TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Twitter's Algorithm: Amplifying Anger, Animosity, and Affective Polarization

106 pointsby abhi9ualmost 2 years ago

13 comments

asimalmost 2 years ago
I hate social media with a passion. In 2000 using msn, yahoo messenger and chat rooms was a joy. Browsing the web was such fun, exploring and finding new information and websites.<p>In 2005 MySpace became all the rage and something shifted. It was the start of the &quot;me&quot; generation. Facebook amplified this and it became all the rage to ask &quot;are you on Facebook?&quot; at parties. Everything was new and exciting but we also had no idea what we were about to get into. The feed became an addictive creation that twitter then took to some next level. The &quot;For you&quot; tab is a joke. It&#x27;s a way to continue to drive clicks, likes, retweets. All of this needs to end now. I honestly don&#x27;t have the answer but I know part of it is probably less technology, less staring at screens and almost zero social media consumption. I so badly want to escape.
评论 #36122221 未加载
评论 #36122254 未加载
评论 #36122555 未加载
评论 #36123356 未加载
评论 #36122234 未加载
评论 #36125592 未加载
评论 #36123249 未加载
评论 #36124930 未加载
anscalmost 2 years ago
I really wish there was a... &quot;casual&quot; or &quot;slower&quot; replacement for Twitter. I created a Twitter account and followed my favorites, unfortunately I keep getting &quot;outrage&quot; tweets recommended. Funny, it&#x27;s like it&#x27;s own little microsphere where people vent about trans, climate, &quot;men are superior to women&quot;, all these 4chan-esque topics. It&#x27;s not even about about being correct, it&#x27;s just about short &quot;gotcha!&quot; posts that sound good but are riddled with logical fallacies and ludicrous points if you think about it for a second.<p>I, on the other hand, go to HN to complain apparently. :-)
评论 #36122366 未加载
评论 #36122441 未加载
评论 #36122446 未加载
评论 #36122368 未加载
评论 #36122607 未加载
评论 #36127326 未加载
评论 #36122443 未加载
评论 #36122359 未加载
trabant00almost 2 years ago
Enough outrage against the outrage. If any other content would have worked then media and social media would use it. The algorithms reflect what people spend time on. My recommendations are motorcycles, computers, video games, watches, etc. Still a useless time-sink comprised of mostly thinly veiled advertising but nothing rage inducing.<p>Nobody else is responsible you click on political stuff. Nobody is responsible you join every moral crusade that gets started. Even on HN where there is no personalized algorithm, even on technical issues, most people vote and comment with their moral&#x2F;political sensibilities a lot of the time.<p>So in conclusion it&#x27;s not them, it you.
评论 #36122512 未加载
yodelshadyalmost 2 years ago
Even without the Algorithm, Twitter would be awful.<p>Quite simply: when was the last time you went in Twitter to see an intelligent argument <i>that opposes yours</i>? Never, because such a thing simply requires more than 280 characters, and <i>multiple</i> cited sources.<p>280 chars with a single source <i>is</i> enough for a &quot;look how bad my enemies are&quot; hot take, however.<p>It&#x27;s a shame, because getting off social media fully is <i>hard</i> now - so much actual research and professional interaction requires it as well. But eurgh.
评论 #36122667 未加载
评论 #36123194 未加载
评论 #36126805 未加载
评论 #36127831 未加载
ggmalmost 2 years ago
Just quit. I did, never regretted it. It&#x27;s toxic shit, owned by a toxic shit.
zhte415almost 2 years ago
&gt; Encouragingly, the recently passed EU Digital Services Act (2022) mandates that large online platforms offer a non-algorithmic way of viewing content, such as a reverse-chronological feed [Vincent, 2022]. This development may soon make it possible to replicate our methodology across multiple platforms of interest. We hope our study will inspire future efforts in this direction.
seanhunteralmost 2 years ago
If you read the &quot;Ranking by Engagement&quot; blog post[1] you can understand why.<p>Ranking by expected user engagment increases user retention but is negatively related to quality.<p>[1] <a href="https:&#x2F;&#x2F;integrityinstitute.org&#x2F;blog&#x2F;ranking-by-engagement" rel="nofollow">https:&#x2F;&#x2F;integrityinstitute.org&#x2F;blog&#x2F;ranking-by-engagement</a>
x-complexityalmost 2 years ago
