Just a data point: I founded a Lisp startup together with a bunch of experienced Lisp hacker buddies from the SBCL community. Sadly and reluctantly, we found Lisp awkward and ended up rewriting everything in C, and then never looked back.

These days I am developing such software with LuaJIT and that is working much better for me than either C or Lisp.

One thing I learned along the way is that many tales of Lisp heroism are actually anti-paradigms. Once upon a time when I read about ITA Software rewriting CONS to better suit their application I thought it was impressive; now I see it as a farcical workaround for having chosen an ill-suited runtime system and sticking with it (and generally an indictment of Lisp not providing a practical performance model for the heap.)

Lispers are too expert at spinning bugs as features. "It's insanely complex, every line could be an interaction with undefined behavior or a race condition or an unexpected heap allocation" becomes "suitable only in the hands of trained specialists, like a chef's knife or a surgeon's scalpel or a Jedi's light saber."

I feel like we need to have a shared "our emperor didn't have any clothes" moment with regards to Paul Graham's essays.

(I say this as somebody who does love Lisp and will probably do a lot more Lisp work in the future but only on a project that is a peculiarly good fit.)