Your explanation looks correct! The practical application is having a webserver that runs a bunch of requests in parallel but that can all share the same memoi

Your explanation looks correct! The practical application is having a webserver that runs a bunch of requests in parallel but that can all share the same memoization cache. We don't want to lock the full memoization cache and therefore block all the other requests when one thread writes a new value.