Well, also notice that I said "bit-flipped input." I.e., the use case I believed to be talking about and responding to was data corruption detection (which is a

Well, also notice that I said "bit-flipped input." I.e., the use case I believed to be talking about and responding to was data corruption detection (which is also associated with "checksum" as a term of art, as opposed to a generic "hash"). I agree that hashing for identity is a different use case than checksumming for data integrity. FWIW I think commonly used checksums are basically ideal ("equally likely" to come to a different value) on single-bit flipped inputs, but don't quote me on that.