Why I started a free, on-demand curation service for Netflix | Hacker News Reader

Why I started a free, on-demand curation service for Netflix

6 pointsJMC08x9 comments

I started Sense (www.wemakesense.co) because I increasingly became concerned with how Netflix was handling the recommendation portion of their product. I saw an ever-consuming reliance on data and computers that strayed away from what recommendations should be about. I wanted to make recommendations more human.

Below are some quick, short examples of the issues that Sense is working on correcting:

- Your recommended 'Top Picks' is a list of 25+ movies

This is a subset list of a list. You're already swarmed with movie choices in Netflix. You still have to decide between a ton of options here

- 1-5 star rating systems and predictors miss the point of recommendations

The goal of a recommender is to give you a movie that you'll enjoy spending your time in. It's not to give you something you'll like '4.223' or '3.4445' stars out of 5. The essence of a recommender is lost in this detail

- Recommending fully done by computers. No personal touch.

Human curation can be infused into the system without total overhaul

- You can love Fantasia without being a kid

After watching Fantasia, you'll be presented with suggestions to watch children's movies next. This is because of a reliance on a 'kids' tag to the movie. But Netflix knows more about me than that. They know I am a male in my 20s. I can appreciate Fantasia for other reasons

- 'Top Picks' and 'More like this' are a scatterplot of all kinds of different movies

This does little to center in on the things people enjoy within a specific movie. Instead it feels like darts being thrown at a dartboard

Unless and until we can recommend things that aren't available via netflix, this invention remains another crowdsourced, yet ironically still proprietary, commercial product. Nexflix already had my viewing habits, my daily routine, and my money. They shouldn't need my help too.

Netflix needs/wants in-house recommendations because any external recommendation service risks drawing attention to their limited offerings. Netflix has around 20,000 titles for Americans, closer to 6000 it seems for canadians. That's only slightly better than a well-stocked blockbuster video. Nearly every old show recommended to me by friends, by actual flesh-and-blood people, isn't there.

Also, What happens to recommendations once the shows they recommend are dropped from the netflix lineup?

We have a system for knowing what is on and isn't on Netflix at any given time - We only give you picks that are currently on the platform (in US)

Ah. So this is totally useless for we canadians. Every time a Canadian hits "this content isn't available in your region" a pirate is born.

Clickable: http://www.wemakesense.co/

I agree recommendation engines, as a rule, are pretty bad. Would be interested to hear more about you approach this problem. Does your approach scale? What's the eventual business model if it's free? Paid placement?

The internal process that a human (you in this instance) is doing can be process mapped out, engineered and then implemented as a automated software solution.

You have just got more criteria then Netflix implementation.

Key part of the criteria: limiting what is considered appropriate enough to be pushed to users as a recommendation

I'm just using it as we speak, pretty cool, can't wait for the movie suggestions. I wish such service exists for Netflix.ca though. Also great idea for books reco ;)

I'm actually working on a service just like this, but not specifically for Netflix. Since a lot of households nowadays have multiple streaming services and I think we should't forget that people still actually buy/rent movies online a service that not only provides the consumer with great movie exploring possibilities but also the possibility to actually see where they can stream, buy or rent the movie.

Reply on news.ycombinator.com