I think this should be move up in the allowed example, earlier in this context.

Also, how is this different from the behavior above in practical terms?

This was a spec that described incorrect behavior. As noted in #14420, the previous behavior of this test was to auto-correct from var.nil? ? nil : var.foo * 42 to var&.foo * 42, which is not safe. To make it safe, the correction would have to be var&.foo&.*(42), but that would reduce readability.

This is a more complex version of examples already documented.

#   foo ? foo[index] : nil    # Ignored `foo&.[](index)` due to unclear readability benefit.                                                                                           
#   foo ? foo[idx] = v : nil  # Ignored `foo&.[]=(idx, v)` due to unclear readability benefit.                                                                                         
#   foo ? foo * 42 : nil      # Ignored `foo&.*(42)` due to unclear readability benefit.  

https://docs.rubocop.org/rubocop/1.79/cops_style.html#examples-stylesafenavigation

So the documentation has not been updated.

I meant that https://github.com/rubocop/rubocop/pull/14421/files/da0e3e7a1fb6bf65a17f32fcfa3e5e44f9aec36d#diff-736e663e58ab853ebfddd666306d7bddb37502d4cde5c1b8703cac7f1b20dcc8R871 is essentially the same.

And if you go up in the specs group you'll see that the allowed examples are earlier, so I think those should be grouped together.

Ah, I see. In cases like foo.*(42) or foo.[](42), where the method call already includes a dot, converting them to safe navigation like foo&.*(42) or foo&.[42] doesn't reduce readability, so an offense is registered.

On the other hand, for calls like foo * 42 or foo[42], which don't include a dot, no offense is registered in order to preserve readability.

So, there's a distinction based on whether operator method call and index access are written with a dot.
I structured the specs to reflect this symmetry.

Oops, I misunderstood the intent. I've updated it to keep the diff simple.