Note that the comparative experiments in the original Bloom filter paper are very sensitive to the implementation details of the other Bloom filters they are co

Note that the comparative experiments in the original Bloom filter paper are very sensitive to the implementation details of the other Bloom filters they are comparing to. For instance, in the implementation of blocked Bloom filters, they get a false positive rate of 0.43% when using 13 bits per element. I get 0.27%, perhaps because in each small Bloom filter I am using a split Bloom filter. I emailed one of the authors of the original paper, and, while their Cuckoo filter implementation is available, they have lost their experimental setup and comparison code.