Comment by tromp

This reminds me of the HyperLoglog [1] algorithm for approximating the cardinality of a huge set with limited memory.

[1] https://en.wikipedia.org/wiki/HyperLogLog