Should machine-translated chapters be appropriately tagged?

[AlexEliot]

While a suggestion for this has already been made here, the matter was quickly dropped due to the short-lived characteristic of suggestion threads. This post is not meant to provide a suggestion, but rather invite discussion.

So let me start it off with a reason in its favor: considering there are countless chapters that are preceded by a disclaimer page specifying the chapter is machine-translated, an MTL tag will relieve scanlators of the need for such a page, and readers will be properly informed in a more streamlined way. The tag would only be decorative; meaning that it will not be filterable, ensuring no kind of penalty is given, so there would be no effect of chapters tagged as machine-translated getting less publicity than those without it.

Would that solve the MTL problem once and for all? If not, what is preventing it?

 

[EOTFOFYL]

A warning label
For potential eye cancer
Would be welcoming.

 

[Angry_Panda]

I agree with this idea in theory, but it comes with its own set of complications.

One problem I see with this is - for lack of a better term - 'enforcement.' Is it strictly 'on-your-honor,' with the community being relied upon to report an untagged chapter as MTL (which seems ripe for abuse), and repeated untagged chapters incurring some sort of penalty for the uploader? Or is there no mechanism for requiring the tag be used, which ultimately defeats the purpose? I suspect there are going to be some groups that absolutely refuse to use the tag on principle, regardless of how obvious their use of MTL is. And there will be some readers who will insist on having the tag used if it's available. These two groups are clearly not going to get along well.

Another problem would be defining what qualifies as MTL - if a translator uses MTL as a starting point to save time, but then does proofing/editing on the output to make sure syntax, connotation, etc. are appropriate and faithful to the source, is it an MTL chapter? It's pretty easy to catch the translations that just copy/paste from the output without any double checking, but there's a good bit of gray area in between a fully manual translation and fully MTL.

I do think it's worth discussing, though as with many things, I don't think there's a single solution that will satisfy the overwhelming majority of the community.

 

[pandascepter]

YES FUCKING YES. IMPLENT THAT SHIT RIGHT NOW. so my day arent ruined

 

[77BLc9JwJGpLVARkamJx]

Use the block button for shitty translation, mtl or not.

 

[AlexEliot]

I agree with this idea in theory, but it comes with its own set of complications.

One problem I see with this is - for lack of a better term - 'enforcement.' Is it strictly 'on-your-honor,' with the community being relied upon to report an untagged chapter as MTL (which seems ripe for abuse), and repeated untagged chapters incurring some sort of penalty for the uploader? Or is there no mechanism for requiring the tag be used, which ultimately defeats the purpose? I suspect there are going to be some groups that absolutely refuse to use the tag on principle, regardless of how obvious their use of MTL is. And there will be some readers who will insist on having the tag used if it's available. These two groups are clearly not going to get along well.

Another problem would be defining what qualifies as MTL - if a translator uses MTL as a starting point to save time, but then does proofing/editing on the output to make sure syntax, connotation, etc. are appropriate and faithful to the source, is it an MTL chapter? It's pretty easy to catch the translations that just copy/paste from the output without any double checking, but there's a good bit of gray area in between a fully manual translation and fully MTL.

I do think it's worth discussing, though as with many things, I don't think there's a single solution that will satisfy the overwhelming majority of the community.

Yes, there is no concrete way to determine whether a chapter is machine translated or not, therefore it can only be a pure opt in for the uploader, that is meant to simply reduce the "this chapter is MTL'd" extraneous pages, and save readers time rather than clicking the chapter and then going back (which applies only to those are are appalled by MTL translations).

I'm skeptical on whether the community should partake in this, because it's a very, very sensitive subject. At most, staff should be able to add those tags only when the disclaimer is clear and consice within the extraneous pages of the chapter. Petty arguments about MTL have always been a thing everywhere, so I don't think the tag will create any significant difference other than a simple "add the MTL tag lol" comments. If people hide the fact that they MTL'd a chapter, then there's nothing that can be done about it, neither will it be attempted to.

