For “Really Like” Movies, Trust Your Memory More Than IMDB and Rotten Tomatoes
As of today, there are 1,984 movies in my database. Of those 1,984 movies, I’ve watched 446 movies more than once. The general rule I’ve set for myself is to wait at least 15 years before seeing a movie an additional time. I’ve found that over that time span I retain a general idea of what happens in the movie but forget the specific details of what happens. When I watch the movie the additional time it feels like a new movie to me.
I find that, while I forget the specifics of the movie, I maintain sense of whether I liked it or not and that often times drives whether I want to see it a second time. The question I’m exploring today is; When deciding to watch a movie a second time, should I trust the impression of the movie I remember or the recommendations of IMDB or Rotten Tomatoes?
The data suggests that the answer to this question is pretty clear-cut. The movies that I’ve seen before have a much greater probability that I will “really like” them whether or not they are recommended by IMDB or Rotten Tomatoes.
|When I watched a Movie, I had:||And it was IMDB:||Probability I Will “Really Like”|
|Seen Before||Not Recommended||69.2%|
|Never Seen Before||Recommended||50.6%|
|Never Seen Before||Not Recommended||33.6%|
|When I watched a Movie, I had:||And it was Rotten Tomatoes:||Probability I Will “Really Like”|
|Seen Before||Not Recommended||65.1%|
|Never Seen Before||Recommended||49.8%|
|Never Seen Before||Not Recommended||31.8%|
While there are qualitative differences for IMDB and Rotten Tomatoes recommendations, the movies that your memory wants you to see a second time are clearly superior indicators of whether it will be a “really like” movie or not. And, the probabilities are higher for a non-recommended movie you’ve seen before than a recommended movie you haven’t seen before.
The other aspect of this data is that it reveals the weaknesses of IMDB and Rotten Tomatoes as recommenders of movies you haven’t seen. Strip the data of movies that you have seen before and a recommended movie has only a slightly better chance that you will “really like” it than selecting a movie randomly from the database. It reinforces the notion that these sites are better at weeding out movies that you won’t like rather than identifying movies that you will “really like”. Later this week, in Thursday’s post, I’ll share how this cut of the data looks for the websites that are sensitive to my individual taste in movies. Theoretically, it should better replicate the wisdom of movie memories.
For now, though, the wisdom of crowds of moviegoers and critics can’t compete with those golden memories you have of great movies seen in the past when it comes to “really like” movies.