Breaking away from the anti-social-media consensus that&#x27;s bubbled up here, I haven&#x27;t seen any discussion about how much the baser human psyche craves tribalism &amp; self-grouping more than anything else.<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;pdf&#x2F;2305.16941.pdf" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;pdf&#x2F;2305.16941.pdf</a> (page 5)<p>As shown in the paper itself, regardless of the political tribe, there&#x27;s a persistent positive feeling that comes with reinforcing the viewpoints of the in-group.<p>(page 4) It should also be noted that the personalized tweets are self-reported to be rated slightly higher than the chronological tweets, with political tweets being the exception. In short, when politics is filtered out, the Algorithm is doing its job well.<p>&gt; 1.3 Tweet value to reader<p>&gt; For each tweet, we also have users assess whether they value seeing that tweet or not (“When you use Twitter, do you want to be shown tweets like [@author-handle]’s?”, where users choose between “Yes”, “No”, and “Indifferent”). We find that overall, tweets shown by the algorithm are rated slightly higher (0.05 SD, p = 0.024). Interestingly, however, when restricting to political tweets, those recommended by the personalized algorithm lead to significantly lower user value than those in the chronological timeline (−0.14 SD, p = 0.004).<p>Dataset-wise, it&#x27;s heavily skewed towards left-&amp;-far-left Democrats (page 12, Tables 5 &amp; 6), &amp; young people (page 13, Table 4).
abhi9ualmost 2 years ago
A thread by one of the authors of the paper, giving an overview: <a href="https:&#x2F;&#x2F;twitter.com&#x2F;SmithaMilli&#x2F;status&#x2F;1663222170560999424" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;SmithaMilli&#x2F;status&#x2F;1663222170560999424</a>
SilverBirchalmost 2 years ago
This is pretty unsurprising for anyone who has used twitter. The entire way it&#x27;s set up is that you have clusters of mutual followers who form communities, and they engage in fairly innocuous ways, and then because the site is public-by-default you get inter-group conversations where someone from one of these communities will see something posted by one of the other communities and at this point the two communities go to war. It&#x27;s like if 1 in every 100 posts on Reddit appeared on a different sub-reddit. Think of the engagement when a &#x2F;r&#x2F;conservative post just pops up on &#x2F;r&#x2F;politics. Reddit has obviously seen what happens there and has made big moves to stop that happening, but at twitter that&#x27;s the <i>default</i> way the site works. It&#x27;s fantastic for engagement and no one at twitter wanted to kill the golden goose.<p>Having said that, this study is quite unfortunate in its timing - it managed to time it&#x27;s data collection at the exact point in time where Elon Musk was forcing his employees to manually boost his tweets so that he could get more views than Biden [1]. So basically while the twitter algo had been pretty stable for the last decade they managed to time their data collection over the exact week where the moron decided to start messing with it to stroke his own ego.<p>[1]:<a href="https:&#x2F;&#x2F;proofpointisolation.com&#x2F;browser?url=https%3A%2F%2Fwww.theverge.com%2F2023%2F2%2F14%2F23600358%2Felon-musk-tweets-algorithm-changes-twitter" rel="nofollow">https:&#x2F;&#x2F;proofpointisolation.com&#x2F;browser?url=https%3A%2F%2Fww...</a>
评论 #36123241 未加载
bigcat12345678almost 2 years ago
Essentially, &quot;Amplifying Anger, Animosity, and Affective Polarization&quot; is to give people more &quot;free expression&quot;, right?
neomalmost 2 years ago
Elon said in a recently interview that his new most important metric is to reduce regret time spent on site. An example he gave is that people should get push notifications about tweets they wouldn&#x27;t regret clicking.<p>I don&#x27;t know how twitter has me profiled, but these days it sends me onlyfans models, Matt Wallace, J.K. Rowling, Ed Krassenstein, Bitcoin influencers as recommended tweets to read.<p>Funny thing is, I don&#x27;t follow anyone and nobody follows me, I just have a twitter account to occasionally (once a week?) read what Elon Musk, Jeff Bezos, Bill Gates and some musicians I like are tweeting about. I&#x27;m a liberal leaning centrist, I don&#x27;t care about American politics much, or bitcoin, or onlyfans... I&#x27;m so curious how it decided to start serving me all this content that is so outside of who I am.<p>Also for those curious, this research is out of Berkeley and Cornell:<p>Smitha Milli, Postdoctoral Associate, Cornell Tech<p>Micah Carroll, Ph.D. Candidate, University of California, Berkeley<p>Sashrika Pandey, Undergraduate Researcher, University of California, Berkeley<p>Yike Wang, Undergraduate Researcher, University of California, Berkeley<p>Anca Dragan, Associate Professor, University of California, Berkeley
评论 #36122513 未加载
评论 #36122412 未加载
评论 #36122780 未加载
评论 #36124685 未加载
评论 #36122694 未加载
评论 #36122521 未加载
评论 #36126362 未加载
twittsalmost 2 years ago
All i came across Twitter is trash. It doesn&#x27;t even load properly on my phone. I would rate it as a piece of shit.