I think what many seasoned developers appreciate in a LISP is in part its appreciation and reflection of this pervasive „unknowability“. Clojure goes even further with this (open maps, extensible protocols, motivation behind spec, and many other design choices). I think at some point or other we‘re going to have to accept that static design and formal systems are very nice for any formally specified, well defined context, but quickly become extremely costly and bloated when used in the maelstrom where chaos and order meet.
To me the „LISP curse“ is exactly that - using it you have to face how few assumptions you can actually make about your problem space/the world. It doesn‘t easily let you escape into that illusion of well-boundedness and predictability, there is no go-to approach for solving a particular problem - most answers to any question will be very elaborate and start with „it depends“. It requires a developer to leave that comforting notion, and syntax is of course an easy excuse.