So VERY cool: how to count a billion objects quickly and not using much space
by cliffelam
Somewhat technical, not mathy, and fascinating.
ex: count the number of distinct words in Shakespeare's works:
Hashset uses 10M of space and is perfectly accurate
Linear Counter uses 3K of space and is 1% inaccurate
HyperLogLog uses 512 bytes and is 3% inaccurate