'Relatedness' has little to do with how much DNA is actually shared, which is confusing. If a sexually reproducing organism has a new mutation, how likely is it that it will be passed down to a given child? 50% Granchild? 25%...and so on.
For Hamilton's rule at least, it's more like how much DNA is shared between two members of a population over and above the population average - which means you can actually get negative relatedness, and that relatedness is completely contextual! So perhaps you could have a high relatedness to 2 bananas - if only you, two bananas, and lots of bacteria lived in a population together.