You are confusing sample size with population size. The sampling error depends on the sample size, but not the population size (for large populations). In the example you gave, the sample size you use certainly matters. In your example, it makes no difference though, whether you are using a 52 card deck or a 52000 card deck, your conclusions are the same.
Fine. What is the population size in terms of the volume of geologic data? What is the sample size to date? Let's get those figures, and then we can work out the probabilities.