In my opinion, I qualify a chapter as machine-translated if the scanlator possesses minimal to zero knowledge of Japanese, and has used MTL tools to translate nearly every dialogue of the chapter. Surely denominating it from "MTL" just because the scanlator translated the はい themselves would be silly.

MTL tools are absolutely okay to use for a human translator. They provide invaluable assistance.

 

[Plykiya]

Considering there are countless chapters that are preceded by a disclaimer page specifying the chapter is machine-translated, an MTL tag will relieve scanlators of the need for such a page

There is no requirement to disclose it. When they disclose it, they're trying to receive less flak for uploading a chapter they know is bad but they still want to be praised as a saint for doing what no one else was willing to do. This isn't solving a problem that exists for scanlators.

Readers will be properly informed in a more streamlined way

If they needed a tag to identify something as MTL in the first place, they would have had no clue it was MTL to begin with. There is no way to identify something as machine translated unless they outright admit to it. Despite that, the average reader will now begin reporting anything low quality as machine translated because something that looks this bad definitely has to be machine translated as well.

People aren't going to treat it as something nice that the scanlator can opt-in to use.

It's not even as if we, as veteran staff of a scanlation website, are able to properly identify something as machine translated either if there is no outright confession by the scanlator. So all of these chapters not tagged as MTL are going to get reported because "this is so bad, there's no way it's not MTL" and we are going to be buried in a mountain of essentially unsolvable reports because there is no way we're going to try to get into contact with every single group for every single chapter to milk a confession out of them on whether it was machine translated or not.

This only seems like a good idea if you're not a moderator.

 

[AlexEliot]

There is no requirement to disclose it. When they disclose it, they're trying to receive less flak for uploading a chapter they know is bad but they still want to be praised as a saint for doing what no one else was willing to do. This isn't solving a problem that exists for scanlators.


If they needed a tag to identify something as MTL in the first place, they would have had no clue it was MTL to begin with. There is no way to identify something as machine translated unless they outright admit to it. Despite that, the average reader will now begin reporting anything low quality as machine translated because something that looks this bad definitely has to be machine translated as well.

People aren't going to treat it as something nice that the scanlator can opt-in to use.

It's not even as if we, as veteran staff of a scanlation website, are able to properly identify something as machine translated either if there is no outright confession by the scanlator. So all of these chapters not tagged as MTL are going to get reported because "this is so bad, there's no way it's not MTL" and we are going to be buried in a mountain of essentially unsolvable reports because there is no way we're going to try to get into contact with every single group for every single chapter to milk a confession out of them on whether it was machine translated or not.

This only seems like a good idea if you're not a moderator.

There will never be a "confirmation" involved other than whether the scanlator points that fact out themselves on the chapter. I understand and can already see the piled up "this is MTL" reports but without the scanlator's explicit disclosure these would be immediately rejected.

 

[Plykiya]

There will never be a "confirmation" involved other than whether the scanlator points that fact out themselves on the chapter. I understand and can already see the piled up "this is MTL" reports but without the scanlator's explicit disclosure these would be immediately rejected.

An immediate rejection will still require the moderator to open the chapter, investigate it for any sign of confession, etc.

It's not just a free report reject. It still requires work for something that adds little to no value to the site.

 

[AlexEliot]

An immediate rejection will still require the moderator to open the chapter, investigate it for any sign of confession, etc.

It's not just a free report reject. It still requires work for something that adds little to no value to the site.

I understand the moderator strain, however I disagree that it's little to no value. It's been requested by other people as well, and the idea has received mixed reactions—neither predominantly positive, nor predominantly negative.

 

[AthK]

So all of these chapters not tagged as MTL are going to get reported because "this is so bad, there's no way it's not MTL" and we are going to be buried in a mountain of essentially unsolvable reports because there is no way we're going to try to get into contact with every single group for every single chapter to milk a confession out of them on whether it was machine translated or not.

