The bad words (explicit, offensive, etc.) and their variations are compiled in a rudeWords array variable contained somewhere in the comments plugin engine which then used to check the converted tag against the words that are in the variable. It checks it three times before it eventually falls to ABCD since it still deemed to have these bad words after some more thorough conversions.

In that code, the first three letters of the converted, unchecked string are being checked if there's a bad word contained in the rudeWords variable. If it does, the string is converted again in a different way. But if that still have bad words inside, then the ID is reversed so it wouldn't appear to have a bad word inside. But even if the double-checked ID still found to have bad words, then the engine decided to choose the fallback tag ABCD for the purpose. But this is an extremely rare case, most user IDs will turn out a fresh, unchecked visual ID which will mostly pass the first few bad word detection in the process.

Note that rudeWords is a Set in JavaScript, which is just like an array, but it has unique values inside and is a bit different from an array. That's why it uses .has() instead of .includes() in an ordinary array. And the lowercase tag is then being converted to uppercase when being displayed in the comments.