utf8_bin collation on UNIQUE columns needs to be used with caution. In latin1
f == F so a duplicate error would be thrown if a user entered in 'F' when 'f' already existed.
With utf8_bin f != F so, make sure to normalize your data before sticking it into the dblayer i.e. put all email addresses, tags, etc in your db as lowercase :)
Thanks for reminding me Peter Z!!