But you can't send the new emoji over basic SMS, because SMS, uses a variant of UTF-16 from the era when people thought 16 bits was big enough. (So do Java and Windows, although there are hacks in both to get past 2 bytes.)
The new emoji are all up in the astral planes, beyond 2 bytes.
Hacks? It's called UTF-16 surrogate pair. Not hacks. UCS-2 officially, but UCS-2 has been legacy for almost 20 years now. Modern phones select the encoding dynamically and use UTF-16 if the message cannot be encoded otherwise. You'll get up to 160 7-bit characters (some symbols take two characters) or 70 UTF-16 code units in unicode mode.
I just wish everyone started to use UTF-8 already and dropped all the other nonsense.
Emoji has always been beyond 2 bytes. The Unicode spec also includes "surrogate pairs" which allows a higher plane code point to be represented as 4 bytes.
for those not getting the joke: mysql has a thing called "utf8" which is not in fact utf8 and will (depending on settings) either truncate text when it meets a 4+ byte character, or raise an error.
It also supports real utf8 in more recent versions calling it "utf8mb4"
This made filing a bug about Thunderbird not sizing astral plane code points correctly slightly more hilarious than it should have been (Mozilla's Bugzilla instance runs on MySQL)...
If that were the case, astral characters would be removed (e.g. shavian or the emoji block) but BMP symbols like U+263F MERCURY or U+262C ADI SHAKTI would be left alone, and they're not: "", ""
Since U+A420 YI SYLLABLE JJUOX goes through (ꐠ) HN likely explicitly strips most symbols (but not all, U+00B6 PILCROW SIGN (¶) is unmolested)
