses dimensions" instead of as "a style that's [on another level]/[overpowered]", and generally being off enough to require extensive rewrites by a later group EDIT: Just did some side-by-side comparisons, and the rewrites were generally minor, here's an example on the upper end.

From that basis of demonstrated issues, I pointed to other scans that I have questions about.

(I think this is a useful example to keep discussing; I find it important to know whether this sort of thing can cause us to disqualify a series, and at what level of severity it would do so)

And no, that does not have to be a plot hole. A plot hole is merely an example of something that could lead to reasonable suspicion. Point is that there must be some justification for the assumption of a translation being wrong before we take some action on it.
If you want more methods, you could, for instance, find the original text and machine translate it. Now, your machine translation will of course not be reliable at all, but when you spot that some word is used and look up in the dictionary, to discover that it could have some other relevantly different meaning that also makes sense in context, that is enough to generate a reasonable doubt. Then one can investigate, maybe clear up the issue one way or another, or decide that we don't have the means to decide on that and take actions like deleting the profile.
But that must come from more of a ground than "I just don't like that sentence" or "this explanation is vague so let's hope the Japanese version isn't for some reason".

Another reasonable suspicion can also be if it has a history of being wrong in big ways a lot already. We already practised that. Fairly sure we have deleted verses like "Demon Bane" and "Problem Children Are Coming from Another World, Aren't They?" in the past for reasons like that, and it's an ok practice to do that when known translation issues just make things unmanagable.

When it comes to translation, you will never have the ability to investigate without restriction, be it MTL, good fan translation, bad fan translation or official translation. We can't ensure to never run into a matter where a translation was wrong and we just couldn't find out. However, it's not practical to be so paranoid about it that we can't work on translated media. Maybe we are wrong sometimes, but then we just correct it once we find out. No harm done.
Not even trying in fear of failure was never a good approach to anything.

To finish this, let me also give an example for what I consider a reasonable suspicion:
In the official translation of the novel "The Executioner and her Way of Life" there is a character called Nono Hoshizaki who wields the pure concept of "Star".
I have a reasonable suspicion that this translation is imprecise. Why?
Because the powers she displays aren't all star-based, but also include things like connecting to the memories of the planet.
The suspicion is that it might be the pure concept of "heavenly body"/"celestial body" in a more literal translation. (i.e. including planets and stars)
The fact that I know that there is the word "hoshi", which is ambiguous in that fashion, aids in supporting this possibility.
Can I confirm that is the case? Not without at least having the raws. So if I made a profile for her, one might need to discuss what to do in that regard.

Thanks for providing these examples, they all seem like acceptable ways to disqualify a translation to me.

I assume that "big translation issues" would be things like the case I've mentioned before, where a translator note said that they decided not to translate a conversation since it was long and boring.

 

[Executor_N0]

That's not the basis I provided for this example of Medaka Box.

From my first post, I've said that Medaka Box has an incorrect translation, including ones that are site-relevant like translating a phrase as "a style that surpasses dimensions" instead of as "a style that's [on another level]/[overpowered]", and generally being off enough to require extensive rewrites by a later group.

Technically, that isn't a wrong translation. That some people interpret "dimension" as something literal for our system is a problem with the interpretation from this side, not the translation itself. Dimension is also used with the meaning for a level or grade in English, something of "tremendous dimensions" or "dimensions off-the-chart", for example, so the fact that the original says "次元を越えるスタイル" is fine. If that was the translation from an old fan-translation, I wouldn't call it a problem from the translator.

There are some translations that are really" out there", but I wouldn't say that one is one of them. A big challenge would be defining exactly what are the types of questionable translation are that can raise reasonable doubt that we can actually put the blame on the translation, instead of being a problem of interpretation that comes from people reading too much into stuff to put into a high-level in this wiki.

 

[Agnaa]

That's not the basis I provided for this example of Medaka Box.

From my first post, I've said that Medaka Box has an incorrect translation, including ones that are site-relevant like translating a phrase as "a style that surpasses dimensions" instead of as "a style that's [on another level]/[overpowered]", and generally being off enough to require extensive rewrites by a later group EDIT: Just did some side-by-side comparisons, and the rewrites were generally minor, here's an example on the upper end.

From that basis of demonstrated issues, I pointed to other scans that I have questions about.

(I think this is a useful example to keep discussing; I find it important to know whether this sort of thing can cause us to disqualify a series, and at what level of severity it would do so)

After looking into it more, I no longer think this is a good example.

Looking into more substantive changes than grammar rearrangements, I found a case; they moved from this to this, despite the raws not saying any of the text they added (he was kinda repeating himself, the translation should make the repetition less direct, rather than adding new text).

Still, I would like us to have a rough idea on how bad grammar/mistranslations get before we treat the whole translation as suspect.

 

[Agnaa]

Bump.

 

[Pluto321]

Bump