Friday 22 June 2007

Auto-generated words minus profanity

I read an article on Boing Boing recently about some programmers who had to make sure randomly-generated strings contained no profanity. They were sitting around a table brainstorming rude words to check for when the intern suggested they drop the vowels and use "base 30" (ie 0-9 plus the remainder of the alphabet) and that seemed to solve all their problems.

Anyone with a little creativity can see that it's still possible to convey profanity in such a system: use 1 for I, 0 for O and V for U and you've reinstated about 75% of English profanity right there. If you stretch a bit to using 4 for A and 3 for E, you haven't dropped the vowels at all.

Mokalus of Borg

PS - I wonder if they ever came up against that problem again.
PPS - Even with profanity filtering, sometimes you get insults.

No comments: