Board Thread:Consensus Track/@comment-34115-20180618163709/@comment-3279282-20180624010204

That still doesn't create a clear line between databook translations and text based on databook translations. Imagine a specific sentence that describes how a technique works. Due to copyright violations, we wouldn't be allowed to use a translation of that sentence. Dantman says we'd have to reword it. Now, to reword it, we would have to find a sentence that conveys the same meaning, but uses other words. However, due to the nature of translations, this new sentence would also be a viable translation, since it shares the same meaning. Thus, it is impossible for us to find a sentence that conveys the same meaning, but cannot be translated from the original. This is not often the case for general-language translations, but for translations within a specific area (such as legal translations), there is often not much room left and the translator is bound to use specific terminology.

So, who or what determines when a text is far away enough from the original to not be a translation, but still conveys what the original wanted to say?