Hacker News Reader

Comment by JanneVee | Hacker News Reader

"The algorithm" is irredeemable. Its sole purpose is to prolong engagement so they can sneak ads in front of your eyeballs. Open sourcing it would not change its purpose and intended design. The design does not take into account that it can distort users worldview to the point of delusion and hinder or even set back progress of society if enough people are deluded in a group. I left twitter for the sole reason of that I saw a huge group of people feeding their worst tribal tendencies and celebrate the death of a human being. This is the very same mechanisms that has fed into pogroms and other atrocities through out history. "The algorithm" presented me with these celebratory tweets as a recommendation in my feed of only following highly technical people and projects with in software.

View on HN

But what if "the algorithm" is turns out to be rather basic? Maybe it just gives people more of what they like. Filter bubbles, polarization and herd mentality could be a natural, inevitable consequence of any Twitter-like UI and a central database.

The algo surely is quite basic. I believe JanneVee is saying that a "rather basic" algorithm that tries to maximize engagement drifts towards promoting the more divisive posts that amplify our tribalistic tendencies.

And by giving more visibility to polarizing content it incentivises others to create more polarising content.

The algorithm therefore is a positive feedback loop of polarization. It doesn't just show what we are, but amplifies our worst (or tribalistic) tendencies.

Relevant Howard Stern reference: https://www.youtube.com/watch?v=9G6xu-J_Dmc

The problem is that I tried to put myself into a software development filter bubble but yet the herd mentality and polarization bled through the filter. It is not a single issue of "algorithm = bad". It is multiple issues intersecting with psychology. It is the subversion of willpower by hijacking the dopamine that you feel rewarded to keep scrolling, it is the fracturing the shared reality by presenting different content to different people depending on what they engaged with and a number of other issues... all this makes it a complicated mess where one set of people starve themselves to death because they think they are fat according to their feed and other where people eat themselves to death because fat is beautiful according to their reality. You can fix one issue but then the next issue props up and all of a sudden the complications of human psychology makes the unintended consequences even more dire. All this to make you watch ads and buy things.

Jaron Lanier’s claim is the feedback mechanism (clicks, engagement) isn’t nuanced enough [1].

i.e. If the algorithm had more insight into a given person’s emotional (and intellectual?) response, the outcomes could be better than the status quo.

[1] https://open.spotify.com/episode/3DBNkrAY5sCIX2sqHpZ567?si=M...

The word here is "could"... It can and has been used for outcomes that hasn't favored the users interests. And that is in my view a worse outcome than status quo.

I definitely hedged yeah.

> The problem is that I tried to put myself into a software development filter bubble but yet the herd mentality and polarization bled through the filter.

But you understand that technology have social consequences? Discussing them is vital part of discussing technology. "Moving fast and breaking things" is fun and all, but when you endangering peoples lives and wellbeing, maybe you should think twice about what you are doing.

Technology is a slave to capital.

Talking about technology as though it has any moral imperative other than "make more money" in the West is naive at best. At worst it is not only boring to anyone over 21, but it's counter productive to improving the world. Our newest toys are there to distract from the woman behind the curtain.

In short: let me read what I want and save your moralizing for the people who actually make decisions. None of which are technologists (any more).

>Technology is a slave to capital.

You mean it's slave to people controling capital? Because capital by itself isn't sentient or capable of making decisions.

>Talking about technology as though it has any moral imperative other than "make more money" in the West is naive at best.

Technology have moral consequences, not a "moral imperative". Technology doesn't care if it counts number of Jews, carrots or cans of Zyklon-b. But people should, and normal people do. Tehnology shouldn't be an excuse to do whatever you want, because "computer said so". Someone programed/tauhgt that computer. Someone is responsible for it.

>At worst it is not only boring to anyone over 21, but it's counter productive to improving the world.

Capital, corporations and billionaires are human inventions, so does kings, slavery, white supremacy, misogyny. Kings didn't stopped to be kings just because everyone asked them nicely. It took time but in the end, kings were no more. Because in the end they depend on ordinary people to do their biddings. Jeff Bezos doesn't check if drivers reached their quota of urin filled bottles, people (directly or indirectly) do this.

>In short: let me read what I want and save your moralizing for the people who actually make decisions. None of which are technologists (any more).

Not talking about problematic technologies, decisions made by people in power doesn't make them go away. Pretending bad things didn't happen is delusional, but just talking doesn't help either. Actions are also needed, even if they are inconvenient or dangerous. There are many ways change can be brought, protests, civil disobedience and more. But pretending everything is ok, will make every call to action fail, even before it started.

> But you understand that technology have social consequences?

Yes that is why I started of this whole thread with: > "The algorithm" is irredeemable.

My personal opinion is formed by tweets that where not about the social or societal consequences of software development but totally unrelated events from people I didn't follow.

Bingo. Filter bubbles are nothing new. Newspapers are lo-fi filter bubbles. Each newspaper occupied a different part of the political spectrum with its own biases. The concept of "following" is also a filter bubble. People will naturally follow other Twitter users with similar views, until the point where most of their feed consists of like-minded tweets.

We are independently blaming every social media platform for what is, fundamentally, a human problem.

Well the problem I'm trying to point out is that yes it is a fundamentally human problem. The problem is that a machine is amplifying this. There is a curation of content that is personal that may lock you harder in the filter bubble than a newspaper trying to cater to a wide audience. To cater to wider audience occasionally the newspaper will contain something that outside of your safe little bubble. It is a bigger chance that you expand and/or evolve your views because of this. Of course the option is to be closed minded, but that is a choice by the reader and not the social media platform which do it for you automatically to keep you using the platform.

