I really like the idea of a text-only social network, so I hope this takes off. I think the author should consider supporting other languages though. Seeing romanized text isn't very helpful to people that actually want to read or write in other character sets. It's a cute feature though.
>Subreply was created by Lucian Marin from the desire of a having a simple to use, English only, public forum that has nothing in common with ancient and untrustworthy social networks.
[...]
>Limitations
>480 characters per reply ASCII only because it works everywhere
I could see this being a xenophobic thing, but I could also see it being more about the limited character set (for minimalism). I'm not making a claim about the motivations, but either way your desire seems counter to his vision.
For example, each english word has ~7 ascii chars, but each chinese / japanese / korean words only has ~2 unicode chars
Maybe it's a moderation problem?
I personally wouldn't feel comfortable running a non-english service, as I feel it would be too hard to moderate it effectively (or at least until you rearch the point where you have paid or volunteer moderators fluent in both the "main" language (to communicate rules and policy clearly) and any other supported langage). You can try auto-translating foreign text but you won't necessarily get a good translation, let alone handle cultural references (not Korean, but e.g. [1])and so on.
[1] https://www.italki.com/article/92/the-dark-meaning-of-5-chin...
(I wonder how it would tell the difference between Russian, Ukrainian and Bulgarian though?)
For maximum confusion there are also cyrillizations, how would it go with that:)
>>> probabilities = {x.lang: x.prob for x in langdetect.detect_langs(text)
Works for me, for differentiating Russian and Ukrainian text.
Looks simple enough that someone could clone it to run "Subreply for Korean."
IMO: I find mixed language social networks and threads difficult because I usually don't understand the other language.