Hacker News Reader

Comment by moonchrome | Hacker News Reader

All this talk about "algorithmic feedback loops" is overselling the quality of said algos (eg. I get more interesting suggestions when I open YT incognito then I do when I'm signed in, I have to scroll through a bunch of stuff I have 0 interest on FB when I open it, etc.) and their importance.

It's completely missing the actual attention grabbing factor here - there's a huge amount of content for consumption on demand in those networks. Unlike TV or radio where at most you could have 100s of channels simultaneously available, there's billions of articles/videos/posts/songs/articles/etc. available with a few clicks.

At best algos get you there faster.

View on HN

I am not sure. This is just a data point of 1.

I used to browse Instagram casually. Like, once is a blue moon. I would see updates from my network and spend a few mins here and there. This was probably like over year or two back. (not 100% sure)

And then a few months back I just could not take my eyes off of Instagram reels. This feature wasn't there before I think. I was just completely hooked. It was pretty interesting. I had to delete my instagram off of my phone.

They definitely are figuring out the formula to retain people's attention.

Again, this is just a data point of one.

>And then a few months back I just could not take my eyes off of Instagram reels. This feature wasn't there before I think. I was just completely hooked. It was pretty interesting. I had to delete my instagram off of my phone.

Sounds like they started showing you different kind of content. I never got into Instagram because I'm just not into the kind of content that gets published there. Not arguing that content presentation isn't a factor here, but I would argue it's the content that's addicting and even if you remove the suggestion algorithms and just did social sharing/recommendations you'd still end up with similar results - viral stuff got shared in chats/groups/mail chains/etc. now it gets served up to your feed directly.

Not saying suggestion algorithms don't improve the experience if they work, but I wouldn't really attribute that much value to them, ultimately we have an industry of influencers trying to create content that will grab your attention.

I go on Facebook about once a month. I’ve noticed that Facebook’s friend recommendation engine now sprinkles in some beautiful women, with whom I have no friends in common. I don’t know if someone tuned the algorithm this way, or if it tuned itself. It feels like the desperate move of a platform that realizes it’s losing control of someone.

Did you see that movie about Facebook? According to that movie, what you described is literally what they do.

Perhaps you just got lucky. Mine suggests tons of people from India and the middle east who I couldn't possibly know. I also check it very rarely. Presumably the algorithm must be picking the (very-slightly) best from many many low-probability guesses.

I had the same experience - they’re incredibly addictive, far more than your standard Instagram.

I think this media is different because of the short duration of each item and the endless novelty. You say to yourself you you will just watch another YouTube video - it’s just a few minutes long and before you know it you’ve wasted half an hour. Where as with a TV programme you have to commit to 30-60 minutes.

Inconsistent reward is more addictive than consistent reward. This has been known for a long time. That's how they make it addictive.

It has the opposite effect on me, I barely open facebook anymore because of how irrelevant it is and I skip the landing page on YT and just search what I'm interested in.

I'm not sure how that's relevant. I'm sure it's a giant Bell curve

This doesn't require algorithmic feedback loops. People simply would seek out (and create) content which provides inconsistent rewards. A slot machine simply needs a random number generator with a certain small chance of jackpot. The null model here is what can be achieved without individualization.

That doesn't mean it can't be optimized with algorithms. Obviously that's what they think they're measuring and doing, otherwise it would be random.

I'm not sure "randomness" is the general property, but rather unexpectedness. The content-provider may well be following a "formula" like with fiction plots, that give the protagonist setbacks and victories. Facebook uses approaches which try to profit within the constraints of a relatively small group of programmers trying to make money off billions of users. It doesn't mean these approaches are the best overall at serving the most desirable content to any given niche of users. Or even the best at making money qwithin those constraints. If are to believe in "network effects" (the other popular basis for calling for a breakup of Facebook) Facebook is simply succeeding because it is there first. Rather independent of this claim that they are winning by being the most competant.

Dono what's with your strawmen arguments here or what your point is besides being argumentative, but I didn't say anything about best. I'm gonna just assume they all see some value in it and you don't have any inside knowledge.

The point of the thread is that these algorithms are not a special new threat, but it's the availability of content itself, targetted or not.

Since we're analyzing each other's psychology now, I do think that insecure and defensive behavior like your post is a real problem with social media. I am not trying to bully or hurt you personally, but simply share some thoughts on a topic.

"All this talk about "algorithmic feedback loops" is overselling the quality of said algos"

So it's you agree that the villains are evil, but your defence for them is that they are inept?

So they will destroy attention even further once the bugs in the algos are tuned out?

Have you tried TikTok? In my experience their algorithm is phenomenal, strikinga good balance between delivering extremely relevant content and trying new content you might like

Reply on news.ycombinator.com