IDK some of you probably know this but it looks like both Google Translate and DeepL use English as their "pivot" language, e.g. they don't "really" translate from, say, Chinese to, say, German directly, they actually quickly translate Chinese to English and then German.
@sidereal that seems like it creates a strong likelihood or unnecessary mistakes, no?
@riotmuffin Yes but it is easier than actually developing comprehensive translation software, and most of the English-speaking internet where the programmers' paychecks come from will never notice.
@sidereal @riotmuffin Makes sense, a quick search suggests google translate supports > 133 languages. If there was a direct translator between each of them, that would be more than 133 factorial. In the ballpark of 1*10^226 translators (a 1 with 226 zeroes after it in case anyone doesn't know the notation). Just names for each translator would take massively more storage than we could possibly produce.
@pavel @sidereal @riotmuffin Agh right, factorial would be combinations of all of them. Combinations of just two is in the ballpark of 8000 pairs (assuming I did the math right this time).