This is a consequence of the rules for block quotes and lazy continuations (and note that CommonMark’s behavior is not idiosyncratic here: Babelmark 2 - Compare markdown implementations).
To understand it better, first think about a case like this:
> foo
bar
That’s standard laziness – the whole thing is a block quote. Now, what about
> foo
bar
>baz
Laziness means that you can omit the leading >
, but you don’t have to omit it; here, it is omitted in the second line but not the third, but again, the whole thing is a single blockquote containing a paragraph with “foo bar baz”.
Okay, now we can put that whole thing in a blockquote:
>> foo
> bar
>>baz
This should give us a blockquote whose contents are the contents of the previous thing – so we have a blockquote containing a blockquote containing a paragraph with “foo bar baz”.
If we add another layer of >
to put this whole thing in a block quote, we get
>>> foo
>> bar
>>>baz
which as expected is a triply nested blockquote containing a paragraph with “foo bar baz”.
But now, by the rules for laziness, we can omit the opening >
on the second and third line, without changing the meaning. So we get our Example 160.
I hope that helps explain why Example 160 is interpreted the way it is. (Obviously, it would not be good style to write Example 160, but the rules do give it a meaning.)
I’ll agree that the eye naturally parses Example 160 as containing three separate levels of blockquote nesting. But, given Markdown’s commitment to laziness, it just can’t be interpreted that way.