You can learn a little about how youtube recommendations and facebook's feed work in this interview I did with someone who worked on both of them.

It's nothing magical. It's just a big system trying to find the things it thinks you are most likely to find engaging.

https://corecursive.com/061-reinforcement-learning/

I know that they aren't magical. But the intent is to present engaging content to you so you stick around longer so you can be datamined and targeted with sponsored content. The issue I take with it is that there is a consequence almost of philosophical dimension of free will. There is societal and social cost of that particular business model that relies on driving engagement to present ads because it lacks the dimensions that are harmful.

There are similar algorithms driving advertisement and one of the ads that has been targeted at me lately is a nicotine source two years after I kicked my nicotine habit. https://en.wikipedia.org/wiki/Snus The intention is clear, to make me pick it up again. Given this is simply a personal anecdote but it is eerily accurate sometimes.

I mean in the interview that idea doubted, the maximizing of ad revenue.

At least with facebook, but I assume with others, they are using engagement as a proxy for delivery value to you. That is the goal they are trying to maximize for.

So they want to maximize how much pleasure you get from the service, so they maximize engagement, but they don't necessarily maximize revenue, because that would hurt engagement.

To quote myself "Its sole purpose is to prolong engagement...", so we are on the same page here. But there is downside to this, it often hijacks your dopamine system and can be addicting. It doesn't only do that, occasionally you get content that you disagree with and interact with because of it. That also gives you pleasure to tell someone that they are stupid. This is why there is a such disproportionally high reach of e.g. flat earther content. People engage with the content to tell them that they are stupid all the time, but sometimes it sucks someone in to their rabbit hole. It is a fairly benign example... so what about anti-vaxxers or QAnon? These people are enjoying themselves to the point that they delusions take hold. And I blame the maximization of engagement.

Yeah! Totally agree. I thought you were saying they were trying to max revenue pee user. My mistake.

You re making a good case about why twitter is overrated as a medium and should not be taken seriously. the vast majority are crazies, who don't matter. The rich men are interested in controlling it because there is a small minority in it, the journalists, who are using it as a measuring stick and as a way to coordinate their stories. While that's true, those journalists are not a particularly sticky crowd and will move elsewhere if some or other rich man thinks he can control them. I think substack is their next campsite, and it could easily launch a microblog timeline.

First, everyone matters.

Second, part of the problem is diving into the delusion by picking a side. One such side is men vs women.

I would argue that almost the entire problem with the internet(including Twitter) is advertising. All the problems stem from it.

I ve posted thousands of comments, they never mattered.

I don't even think twitter issue is "the algorithm", but the UI itself. The issues is not even bubbles, there is nothing wrong with bubbles, we even celebrate healthy bubbles thats bring fun, odd topics, etc. The issue is bubble shiftings to extremes with nothing to moderate it. As the bubble become more and more extreme, and showing their behavior in public, the behavior become "normal" and allows it to become even more extreme.

Getting rid of the "trending" block, with something else, like "trending tag within people you follow" would get rid of this normalisation of extreme behavior.

Then you need moderation.

Seeking engagement is not necessarly evil.

GitHub increased engagement, and it lead to more cooperation in the open source software, that's a plus for everyone.

The issue of a lot of social networks, is they want engagement, even if it cause addiction.

I think twitter social network "model" is salvagable in something healty, but it needs a lot of changes.

I use the Minimal Twitter extension to remove everything. I have a chronological twitter feet with no ads and no suggestions. I go in maybe 2-3 times a day for 5-10 minutes, scroll down and like/respond as desired, then leave once I recognize the tweets from last time (or I get bored).

I'd be happy to pay $1 or $2/mo for this directly from Twitter, but until that's an option I don't feel remotely bad removing my sliver of ad revenue from their bottom line.

The UI is like that because of the above commenter's point - the business model itself. It's not that Twitter isn't technically capable of tweaking the UI or giving you more settings to customize your experience, it's that giving you those settings would mean you use them against their own business model.

> Seeking engagement is not necessarly evil. GitHub increased engagement [...]

At this point it really depends on what you mean by "engagement". GitHub's definition of "engagement" is definitely not the same as Twitter's.

Technically, Twitter's goal is not "engagement" but "time spent looking at ads". It just so happens that "engagement" is a very good proxy for that. In Twitter's case, increasing "engagement" is definitely evil.

I also don't believe GitHub's goal is to increase engagement. Their goal is ultimately to deliver features useful to their paying customers - companies that build software, where as Twitter's (and any other ad-supported service's) goal is to serve more ad impressions (or at least make advertisers believe they do).

Twitter, and GitHub goal is not "time spent looking at ads" or "deliver feature" but "make money". I bet/hope that if Musk succeed in taking over twitter, he will implement a patreon style subscription, there is a lot of money there thats twitter leave on the table.

> he will implement a patreon style subscription, there is a lot of money there thats twitter leave on the table

I don't think that's any better than the current scheme; infact, it could be worse becasue the current ad system has centralized control which can check on extremist echo-chambers. I suspect Patreon-style funding will increase polarization[1], with the "content-creators" optimizing for what brings in the most money, not necessarily their thoughtful ideas.

1. With a side of increased brigading, coordinated harassment,doxxing and swatting; but that'll be the price of "free" speech.

Current ad system try to sell weapon related merchs to far right extremist, etc, I don't believe in your "centralized control".

I’ve seen the work that goes into spotting & mitigating TOS-violating ads: you don’t have to believe in it for it to exist.

Reply on news.ycombinator.com