@kaoudis I think what you describe is a slightly different case from the one in the post: you want to explicitly detect failure, while the author doesn't (note that there are no checks for empty).
Considering that even the Big Fat Policy Cat mentioned in the post couldn't prevent digression from the stated rule, it's a good question if trusting an API in a system with nullable types is a good idea in the first place...
Anyway, Mr.Chen usually answers comments, so it may worth to bring up these points at the source too!