Again, if you can't stick the landing, you might as well not have any payload on it. So if you're worried about cost, keep testing until you can stick the landing with dummy mass. Once that works, send the real payload. Otherwise, you're just wasting payload.
The mindset difference seems to be that if there's no human on board, so no problemo wasting a lander if something goes wrong. That's just a bad attitude (as well as yaw and roll). If you designed everything with "baby on board" hanging in the window, you'd probably not cut so many corners so sharply. Otherwise, why not just light your cigars with hundred dollar bills. How would you feel if you were on the team building the payload, but the lander guys keep fucking up so you just wasted however much time you spent because "meh, we're just testing". In sports, there's a saying "practice like you play because you play like you practice".
Okay but successfully landing an inanimate carbon rod is easy, but why?
Who said it was easy? I'm saying they are not giving it enough respect because of the attitude of "it's only a test". That's bad. It's still expensive to get to that point. They have become complacent/lazy with the luxury of being able to iterate. Rather than spending money on engineering testing, they just build "real things" that don't work and improve the failed things. Never mind that if procedure 10 failed, you never get to test procedure 11+. So your next launch fails at procedure 11. It's just a bad attitude.
Though, of course, I wonder how many landings they are planning to do, and how many of them they need to do to compensate for each failure to land.