Back
Eli Pariser: The Filter Bubble (2011) 4 stars

"An eye-opening account of how the hidden rise of personalization on the Internet is controlling-and …

...at this point, Netflix can predict how much you'll like a given movie within about half a star.

— The Filter Bubble by 

I'm curious where this conclusion comes from? The wording makes the claim a little nonsensical, but it sounds like it's based on actually metrics about how people (in 2011) responded to netflix's recs? But there must be an unstated assumption here about how much prior use the user has on which the recommendation is made.

I also wonder how well this holds up today, when streaming video is so fragmented. Netflix has basically none of the movies I like on it, so it does a very poor job if recommending them to me.

replied to mouse's status

I guess I also wish, more generally, that we were more explicit when talking about the efficacy of recommendation algorithms, that they inherently can only recommend to you within their own content, and contextualized what a successful the recommendation means, specifically.