Not advocating for MD5, but it does strongly imply the use of cryptographic hash functions for hash tables, which is almost never the case, and a useful distinction to explain.

Cryptographic hash functions are trying to make the original data unrecoverable in any way, and essentially distributing as evenly throughout their output space as possible. They are also often trying to be slow to compute to prevent brute forcing.

On the other hand, hashing for a hash table doesn’t need to be secure in the same way (hence Python’s smaller bits just returning themselves as their hashed values). They also don’t need to distribute themselves evenly. They need to be fast, unlike a cryptographic hash, and they need to restrict the data to a fixed size.