Then just make it so that whenever a person reports something as MTL, they get a pop-up message explaining that this tag is opt-in by the scanlator and has nothing to do with moderators of MangaDex. And maybe only pass that report to moderators if user has power user+ permissions. Since it’s assumed that power users should know about site functioning. And if someone abuses it, just remove that permission from them.

 

[Teasday]

As a quick comment on the technical side of things, we've long held that adding any kind of "tag" data to chapters is just not going to fly. It's a waste of space. As such we have two options: tagging the title (which would require new separate entries dedicated just for the machine translations) or tagging the group. The latter is a more reasonable option, but as it's essentially a "hey, block me" tag, no group is going to want to be associated with it (and thus would predictably simply lie about using MTL) and our group blacklisting solution is going to get even worse to deal with than it already is.

As I see it, the only reason some uploaders willingly disclose they're using MTL is because there are no negative consequences to doing so. An MTL tag would be a negative consequence, the disclosures would stop and people would resort to spamming the moderator queue to get groups tagged instead of just blocking them. It seems to me that adding the tag would in fact make it less clear who the machine translators are.

 

[AlexEliot]

As a quick comment on the technical side of things, we've long held that adding any kind of "tag" data to chapters is just not going to fly. It's a waste of space. As such we have two options: tagging the title (which would require new separate entries dedicated just for the machine translations) or tagging the group. The latter is a more reasonable option, but as it's essentially a "hey, block me" tag, no group is going to want to be associated with it (and thus would predictably simply lie about using MTL) and our group blacklisting solution is going to get even worse to deal with than it already is.

As I see it, the only reason some uploaders willingly disclose they're using MTL is because there are no negative consequences to doing so. An MTL tag would be a negative consequence, the disclosures would stop and people would resort to spamming the moderator queue to get groups tagged instead of just blocking them. It seems to me that adding the tag would in fact make it less clear who the machine translators are.

That's reasonable. If chapter tags don't become a thing, then there's no way I can argue in its favor. Any other way will stigmatize groups/uploaders.

 

[tristan9]

we've long held that adding any kind of "tag" data to chapters is just not going to fly. It's a waste of space

If there's broad agreement that we need this, we can iron out the technical requirements. It's not so impossibly bad, realistically.

If not, what is preventing it?

Aside from the issues of self-reporting and arbitraging raised here and in the past, it's also a bit weird of a solution to me.

The MTL part of the translation is usually not the actual problem from a reader's perspective. Instead, it's the fact that the translation is poor. I see no difference between a garbage MTL and a garbage human TL. The latter just happens to usually be less shit (because it's selecting for people that actually at least bother trying), but in many cases is still bad enough that it is difficult/impossible to read.

Similarly, I see no difference in reading experience decrease between a trash script and a trash typesetting. All the Arial/no-antialiasing scanlations are exactly as insufferable to me. Maybe even more in quite a few cases.

What we try to achieve is allowing users to know ahead of time that a scanlation is gonna be so bad they're better off just not reading it. And I don't feel an MTL tag as proposed has any chance of actually achieving this.

 

[AlexEliot]

If there's broad agreement that we need this, we can iron out the technical requirements. It's not so impossibly bad, realistically.


Aside from the issues of self-reporting and arbitraging raised here and in the past, it's also a bit weird of a solution to me.

The MTL part of the translation is usually not the actual problem from a reader's perspective. Instead, it's the fact that the translation is poor. I see no difference between a garbage MTL and a garbage human TL. The latter just happens to usually be less shit (because it's selecting for people that actually at least bother trying), but in many cases is still bad enough that it is difficult/impossible to read.

Similarly, I see no difference in reading experience decrease between a trash script and a trash typesetting. All the Arial/no-antialiasing scanlations are exactly as insufferable to me. Maybe even more in quite a few cases.

