I’ve always hated how localization agencies play favorites.
Working with the same few people per language, giving them the bulk of the workload, and not letting any of them fully baton pass unless they get too expensive or croak.
Of course I get why it happens, but that doesn’t make me hate it any less.
While freelancing, it annoyed me to no end how there were no chances to prove my worth at most agencies just because someone got there first.
Had I kept waiting for a shot, I would probably be long gone from this industry by now. So, as a personal vendetta, I always try to take a chance on one applicant with each recruiting round at loclait.
We’re not talking mentorship, cause no offense, I have no time for or interest in babysitting. It’s more like, ‘you don’t quite clear my standards, but you’re decent and passionate enough to try out.‘
So while not as involved, it does require quite a bit of involvement still.
Recently, I took a chance on one such passionate translator and paired them with a more experienced one, as usual, on a pretty important project. They didn’t excel in our test, but they were promising.
Normally, I would not have given them half a project as a first task, let alone a prestige one to boot. But this was a rare already released title—playthroughs are available, the game itself is available (and was provided), and I wrote a mini style guide for them too.
They did not do great. Despite multiple discussions and 1-to-1 guidance sessions during the project, their output never quite stabilized, let alone matched our second translator’s.
There were structural mistakes, comprehension mistakes, mistakes that’d only make sense in the complete absence of references, and no preservation of voice whatsoever.
I ended up rewriting almost all of their contribution—half the project—while scrambling to wrap up what I’m actually getting paid to do in time.
There was very little difference between their output and AI output. Both drifted off base even with context provided, lacked consistency, and needed a logical correction pass before even attempting to preserve character voice.
The only difference is AI is a $20 subscription that would at least give me plenty of time to edit its flawed output instead of surprising me with it in the final phase.
But I’ll always bet on humans, because a human can do better, be better.
We (actually, I) ate the loss, paid them in full, and provided quite the wordy performance review, making something very clear:
Some degree of misalignment is expected in first collabs, so when possible, your first fuck up is on us to clean up, but the next is always on you to fix.
As for the one after… welp, you’re still on cleanup duty. But there won’t be a next, I’m afraid.
Will things work out? That remains to be seen. That’s probably another (less risky, deffo less prestige) collab away.
Even if they don’t, I’ll never stop betting on humans. I’ll just stop betting on one person.
By Seif Chaib.
Originally posted on Linkedin.