5 ms·

It's going to be paradise when we can find our natural best friends AT SCALE Here's my second most recent idea on how to do it, haven't written up the latest y

It's going to be paradise when we can find our natural best friends AT SCALE Here's my second most recent idea on how to do it, haven't written up the latest yet: http://zeroprecedent.com/A%E2%9D%AF%E2%9D%AEON%20deck%20v.3.pdf http://zeroprecedent.com/A%E2%9D%AF%E2%9D%AEON%20deck%20v.3....

npilk 4y ago
Interesting concept but what seems more compelling to me would be more like a Tiktok/Reddit hybrid. You log in and see content that "the algorithm" knows you'll like, but with an added element of community from people "the algorithm" knows you'll find cool/funny/insightful/etc. Like a simulacrum of the Facebook feed but instead of friends and family it's this "tribe". No need for labels or even naming those people. Feels like this is what social media is trying to achieve anyway but they're mostly stuck with the existing social graph as a starting point...
- EZ-Cheeze 4y ago
  I actually mention extremely good content aggregation in there, but more as a byproduct of good matching than as a way to arrive at it. But it really could work the other way around, with the things you and everyone else likes providing strong signal that can be correlated. I hope this is the year we all get a thousand true friends! Instant connection every time
telehaxor 4y ago
"Machine learning will help you find your best friends." Yes, but how? How will it avoid optimizing on the wrong thing? "It will gather a lot of data and then use algorithms to predict your affinities from it." Yes, but how? What algorithm would do that? Ordinary recommender systems, neural networks, something else? And tight knit communities of close friends rarely have a lot of churn, but at-scale systems often do - how will it reconcile these tendencies? I guess I was looking for something a little more like a research paper.
- EZ-Cheeze 4y ago
  Uh yeah I'm a filthy non-technical. But it's obviously possible, right? Current social networks are like those bikes with a huge front tire. Someone make the tires equal already

npilk 4y ago

Interesting concept but what seems more compelling to me would be more like a Tiktok/Reddit hybrid. You log in and see content that "the algorithm" knows you'll like, but with an added element of community from people "the algorithm" knows you'll find cool/funny/insightful/etc. Like a simulacrum of the Facebook feed but instead of friends and family it's this "tribe". No need for labels or even naming those people. Feels like this is what social media is trying to achieve anyway but they're mostly stuck with the existing social graph as a starting point...

EZ-Cheeze 4y ago
I actually mention extremely good content aggregation in there, but more as a byproduct of good matching than as a way to arrive at it. But it really could work the other way around, with the things you and everyone else likes providing strong signal that can be correlated. I hope this is the year we all get a thousand true friends! Instant connection every time

telehaxor 4y ago

"Machine learning will help you find your best friends." Yes, but how? How will it avoid optimizing on the wrong thing? "It will gather a lot of data and then use algorithms to predict your affinities from it." Yes, but how? What algorithm would do that? Ordinary recommender systems, neural networks, something else? And tight knit communities of close friends rarely have a lot of churn, but at-scale systems often do - how will it reconcile these tendencies? I guess I was looking for something a little more like a research paper.

EZ-Cheeze 4y ago
Uh yeah I'm a filthy non-technical. But it's obviously possible, right? Current social networks are like those bikes with a huge front tire. Someone make the tires equal already