A better ranking system is critical

While ranking movies to make a list is a great idea, the current ranking algorithm is simply not satisfying.
I want to have a list where if movie A is ranked at the position n and movie B is ranked at the position n+1, then it means that I prefer movie A over movie B, and this no matter the value of n. Let me call this a fully ordered list.
This is not the case with the current ranking algorithm where some of my favorite movies are buried deep down in the list while some movies I like more or less at in my top one hundred. Worst, a fully ordered list is impossible until all the movies are compared 2 by 2 which will simply take forever! If a list has N movies in it, it will take N! comparisons to achieve this, and that is, if comparisons between movies are not repeated as they are now. Hence, after the novelty has waned out, I realized I will never get a fully ordered list and I gave up.
Bottom line: you need a better ranking algorithm which produce always produce a fully ordered list, and that does not require to compare all the movies 2 by 2. I'm sure you already thought about it and already got suggestions from other users but here is my two cents.
Suppose a user starts a list. Give him the choice between two movies, say batman begins vs. star wars. He picks star wars. His list is now
1. Star wars
2. Batman begins
It is fully ordered.
Now pick a new movies, say ghostbusters. Pick a random point of entry in the list, say 2. The user is asked to choose between ghostbusters and batman begins. He picks ghosbusters. Then you asked it to choose between ghostbusters and star wars. He picks star wars. Since ghostbusters lost, you stop the ranking process and ghostbuster get the rank of the movie that bet it minus 1. The rank of movies below in the list is changed accordingly. The list is now
1. Star wars
2. Ghostbusters
3. Batman begins.
The list is fully ordered
Introduce a new movie, say indiana jones and temple of doom. Pick a random point of entry in the list say 1. Compare temple of doom to star wars. star wars wins. Compare temple of doom to ghosbusters. Temple of doom wins. Stop the ranking of temple of doom. Its rank is the rank of the movie it just bet. The rank of movies below the list are modified accordingly. The list is now
1. Star wars
2. Temple of doom
3. Ghostbusters
4. Batman begins
Well, I guess you get the idea. The algorithm would go something like this
1. Pick a new movie
2. If the user has seen it, compare it to a random movie on the list
3. If the user prefer the new movie, keep on comparing that movie to movies up the list until the new movie loose (its rank is then the rank of the movie that bet it minus 1) or until it reaches the top of the list (in this case it is the new number 1 movie on the list).
3. If the user prefers the randomly chosen movie over the new movie, keep on comparing the new movie to movies down the list until it wins (in this case, its rank is the rank of the movie it just bet) or until it reaches the bottom of the list (in this case, it is the new least preferred movie of the user).
Also, instead of picking randomly from the list the movie to which the new movie will be initially compared, it will be probably more efficient to use the global rating of the new movie to decide to which movie to compare it. For instance, if the global rating of the dark knight is 6, it is safe to assume that on most list, the dark knight is around the sixth position. Hence, it makes sense to start its ranking process by comparing it to the sixth movie on the list.
Finally, in order to cope with the users changing its preferences, the possibility should be given for the user to ask to rerank a movie. In this case, the process would start by comparing the movie with the movie just on top of it on the list