A complaint sent to Discord (via their "Help & Support" page):
Let's pretend that usernames were always restricted to the Latin alphanumeric character set. Removing case sensitivity alone reduces the number of available usernames by 2^n, wherein n is the number of characters in a given permutation (as the service no longer distinguishes between, for example: ken, Ken, kEn, keN, KEn, kEN, KeN, and KEN; or, for that matter, kent, Kent, kEnt, keNt, kenT, KEnt, KeNt, KenT, kENt, kEnT, keNT, KENt, KEnT, KeNT, kENT, and KENT; as I hope should be obvious, exponential progressions of even small numbers advance quickly).
The elimination of unique tags has an even MORE catastrophic effect, reducing the number of available usernames by 10000^n, wherein n is the number of unique permutations in a given character set (from ken#0000 to ken#9999, for example).
18% of the world population use the Chinese character set, or a derivative thereof. 14% use the Arabic character set. Another 14% use Devanagari (the Indian character set). Those three scripts alone account for 46% - almost HALF - of the world's potential Discord users, and this change prevents them from having usernames in their native script. Throw in Cyrillic and various other minority scripts like Hmong, and it's OVER half. You are alienating a majority of your potential userbase.
Now, there are a large quantity of permutations available in a case-insensitive Latin alphanumeric character set (specifically, 36^n + 36^(n+1), wherein n is the length of a given string, as iterated to the maximum string length), HOWEVER most of these permutations will be linguistic garbage, and thus harder to remember than a meaningful case-sensitive string plus a four-digit number. Or do you expect users to be able to intuitively recall whether a potential new friend's username is "aaaaaaaaaaa" or "aaaaaaaaaaaa"?
