181–190 of 317 posts
I've said it before and I'll say it again, FB users are a kind of digital peasant or serf. To me it feels like we're seeing the genesis of the divide between Morlocks and Eloi.
https://www.theatlantic.com/business/archive/2017/07/connect...
Facebook also has a serious problem in that its news feed is a content recommendation engine with only positive reinforcement but no negative reinforcement. So you end up with a ton of false positives even when actively interacting with the content, and their system doesn't even know how wrong it is. And should you really not like some content, the solution is unfriending the poster, rather than simply matching again…
...note that I can choose to dislike the post itself, or hide the entire source or app if I prefer.
Earlier quoted context omitted.
> If I get notices from pages/groups I dont care about anymore, I either unfollow or mute the next time their post annoys me. I do that as well, but I think it's extremely rare among Facebook users. Everyone else I know either: * Puts up grudgingly with "spam" they no longer want to see, or eventually * Complains "Facebook is full of crap these days, I'm deleting my account". A filter algorithm is a much better exper…
What I want to read is stuff my friends have taken the trouble to type in, all of it whatever the topic. I don't want to read whatever random crap on the web they are reading or what ever. Original user generated content made Facebook and the inability to find my friends' generated content is what makes it not worth reading. I have about 1B ways to get hot memes and trending internet topics. I have relatively few way…
Current version od algorithm is not working.
Similarly with Amazon, products should have some sort of 'elasticity' score where it should learn that recommendations of inelastic products is a waste of screen real-estate. I mean, I doubt the model is giving a high % to most of those recommends - it's likely more a business/UX issue in that they've decided it's worth showing you low-probability recommends instead of a cleaner page (or something more useful).
Youtube, on the hand, seems to be precision tuned to get you to watch easy to digest crap. You consume the crap voraciously but are generally left unfulfilled. This is a more difficult problem where you're rewarding naive views rather than a more difficult to discern 'intrinsic value' metric. As a 'long term' business play the model should probably weight more intellectually challenging content just like fast food restaurants should probably figure out how to sell healthier food because by pedaling crap you're only meeting consumer's immediate needs, not their long term ones.
Earlier quoted context omitted.
The missing detail here of course is...maybe his posts aren't getting engagement outside of mom because his Facebook 'friends' aren't interested in them?
Well, that would seem to be contradicted by the observation that his friends do engage with them if he hides the posts from his mom.
He should try alternating the privacy settings with each post.
It's simple really, just stop participating in that evil experiment. From the outside, you look like morons; talking to opposite sides of an algorithm while interpreting what comes out as reality. It's been proven over and over again that consuming that crap makes everyone feel bad and hate each other. There are plenty of alternatives, but this one is mine: https://github.com/andreas-gone-wild/snackis
Earlier quoted context omitted.
I've heard the argument that this prevents buyer's remorse. Not sure if it has any merit.
wait, wouldn't this cause buyer's remorse? like, if I just finally decided to buy some product ... being bombarded with all the other options suddenly makes me rethink my purchase decision, "oh but what if this other one is more gooder!?!?!"
Earlier quoted context omitted.
YouTube would rather you fall into a trance of continuing your recent viewing habits rather than providing a personalized library. It just makes better business sense.
Why does it make better business sense?
I feel like these are just shitty models. A good recommendation model would get features like "is_mom" and learn that "is_mom" is a shitty predictor of relevance. Similarly with Amazon, products should have some sort of 'elasticity' score where it should learn that recommendations of inelastic products is a waste of screen real-estate. I mean, I doubt the model is giving a high % to most of those recommends - it's li…
Earlier quoted context omitted.
I did this thinking that I don't want to share semi-private pictures of family with strangers, or to spam my family and friends with programming stuff. Do you know there are friend groups you can use to eliminate that problem? "Family and school," "Programmery Peeps," if you post using either of those groups for visibility then it seems like the issue is moot.
Most people don't use them. It is too hard (10 extra seconds is too hard). Facebook is good at figuring out from the pattern of likes who else will be interested in seeing something. It is only when you have an important announcement that they screw up.