Reporting MTL releases

Posted in

#126 by usagi
2021-08-31 at 21:26
< report >
Anyway, sorry for the bad handling, Usagi.
It's ok ) I found that old translation pretty much by chance anyway - and it's not like something bad happened while this release wasn't marked as MTL many years already. Actually, I think it's one of the last eng>ru MTL translations that could be detected casually. Nowadays eng>ru google translate algorithms improved to such extent that it can translate pretty complex texts almost without mistakes. And after light editing it would be nearly impossible to tell whether it was initially translated by machine or human.
In that sense the line "MTL doesn't equate bad tl" can become somewhat outdated in relation to nonasian translations. I mean - if translation algorithms for nonasian languages became that good now - what it means to read new really bad translation of said languages? Does it mean that translator was so incompetent that he could turn pretty decent result of machine work in something glaringly bad? It's.. unlikely, isn't it? Probably it would mean that he didn't bother to edit text at all - first, and second - that it was exactly MTL.
For translation from Japanese this logic do not work though - due to it being too context heavy language - and therefore - too hard to translate by MTL properly.Last modified on 2021-08-31 at 21:32
#127 by beliar
2021-08-31 at 22:12
< report >
Actually, I think it's one of the last eng>ru MTL translations that could be detected casually.

Really? Because I'm highly suspicious that early RU piracy groups like Fargus, Master Media, 7-oj volk, Macho Studio all used MTLs for their translations. Would be great if someone could confirm that, but I'm not gonna ask anyone to commit to such an undertaking.
#128 by poudink
2021-09-01 at 02:41
< report >Was mtl in the early 2000s even good enough for that?
#129 by usagi
2021-09-01 at 03:40
< report >
Really? Because I'm highly suspicious that early RU piracy groups like Fargus, Master Media, 7-oj volk, Macho Studio all used MTLs for their translations.
That translation was done in 2015 whereas those you mentioned were made much, much earlier. So I don't see any contradiction in my statement.
#130 by irx
2021-09-01 at 11:18
< report >
Was mtl in the early 2000s even good enough for that?
Not really, didn't stop the pirates and even some commercial publishers of the era like Akella though. Most if not all of those games may have something like 10-30% of manual translation (typically of the mediocre to decent quality) which at some point turns into an unedited mtl mess, with scripts broken by mtl as a bonus which crash the games to save some sanity of those trying to comprehend them.
#131 by hirary
2021-09-03 at 16:59
< report >All the things that this guy translated (this is not actually a group) are MTL. I talked personally with him one time about that, and it's obv. link
#132 by stealthswor
2021-09-04 at 00:47
< report >r82663
#133 by gambsgambs
2021-09-04 at 02:45
< report >Clover Moon's patch isn't MTL, it's a human translation which uses MTL as a starting point. It just makes more sense to do this than starting with an empty file for efficiency reasons
#134 by ninigi
2021-09-04 at 03:06
< report >^Sureeee, it's not MTL alright.
#135 by gambsgambs
2021-09-04 at 03:19
< report >If bilinguals look at the MTL and change the parts that are wrong or awkward, it is no longer MTL but a human translation. Clover Moon is doing this before releasing their patch

If you argue that edited MTL == MTL, then I could make the argument that there is a very bad machine translator that takes Japanese text as input and outputs a completely empty file. Every other translation group is editing this machine translated text. Therefore all translations are edited MTL.
#136 by shukumeiteki1
2021-09-04 at 03:57
< report >Right now the translation is MTL, they say it's a "placeholder" and want to do a real translation but so far they haven't gathered a team even so until they do it should be tagged as such.

Even a fixed version of a MTL is just that, if the source text was translated via software no matter what you do with it it will still be MTL'd.
#137 by gambsgambs
2021-09-04 at 04:08
< report >They claim to have gotten many Japanese translators: link

And I strongly disagree with “fixed MTL still being MTL”. The idea is to change anything that you yourself wouldn’t have written. So many of the simple sentences would stay the same but the long or complex ones would get massively rewritten. The end product should be entirely indistinguishable from a human translation if they do it correctly; the MTL is literally only for efficiency’s sake and would have no bearing on the finished product
#138 by cubky
2021-09-04 at 04:54
< report >Never thought I would see gambs of all people defending MTLs. What a world we live in.
#139 by adamstan
2021-09-04 at 05:26
< report >@138 - I hate to admit it, but he has a point here. If you just edit MTL's awkward English to make it somewhat readable, then the result is so called "edited MTL", and deserves to be flagged as such, since it still shares the fundamental flaw of MTL.
But if instead you start to actively check and correct the translation errors it made, then the output stops being MTL - you no longer stick to the software's (possibly wrong) translation, but replace it with your own. And if by any chance the software got some of the simplest sentences (like "Here's the school" or "Weather today is nice") right, then there's no harm leaving them in, if you'd translated them the same way.
Some people prefer doing things from scratch, and perhaps it's most efficient way, but there are some people that feel more comfortable "fixing" previous work no matter how bad it is, even if that means the same or more workload.Last modified on 2021-09-04 at 05:34
#140 by moyang
2021-09-04 at 18:44
< report >I added some MTL patches: r82677, r82678, r82679
#141 by usagi
2021-09-07 at 00:23
< report >r81678 is MTL. Proof link see comments
r9424 is MTL. Proof link see comments
#142 by naiohoras
2021-09-08 at 11:22
< report >r80948.1
#143 by poudink
2021-09-08 at 13:11
< report >r47264 is edited mtl according to link
#144 by poudink
2021-09-08 at 17:48
< report >r82785
#145 by poudink
2021-09-08 at 18:14
< report >r82786
#146 by 4digitmen
2021-09-10 at 17:16
< report >r82819 one of the columns link

I'm pretty sure this is a dupe entry to r82663 since Clover Moon is working on itLast modified on 2021-09-10 at 17:16
#147 by lemonov
2021-09-11 at 23:52
< report >r82283
In the comments to this VK post link the translator (Sergey Romanov) pretty much admitted that he doesn't speak Japanese and is translating "using a dictionary". Does this mean that this should be flagged as MTL or not?Last modified on 2021-09-11 at 23:53
#148 by ezezin
2021-09-13 at 04:34
< report >The English translation of r82923 is a MTL. Source: link
#149 by historyeraser
2021-09-15 at 09:28
< report >r82974 MTL according to description
#150 by historyeraser
2021-09-18 at 06:55
< report >untag r82663 as MTL per t17026.8