Actually, it doesn't work. My definition of "work" is "functions correctly, 100% of the time". I noticed a bug today: Some of the lines in the insult files rely on $1 being the nickname of the person to insult. I forgot about this. I'll fix + update it momentarily.
To thise who think what I said above was nonsense, I wish you'd have been more specific so I could actually form a response that is directly relevant. Since you were obscure about what you disagreed with, I'll consider it a disagreement to everything I said.
By writing the insult generating part of the script the way I have, it's simple, secure (providing care is taken to ensure insecure code doesn't end up in the .txt files) and extensible. This script uses a layer of evaluation native to $read. If you open up the .txt files that the insults are read from, you'll notice $identifiers used in it. Those are evaluated when a line is randomly read from the file. This is elementary scripting, and if you don't know this then you shouldn't be running scripts on Hawkee; Running scripts without knowing precisely what the script can do is dangerous. By auditing the files, you can see that my script isn't dangerous. If I were to use a server, the script would either lose abstraction, become a security issue, or that layer of evaluation would be lost and as a result the script would become a lot more complex.
Any questions or comments? Do you disagree with anything there?