I get 1132 spoonfuls as a lower bound.
Big assumptions:
- Soup is considered identical to water for the sake of molecule count and density.
- Soup and water are evenly mixed before each spoonful.
- Each spoonful is guaranteed to contain the correct fraction of soup to water from the mix.
There are ~1.666×10^25 molecules of water in 500 ml (source: WolframAlpha). We seek what power of (500-25)/500 [= 19/20] is small enough to counter this number in order to get to the level of single molecules. This is about 1132.
But like you point out, it's going to be tasting watery a long, long time before that happens. It's 50% rainwater after about 14 spoonfuls (Sanity check: That would be 10 if the container was big enough and no spoonfuls were being removed.). ~90% at 45 spoons and ~99% at 90 spoons.
Gutei's finger.
Also: He who thinks he understands Zen does not understand.