Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Twitter's Recommendation Algorithm

blog.twitter.com

Post reply on HN
Print view View original

661–670 of 1001 posts

Re: Twitter's Recommendation Algorithm

#661

Post by hn2017 »

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…

Per Zoe: The code that was released today doesn't show the parts that actually alter the scores of Elon and other users. The part of the code referenced below just tracks Elon stats (from what I know). Employees removed most PII before the code was released.

https://twitter.com/ZoeSchiffer/status/1641902570921943044?s...

Re: Twitter's Recommendation Algorithm

#662

Post by hot_gril »

gspencley wrote: post #406
koolba wrote: post #90

It's reassuring to know that billion dollar tech companies write CI exactly like I do: https://github.com/twitter/the-algorithm/blob/main/ci/ci.sh Permalink: https://github.com/twitter/the-algorithm/blob/7f90d0ca342b92...

It has been my personal experience, over 25 years in the industry, that often times the bigger the company the worse the code. It's not an absolute rule, I've certainly inherited projects in a consulting capacity that were written by small teams and were atrocious. But more often than not, a small team working for a small company has fewer of the internal "forces" that incur "technical debt." Those forces are things…

Yep. I work for Big Tech, and our CI was lacking for a very long time. Integration testing isn't prioritized at all. Ironic to write code in an overly pedantic language with type safety, spend 3X that time on unit tests with near-100% coverage and what-if analysis, and still have the thing just not work in the end.

For my own projects, I went TDD with integration tests first, and it paid off in both the short and long term. Soon after activated CI (thanks Heroku Pipelines). Actual functionality was more complex than what my day job project does, and my way of doing things wouldn't have been allowed there, yet it was more stable in the end. Amazing what you can do when nobody is telling you what to do.

Re: Twitter's Recommendation Algorithm

#663

Post by suddenclarity »

jeffbee wrote: post #92

Why does anyone use "for you"?

To see what people talk about in your friends circles. It can be interesting in moderation. Similar to skimming the frontpage of HN or recommended list on YouTube. Especially during major news event when your friends might not be the ones posting about it.

Re: Twitter's Recommendation Algorithm

#664

Post by hn2017 »

CathalMullan wrote: post #55
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…

Only used for metrics, apparently. [0] /** * These author ID lists are used purely for metrics collection. We track how often we are * serving Tweets from these authors and how often their tweets are being impressed by users. * This helps us validate in our A/B experimentation platform that we do not ship changes * that negatively impacts one group over others. */ [0]: https://github.com/twitter/the-algorithm/blob/7f…

Per original source, The code that was released today doesn't show the parts that actually alter the scores of Elon and other users. The part of the code referenced below just tracks Elon stats (from what I know). Employees removed most PII before the code was released.

Re: Twitter's Recommendation Algorithm

#666

Post by xmprt »

infogulch wrote: post #657
CathalMullan wrote: post #55

Earlier quoted context omitted.

Only used for metrics, apparently. [0] /** * These author ID lists are used purely for metrics collection. We track how often we are * serving Tweets from these authors and how often their tweets are being impressed by users. * This helps us validate in our A/B experimentation platform that we do not ship changes * that negatively impacts one group over others. */ [0]: https://github.com/twitter/the-algorithm/blob/7f…

So many unnecessarily cynical takes here. Let's say you were in charge of a large legacy system that some segment of customers complain about it not working for them as well as other segments. How would you know whether their complaints are valid unless you measured it? You have to know first. So measure it.

I understand the value of measurements but how does measuring tweets from an individual user help?

Re: Twitter's Recommendation Algorithm

#667

Post by corbulo »

joshuamorton wrote: post #656
corbulo wrote: post #638

Earlier quoted context omitted.

The degree to which it is selective has yet to be determined. Are you claiming total ignorance is superior to partial revelation? I think we would all do ourselves better to go live on a desert island and abandon everything about modern life. A shovel might be useful to bury our heads while we're there.

> Are you claiming total ignorance is superior to partial revelation? I am claiming that this is at least sometimes true, yes. Not always, but sometimes . You're the one claiming that partial revelation is always , without exception, superior to total ignorance. That seems unlikely. Propoganda is often partial revelation, are you saying it is always better to receive only propoganda than to receive no information at…

I think there is objective value to understanding propaganda's origins and goals. Those who do not study history are doomed to repeat it. Those who do not understand propaganda are highly likely to be controlled by it.

Propaganda can be a pretty vague term by the way. Can you describe how the public coming to a better understanding of the inner workings of the worlds most influential social media site is merely propaganda?

Re: Twitter's Recommendation Algorithm

#668

Post by takeda »

GuB-42 wrote: post #648
evolve2k wrote: post #639

Earlier quoted context omitted.

Elons reply “I only learned about it now!”. What a crock of shit. We literally went through reporting a few months back where he was clearly instructing the team to make sure his tweets always come up, for everyone.

I think it was more narcissistic than that. As in: "My tweets are very important, if they don't show up on top, it means the algorithm can't recognize what is important, it needs to be fixed". And the team, who probably didn't see in which way Elon Musk's tweets could be that important besides the fact that he wrote them, they just decided to give Elon Musk's tweets a boost.

The most hilarious things to me was when I was reading some (unrelated to this drama tweet) posted in an article and the top recommendation was Musk's tweet that he doesn't artificially promote his tweets.

Re: Twitter's Recommendation Algorithm

#669

Post by gregcoombe »

iaseiadit wrote: post #659
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…

If you buy a company for $44B and take it private, I for one say you should get your own flag.

"It's one flag, Michael. How much could it cost, $44B?"

Re: Twitter's Recommendation Algorithm

#670

Post by dopa42365 »

bakugo wrote: post #362
corbulo wrote: post #181

Earlier quoted context omitted.

I'm confused, then why not just use your 'followed' feed instead of 'for you'?

On the android app at least, a recent update made it so pressing the home button while you're in the "followed" tab switches to the "for you" tab. It's extremely annoying

If only there was any other way to visit a website.
Post reply on HN