> Shouldn't entirely new life develop all the time on earth?

It's possible OoL requires some exponentially unlikely step. We wouldn't realize that because of observer selection: had the step not occurred, we wouldn't be here to be thinking about it. There's a huge complexity gap between the stuff produced in OoL experiments and the simplest known life.

It's also possible OoL requires conditions that no longer exist on Earth, or that only existed in the early Earth (or wherever life got started in the Solar System). For example, if it depended on the existence of short lived radioactive isotopes or free ammonia.