Well, my way of thinking was like that: your friend gets to the bottom of a well and you yell to him "What's down there?" - and perhaps the well is completely empty and his answer is "Nothing, absolutely nothing!".
But perhaps you see that there is something down there, but can't see what it is - so you ask "What is that down there?" and 'that' refers to that unknown thing that you do see.
That's how I interpret it and that's why I offered an a bit different translation.
Words that have ó in their basic, Nominative form, often use o in other forms. It may be possible that only Nominative and Accusative (if identical to Nominative) have it, but I'm absolutely not sure of that. I went through a few declensions and they seem to confirm it, but that may not be enough of a proof.
I thought about it and this is what I came up with:
When an inflection causes the consonant which follows the ó to move to another syllable, the letter loses its diacritical mark.
- Dół (ó and ł are part of the same syllable)
Do-le (the consonant moves to the next syllable).
Sa-mo-chód (ó and d belong to the same syllable)
Sa-mo-cho-du (the consonant moves to the next syllable).
But, there is a list of exceptions. I could come up with two so far: król and ból.