Cosine similarity
How closely two vectors point the same way, from 1 (identical direction) through 0 (unrelated) to -1 (opposite).
Like you're twelve
Two people stand in a field and point at things. If they point at the same tree, their arms are parallel — that is 1. If one points at a tree and the other at a cloud on the far side, the arms are opposite — that is -1. If one points north and the other east, the arms are unrelated — that is 0.
Notice that how long their arms are never came into it. Only the direction.
Precisely
Cosine similarity is the dot product of two vectors divided by the product of their magnitudes: the cosine of the angle between them, independent of length. That length-independence is why it is the default measure of similarity for vectors of learned features — a doubled-magnitude vector means the same thing.
It is the score behind every similarity search in HeatherDB, and it appears in answers as well as queries: a read reports the similarity between what you asked and what came back, which is the basis of the fidelity signal.