What we try to achieve is allowing users to know ahead of time that a scanlation is gonna be so bad they're better off just not reading it. And I don't feel an MTL tag as proposed has any chance of actually achieving this.

I agree. In my scanlation experience as well I have seen translations that are worse than MTL. Although, I could argue that the likelihood of a bad quality MTL is far greater than that of a human-translated script to be just as awful. There is the imaginary correspondence of "MTL = bad", but the other way around isn't always true. A translation being bad doesn't mean it's necessarily MTL'd. Though the point of the tag was never to warn users about a bad translation, but inform them it was machine translated. It's not possible to find a foolproof idea for bad translations, but I do believe it's possible to cover a quite significant subset of them.

 

[comfykotatsu]

I also see little value in the MTL tag, feels more like something bad.

Likely to happen in my opinion on the staff side of things:
=> mtl tag will most likely be perceived as something negative
=> translators wont use the tag and will stop adding a page saying that the chapter is machine translated
=> devs would have implement a feature nobody uses
=> mods receive false reports
=> mods might start to ignore, avoid or instantly reject any reports for the mtl tag, if it is perceived as a time waste
=> readers might suggest or expect filter for mtl tagged chapters

Very unlikely that any kind of communication will make the mtl tag any less negative, therefore likely to happen on the reader side of things:
=> readers will most likely use the tag to avoid machine translated chapters
=> mangadex pretty much would have encourged those readers
=> translators of machine translated chapters who at least want their chapter opened and checked out wont use the tag
=> to not receive the tag, they wont add another page telling that it is machine translated
=> readers will be even less likely to avoid machine translated chapters

Might possibly happen:
? looks like a two class system to me, normal chapter <=> mtl chapter, one is hated the other one is preferred, if history taught me one thing then that this mostly ends up shitty
? might discourage translators to upload machine translation on mangadex
? might discourage translators who do machine translations from translating mangas
? translators who do machine translations might feel unwanted by the community

Possibly valuable for the backend server:
=> one less page per machine translated chapter
=> could save up some storage space
How valuable is this? No idea.

My proposal, instead of marking chapters with something negative like a mtl tag, mark it with something positive.
=> a star for having a nice translation
=> an arrow up for having a lot of views / reads
=> a rainbow for a lot of comments
=> an elevated border for having great typesetting
=> a crown for what ever, you get the point
Anything that elavates the chapter in a positive way compared to its counter parts.
Posititvity might increase competition, motivation to improve, fame and honor but also possibly fighting and hate between groups.

 

[Aureliano]

There will be a time when a MTL will be as good as a human made one, but that is not the case yet.

You could also implement a "rough translation" tag, or an outright "rewrite" tag, to encompass all shitty translations, MTL or not.

 

[pandascepter]

how about a second rating for the tl? maybe 3rd, 4th, and so on for those godlike cl, rd, ts?

 

[pandascepter]

> What we try to achieve is allowing users to know ahead of time that a scanlation is gonna be so bad they're better off just not reading it. And I don't feel an MTL tag as proposed has any chance of actually achieving this.

Rating in the chapter?

If it's not that 'specially bad for the tag to be implemented, well what will be going to happen for the rest of the chapter? will it just be tagged to that only uploaded chapter? It is also goes to my above suggestion about rating for the TL rating, so, it is best for the reader to input a rating chapter themself which also leads to another problem. People are more likely to give it a 1 star if it not goes as what as they expected.

The best cases for this MTL problem is to give them ahead what is going on to the uploaded chapter asap. as Tistan mentioned earlier.


Perhaps PU, Group Leader, Group member, staff has the power to give em a rating? Which is unbearable thought since an uploaded chapter can be a ton and much of work. How about only the uploader can tag it? I can only see that coming to Pumpkin Night. wheeezeee

 

[sterven]

Too many staff members on general today whats going on? Did someone die?