
Elisheva:
If the Recommended for You titles are an indication of what we'd get from a similar list of users, mine would be along the lines of: here's someone who watched a Thai title in 2018 and hasn't been back on the site since, an assortment of random BL stans and Chinese het rom com fans, and a few dozen who follow me assuming I'm into kdramas because isn't everyone?
So IDK it might work for some, and maybe it should be done for them, but I'd probably be torn between wanting to opt out and hoping it might find me the only other person on the site who really likes Thai ghost movies which turn out to be about grief.
If there's no opt out though, no. The site already spams me enough things I have increasingly less than zero interest in.
That's fair, user matching is an extremely complex algorithm that models haven't got right yet - since it's a slice of our varied, complex, and usually implicit emotional motivations . For example, we know 3 people can enjoy the same content for completely different reasons - as a die hard fan of a cast member, for the story, as a good lunch time companion).
This gets slightly easier with book recommendations - perhaps because textual themes are easier to capture? Not sure. And that's where my inspiration comes from. I use Storygraph & love discovering books to read based on a reader who's as weird as me in reading preferences. Feels like a kindred soul, likely sitting worlds apart.
I found myself craving a similar vibe on MDL. The recommendations (personalized or similar-to-this-drama) are quite off the mark on 'what to watch next' and, other than my years of acquired judgement, word of mouth or boredom end up as my usual way of finding new content.
I second the necessity of an opt-in option (better to get users to explicitly turn such features on, rather than add it sneakily like so many social media products). (fyi: storygraph offers such a privacy setting too.)
At this point this is a pipe dream in finding kindred souls that share my combination of content tastes. I just threw the idea here in case MDL has a data science team up for the challenge.