It's more like editing the raw binary of an executable. It's only more precise if you understand the binary representation and the impact of changes. Otherwise it's incredibly imprecise, as you are just randomly flipping switches.
In the case of genetics, we have a some understanding, but it is fairly limited. There is real potential for unexpected harm, given how much we don't know and the complexity of the code. Of course, we can do some amazing things with the parts we do understand.
Selecting breeding is like using a high level GUI that is decently functional. Like most GUIs, it significantly restricts how you can modify the underlying data. The plus side is that it is harder to put the data in a bad or dangerous state using only the controls offered by the GUI.
In the case of genetics, we have a some understanding, but it is fairly limited. There is real potential for unexpected harm, given how much we don't know and the complexity of the code. Of course, we can do some amazing things with the parts we do understand.
Selecting breeding is like using a high level GUI that is decently functional. Like most GUIs, it significantly restricts how you can modify the underlying data. The plus side is that it is harder to put the data in a bad or dangerous state using only the controls offered by the GUI.