Your bad language filter will not work. Here is why it will not work.
(Inspired by the long-standing Scunthorpe problem, such items as ‘Your spam solution will not work’ and ‘Your calendar reform will not work’, and mostly my recent discovery of the PoliteType Initiative which ticks many of the boxes but not all.)
approach to automatically detect, filter and moderate bad language without the need for human intervention. Your idea will not work. Here is why it will not work.
[ ] Your filter doesn’t let me say ‘Scunthorpe’
[ ] Your filter doesn’t let me say ‘Lightwater’
[ ] Your filter doesn’t let me say ‘Pakistan’
[ ] Your filter doesn’t let me say ‘Enola Gay’
[ ] Your filter doesn’t let me say ‘abbot’
[ ] Your filter doesn’t let me say ‘fatty acid’
[ ] Your filter doesn’t let me say ‘spice’
[ ] Your filter doesn’t let me say ‘classic’
[ ] Your filter doesn’t let me say ‘hearse’
[ ] Your filter doesn’t let me say ‘prickling’
[ ] Your filter doesn’t let me say ‘magna cum laude’
[ ] Your filter doesn’t let me say ‘peacock’
[ ] Your filter doesn’t let me say ‘skyscraper’
[ ] Your filter doesn’t let me say ‘muffled’
More generally, your filter appears to suffer the following flaws:
[ ] Filters many innocuous terms
[ ] Ignores many offensive terms
[ ] Doesn’t account for foreign names
[ ] Doesn’t account for English names
[ ] Doesn’t account for technical terms
[ ] Doesn’t account for everyday language
[ ] Alteration sometimes looks more offensive than the original
[ ] Doesn’t inform the writer what the issue is
If your filter were ever to gain widespread adoption, it would trivially be circumvented in its main function by slur measures such as:
[ ] abbreviating to ‘a-hole‘
[ ] misspelling to ‘phaggot’
[ ] synonymizing to ‘dickslurper’
[ ] character substitution to ‘wh0re’
[ ] Regional Indicator Symbol 🇳
[ ] Inserting empty <b></b> tag pairs mid-word
While also having negative impact in several other areas including
Furthermore, the following philosophical objections apply:
[ ] You’ve never bloody well heard of the Scunthorpe problem, have you?
[ ] It’s still censorship even if you like it and mean well
[ ] Doesn’t account for gradual adoption
[ ] Doesn’t account for indefinite opt-outs and hold-outs
[ ] Regexes don’t work that way
[ ] Regexes can’t work that way
[ ] Fonts shouldn’t work that way
- [ ] Fonts shouldn’t perform man-in-the-middle attacks
- [ ] Fonts shouldn’t replace a two-letter word with a three-word phrase
[ ] Heavily bowdlerized synonyms will alter meaning
[ ] Replacing curse words with funny words results in people reading hostility into those funny words
[ ] Widespread collateral damage to innocent people’s conversations
[ ] Likely to teach people more slurs as they notice the filter’s workings
[ ] New slurs will develop that you haven’t accounted for
[ ] Will require more user effort to deal with than the moderator effort it saves
[ ] Imposes an enormous burden on ESL speakers
[ ] This has been tried and failed so many times that we now have stock phrases like ‘Scunthorpe problem’ and ‘clbuttic mistake’ and your approach doesn’t bring anything new to the table
In closing, I would like to say
[ ] Clever idea, but you have a long way to go before this is usable for the general population
[ ] This was a bad idea from the start, and I’m afraid you’re wasting your time
[ ] This is a downright malicious idea, and you’re making the world just a little bit worse