Why use this instead of putting a bloom filter in front of a fixed hash table? In the unlucky case that two pairs have total hash collision with the IBLT you ca

Why use this instead of putting a bloom filter in front of a fixed hash table? In the unlucky case that two pairs have total hash collision with the IBLT you can't recover anything. Iteration cost is the same. If you need delete just use a count in the filter instead of a hash. If the table fills up, drop entries. I guess the IBLT would be able to recover info if the number of insertions crossed the threshold and deletions brought the count back down.