A policy rarely used for handling collisions in hash tables: discard the old colliding entry! This is appropriate for using a hash table as a cache, e.g. for memoization. Or push the oldest, or one of the older, entries out of a small bucket, of four or eight items, like a 4-way or 8-way set-associative CPU cache.
If you hash the key twice, so that it hashes into two separate buckets — rather like cuckoo hashing — then the chance of survival goes up significantly. You check both buckets before declaring that it’s not found. In this case, you should probably rewrite the entry into both buckets when read-accessing it, so that more-often-read entries will be erased with lower probability. This is an alternative to using 4-way or 8-way set-associative caches; I suspect it will turn out to be more efficient.