More useful QA insights! Imagine a world where we could use QA resources to see how all the various text components fit together during a project, instead of QA only happening AFTER final text is submitted... https://bsky.app/profile/buddywaters.bsky.social/post/3lg7yp2gjqc2t
Reposted from
Buddy Waters
Having worked on this end of things, I'd add that QA are the only group looking at the text in game, and ideally would be catching context issues. But that's hard to teach and few companies value the skill enough to retain testers with it despite it really saving the translation's bacon.
Comments
https://bsky.app/profile/geekycatlady.bsky.social/post/3lgalifnmdc2e
• Unreliable accuracy - MTL likes to make up plausible-sounding things because that's how all generative algorithms operate. Catching and fixing this requires re-translation, which takes time and money, defeating the purpose of emplying MTL to start with.
- Tight deadlines
- Restrictive rules
- Priority on other more urgent bugs
https://bsky.app/profile/ktnguyen.bsky.social/post/3lgc337wd5s27