E.g. looking at the quality of Microsoft's German translations (which seem to be done by a poorly trained bot), the translation is actually counterproductive, because it's full of special terms translated into German that either would not be translated from English in a "native" German text, or where a different translation is commonly used.
My favourite is the Visual Studio translation for Link-Time Code-Generation. Which is translated as "Link-Zeitcode Generierung" which in English means "Link Timecode Generation".
In other countries like Russia or Japan I noticed, as an outside observer, more reliance on translations.
When translating a complication is the degree to which one does it. There are terms which can be translated and some where it becomes weird. In some parallel comment "pull request" was mentioned. Translating that certainly becomes weird, but translating "file" as "Datei" is well established and - when translating - expected.
I do agree that it’s not a walk in the park because it takes loads of resources to provide actual translations and one has to make i18n a feature in the code. But... come on. phpMyAdmin was doing this in 2004.
However finding the right translation level in technical context is hard. For which concepts should one translate the name and which terms should one borrow the English term? The French translate "computer" as "ordinateur", the German "Rechner"/"Rechenmaschine" however is outdated.
Now what about a git repository clone and commit? How much translation will destroy understanding?
Unfortunately that's misunderstanding that fundamentally language relies on way more context than just strings to get the same message across.
You can't just simply translate and have a good UX for other languages.
Not translating isn't really a solution.
Wanting everything to be 100% translated is also unrealistic.