Twitter's Recommendation Algorithm • Hacker News Board

Re: Twitter's Recommendation Algorithm

#851

Post by bluesign » Sat, Apr 01, 2023, 10:16 AM UTC

dmonitor wrote: post #285

phailhaus wrote: post #163

Great! But nothing is going to change until people realize that the problem is the feedback loop . It's not the recommendation engine itself, it's the fact that there's no way "out" of the feed that the engine produces. It recommends you stuff, you have little choice but to engage with it, and then it trains on that information . This is the problem with most of social media today. It is a very well known problem in…

reminds me of a story about a guy who was given a gift, a decorative plate with a rooster on it i think it was. didn’t care for it too much, but out of politeness put it on display on an empty cabinet he had. a while later someone noticed he had it and figured he liked it, so got him a similar decorative plate with a rooster on it. again, out of politeness, he put it next to the old one. now other people started to t…

and in the end, he starts to like that decorative plates with roosters or at least thinks rooster on plate is a nice decoration if it is produced that much.

Re: Twitter's Recommendation Algorithm

#852

Post by kilianinbox » Sat, Apr 01, 2023, 10:18 AM UTC

Summary this far • Code from Twitter's algorithm GitHub repository shared • Algorithm checks for specific author types (e.g., Elon Musk, power users, Democrats, Republicans) • Author ID lists used for metrics collection in A/B experimentation platform • Metrics tracked in A/B tests to avoid negative impacts on specific groups • VIPs like Musk, LeBron James, AOC used as indicators for algorithm's behavior • Algorithm changes that negatively affect Musk unlikely to go live • Speculation about code changes pre- and post-Elon's purchase of Twitter • Discussion on the importance of measuring and testing for potential biases • Debate on moral decisions in the context of Twitter's algorithm and content moderation

Re: Twitter's Recommendation Algorithm

#853

Post by oneeyedpigeon » Sat, Apr 01, 2023, 10:22 AM UTC

misiti3780 wrote: post #309

cubefox wrote: post #155

Earlier quoted context omitted.

This is a very cynical take. They should be commended for publishing recommendation code at all, which no other major social network does.

[flagged]

Because the way he acts gives people every right to. I agree that he may be misrepresented, but if he is, then he has to shoulder at least some of the blame.

Re: Twitter's Recommendation Algorithm

#854

Post by helsinkiandrew » Sat, Apr 01, 2023, 10:28 AM UTC

ivalm wrote: post #837

helsinkiandrew wrote: post #825

Earlier quoted context omitted.

> But the underlying policies and models are almost entirely missing... Without those, we can't evaluate the behavior and possible effects of "the algorithm And neither can spammers find and test the cracks and edge cases that would allow them to break the system, that does sound reasonable to me. If they were public there would be an arms race between spammers/those wishing to game the system and Twitter engineers.

Then don’t pretend to release “the algorithm.”

They’re explaining how it works without giving the specifics. Much like the US military explains how the nuclear deterrent works without disclosing detailed plans and control codes.

Re: Twitter's Recommendation Algorithm

#855

Post by oneeyedpigeon » Sat, Apr 01, 2023, 10:33 AM UTC

cdash wrote: post #753

anigbrowl wrote: post #127

Earlier quoted context omitted.

Inaccurate. Musk stated that people you follow will continue to show up. I only use the 'for you' feed when I'm bored and want stupid dopamine hits, I leave it on Following almost all the time. But that's on desktop, my understanding it that it keeps resetting itself for mobile users (of whom I am not one).

I have had it reset a couple times on mobile a few months ago but at least recently I have not noticed it happening anymore.

Just pressing the home icon twice takes it back to For You.

Re: Twitter's Recommendation Algorithm

#856

Post by oneeyedpigeon » Sat, Apr 01, 2023, 10:43 AM UTC

woodduck wrote: post #742

jillesvangurp wrote: post #711

The irony is that I prefer Mastodon's sort by time and don't try to be clever approach to this expensive and futile attempt to feed me an endless stream of click bait. I objectively spend more time on Mastodon than on Twitter at this point. It's more engaging for me. It's how Twitter used to work when it was still nice to use. If Twitter wants to put a stop to the user exodus and save lots of money in the process, he…

They did add an off switch to the for you feed. There's a tab for following, albeit not sorted by time.

But it's not sticky and it's easy to miss when it forces you back to For You. Less of a problem on desktop because you can write a trivial extension to force Following, but not an option on mobile, unfortunately.

Following IS sorted by time, though.

Re: Twitter's Recommendation Algorithm

#857

Post by tinyhouse » Sat, Apr 01, 2023, 10:46 AM UTC

GaryNumanVevo wrote: post #32

tric wrote: post #11

From https://github.com/twitter/the-algorithm/blob/7f90d0ca342b92... ( "author_is_elon", candidate => candidate .getOrElse(AuthorIdFeature, None).contains(candidate.getOrElse(DDGStatsElonFeature, 0L))), ( "author_is_power_user", candidate => candidate .getOrElse(AuthorIdFeature, None) .exists(candidate.getOrElse(DDGStatsVitsFeature, Set.empty[Long]).contains)), ( "author_is_democrat", candidate => candidate .getOrEls…

I wonder who's on the "VIT" (Very Important Tweeter) list?

VITs are verified accounts.

Re: Twitter's Recommendation Algorithm

#858

Post by oneeyedpigeon » Sat, Apr 01, 2023, 10:46 AM UTC

mlindner wrote: post #789

qbasic_forever wrote: post #532

Earlier quoted context omitted.

Yes that weekend everyone started seeing Elon's tweets in their timelines: https://www.theverge.com/2023/2/14/23600358/elon-musk-tweets... You seem weirdly protective of Elon fyi

And Elon later explained that that had nothing to do from him making any specific request that his tweets be valued. This is largely a false creation without evidence to back it up. People are assuming intention behind something that's much more easily explained as a bug, one that was quickly fixed. If it wasn't a bug, why would it disappear if people assume Elon is narcissistic as people claim? It'd still be priorit…

It's an incredible coincidence that, out of hundreds of millions of accounts, this bug happened to elevate the account of the person who owns twitter. I'm not saying it was 100% intentional, but it does suggest Musk's account is 'special' in some way.

Re: Twitter's Recommendation Algorithm

#859

Post by concordDance » Sat, Apr 01, 2023, 10:49 AM UTC

philote wrote: post #434

cubefox wrote: post #155

Earlier quoted context omitted.

This is a very cynical take. They should be commended for publishing recommendation code at all, which no other major social network does.

Hey, we can get even more cynical. Why should we trust that this code is even similar to what they run in production currently?

I can't imagine deliberately special casing Elon's account in something they made from scratch to fool people.

Re: Twitter's Recommendation Algorithm

#860

Post by oneeyedpigeon » Sat, Apr 01, 2023, 10:49 AM UTC

slim wrote: post #810

qbasic_forever wrote: post #267

Earlier quoted context omitted.

IIRC it was very recent, there was a Twitter engineer that was fired after explaining to Elon that the algorithm was not biased against him: https://www.salon.com/2023/02/10/petulant-elon-musk-fired-tw... Almost certainly after that event Elon had them explicitly bump his tweets in their reach.

Musk expressed his disappointment with engineers' work and *told them to track* how many times his tweets get recommended, one worker told the outlet. so we know exactly when and why "author_is_elon" was added to the code

That's one possible explanation. If you were implementing that change, would you do it by embedding an "elon" variable in the codebase, or would you make the specific account a parameter?