Struck for the name of a movie, tried to search for "Anonymous Authors" [without quotations] which came up with no predictive suggestions in dropdown nor any search results as the movie title is "Authors Anonymous". Also in tag search, the same search implementation is concurrent.

The order should not matter, many users may give up and in longer titles, the number of permutations will increase many fold. Finally, the movie title in the database may not match the exact title word order that users will commonly know the movie by, or may not match one of the many various titles it may also be known by so to maintain such a totalitarian stance that MovieLens database of movie title's is perfect in its order is asinine as well. With tags this becames an even bigger problem as there is no official authority, it is possible that "Word-1 Word-2" may have a different meaning to "Word-2 Word-1" [minority scenario] but at the very least show the user to prevent potential duplicates being added and improve ability to search.

This is sloppy programming. Search for each word individually if they are present in the movie title, tag, or <other> in the database. Only if that is mastered, add more complexity such as ranking exact matches of the search string higher, weighting first search term higher, or if the words are in the correct order.

Search of: carrot apple
Should match all of: "...apple carrot...", "...apple banana carrot...", "...carrot apple ..." at the very least.