Charles Stross discusses some of the second-order effects should the US Secret Service actually get the sarcasm-detection software they’re reportedly looking for:
… But then the Internet happened, and it just so happened to coincide with a flowering of highly politicized and canalized news media channels such that at any given time, whoever is POTUS, around 10% of the US population are convinced that they’re a baby-eating lizard-alien in a fleshsuit who is plotting to bring about the downfall of civilization, rather than a middle-aged male politician in a business suit.
Well now, here’s the thing: automating sarcasm detection is easy. It’s so easy they teach it in first year computer science courses; it’s an obvious application of AI. (You just get your Turing-test-passing AI that understands all the shared assumptions and social conventions that human-human conversation rely on to identify those statements that explicitly contradict beliefs that the conversationalist implicitly holds. So if I say “it’s easy to earn a living as a novelist” and the AI knows that most novelists don’t believe this and that I am a member of the set of all novelists, the AI can infer that I am being sarcastic. Or I’m an outlier. Or I’m trying to impress a date. Or I’m secretly plotting to assassinate the POTUS.)
Of course, we in the real world know that shaved apes like us never saw a system we didn’t want to game. So in the event that sarcasm detectors ever get a false positive rate of less than 99% (or a false negative rate of less than 1%) I predict that everybody will start deploying sarcasm as a standard conversational gambit on the internet.
Wait … I thought everyone already did?
Trolling the secret service will become a competitive sport, the goal being to not receive a visit from the SS in response to your totally serious threat to kill the resident of 1600 Pennsylvania Avenue. Al Qaida terrrrst training camps will hold tutorials on metonymy, aggressive irony, cynical detachment, and sarcasm as a camouflage tactic for suicide bombers. Post-modernist pranks will draw down the full might of law enforcement by mistake, while actual death threats go encoded as LOLCat macros. Any attempt to algorithmically detect sarcasm will fail because sarcasm is self-referential and the awareness that a sarcasm detector may be in use will change the intent behind the message.
As the very first commenter points out, a problem with this is that a substantial proportion of software developers (as indicated by their position on the Asperger/Autism spectrum) find it very difficult to detect sarcasm in real life…