The original point was: "The idea that you can engineer yourself an addictive game is a silly concept."

The rebuttal was: "this is literally what casinos do"

The rebuttal to that was: "well, some casinos fail"

Which isn't really a point. Some casinos fail, that doesn't show that they don't optimize for addiction. It just indicates that optimizing for addiction is necessary but not sufficient for a casino to maintain solvency.

I think we have fundamentally different of andrewla's original point, and that is OK.

I read it as pointing out that success isn't guaranteed with an addictive game model.

I did not read it as saying that some games don't or cant optimize for addictiveness.