This worries me a lot. My privacy threat model since a long time has been to at least be anonymous enough that someone would have to be specifically assigned to digging through my file for days to discover my identity. With AI and surveillance by default becoming the norm, I can imagine it would be trivial to defeat my tricks and safeguards with a thorough enough dataset.

Semantic analysis of online profiles kind of feels to me like the digital version of “gait analysis” in the real-world. A thing that could be anywhere, that I don’t know how I could reliably defeat, without developing foot problems from having a pebble in my shoe all the time.

So if you see me throw out a braindead one-liner or a confident take on something completely outside my expertise, you can consider it just me throwing some “silly walks” in there to disrupt analysis of the system.