Nanocrowd: A new kind of movie recommendation engine
It looks like word salad, but Nanocrowd's movie recommendation system is both unique and very good.
Nanocrowd is a little video discovery app for the iPad that approaches the problem of finding you a movie to watch in a new way.
Rather than collect your viewing history and asking you to rank what you've seen (as Netflix and Tivo do), Nanocrowd lets you start with one movie and then it tries to tease out what it is about that film that you're liking right now. Then it finds more like that.
For example, if you say you liked "District 9," the app will give you a list of three-word options to narrow down recommendations that spawn from that. Choose "Gory Aliens Monster" and the app will recommend "Altered" and "Horror Planet." Or select "Space Humans Humanity" and you'll get "Contact" and "Blade Runner." It's easy to see how both lists are related to the seed movie, yet they clearly address two different moods someone who likes "District 9" might be in.
The app can launch a streaming app on your iPhone if a movie you want to watch is on a service like Netflix, or it can kick off a search on iTunes or the Amazon site.
Nanocrowd created its categories by analyzing the words people use on other reviews sites to describe movies. The recommendations were good when I tried the service, so based on the results, I'd say the semantic technology the company has works well.
It's a shame the app looks so bland, though. The design is just not up to modern standards. And while the descriptive word triplets do make some sense, it is awkward to read through lists of jumbles like "Humorous Russia Weapons," "War Conflict Peace," and "Jokes Comical Lovable" (guess the movie, the answer is below) when trying to find a flick.
Nanocrowd does movies only, not TV shows. The company may add them later.
This is an acquisition candidate for a streaming company like Netflix, or considering Netflix's likely future fortunes, maybe Amazon or Apple. Nanocrowd can make a few bucks by steering consumers to buy or rent online movies, but I think the best possible outcome for this company is for it to sell the cool algorithm soonest. That would explain why the company hasn't put a ton of work into the user front-end, too.
(Quiz answer: "Dr. Strangelove")