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

671–680 of 1001 posts

Re: Twitter's Recommendation Algorithm

#671

Post by davesque »

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.

Yeah, but then what do you do after you measure it? Nothing? No, you make decisions differently so as not to offend whoever is part of the criteria. For example, can we agree that we don't want an "author_is_flat_earther" flag? Because who gives a shit if Twitter makes a change to their recommendation engine that negatively affects flag earthers? Just because something is only used for A/B testing doesn't make it completely inert.

Re: Twitter's Recommendation Algorithm

#672

Post by nonethewiser »

delecti wrote: post #422
lhnz wrote: post #320

Earlier quoted context omitted.

I think the decade old comment related to a different part of the code regarding the number of followers you have in relation to the number of accounts you follow. (Everybody on the call wants to remove this: I wonder why they haven't yet.)

Chesterton's Fence. In a sufficiently large system, you should be hesitant to remove things unless you're sure you know why it was added, and all the things that have come to depend on it since. I've definitely been hesitant to remove things I was pretty confident weren't used anymore, just because I didn't want to deal with the repercussions if I was wrong.

I didn’t know there was a name for this. Thanks for sharing.

I’ve definitely been bitten by this. You always have to weigh the chance you break something against the upside. If you’re actually fixing a bug, fine. But just refactoring to make something cleaner? Or deleting because it seems like it’s not doing anything, even after doing some research? Think again.

Re: Twitter's Recommendation Algorithm

#673

Post by nonethewiser »

tablespoon wrote: post #444
jandrese wrote: post #373

Earlier quoted context omitted.

I thought it was interesting how it explicitly doesn't boost independents. So much of the two-party system is self-reinforcing.

> I thought it was interesting how it explicitly doesn't boost independents. So much of the two-party system is self-reinforcing. Is it boosting? Others are claiming this code is just for metrics collection: https://news.ycombinator.com/item?id=35391896 . But on the topic of Democrats vs. Republican vs. independent; a big factor may be that "Democrat" and "Republican" are much more cohesive groups and therefore much…

Metrics collection so they can avoid bias, allegedly

Re: Twitter's Recommendation Algorithm

#674

Post by favorited »

djbusby wrote: post #651
philistine wrote: post #455

Earlier quoted context omitted.

American's lax attitude towards cultivating more than two parties is literally killing the republic from the inside.

IIRC Some states you cannot vote outside your declared party (Washington)

That's only for primaries, which traditionally were only open to members of a party.

Re: Twitter's Recommendation Algorithm

#675

Post by dorkwood »

xmprt wrote: post #666
infogulch wrote: post #657

Earlier quoted context omitted.

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?

If engagement on the tweets of that user goes down after a change has been implemented, you can roll back the change to prevent that user from being negatively impacted.

Re: Twitter's Recommendation Algorithm

#676

Post by nonethewiser »

XorNot wrote: post #556
sschueller wrote: post #237

Earlier quoted context omitted.

Well someone just asked about it in the live spaces[1] Elon is hosting and he said that should not be there. An engineer said afterwards it is just for metrics but then Elon chimed in again and said "we should get rid of it, it should be gone." [1] https://twitter.com/elonmusk/status/1641880448061120513?s=20

Of course he did because it makes him look bad and he's desperate for praise and attention. What he wanted was everything that feature provides, without it ever being shown that it's there. But since he refuses to hire PR people and almost certainly came up with this idea in the last few days, no one was paid to hide its existence. The next story out of Twitter will be the remaining engineers being threatened because…

This is pure speculation.

Re: Twitter's Recommendation Algorithm

#678

Post by spyder »

culi wrote: post #421
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…

what is vits? private val DarkRequestAnnotation = "clnt/has_dark_request" private val Democrats = "democrats" private val Republicans = "republicans" private val Elon = "elon" private val Vits = "vits"

Very likely this: https://create.twitter.com/en/articles/best-practice/2021/pa...

Re: Twitter's Recommendation Algorithm

#679

Post by slg »

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.

You can dismiss the complaint without measurement if you are confident in two things:

1. Your system does nothing to actually segment this specific group by their identity.

2. You are confident that the systems you have set up to reward good behavior and punish bad behavior are accurate.

If both of those are true, you know that even if the group is being disproportionately negatively impacted by some form of recommendation/moderation, that it is only because that group disproportionately participates in behavior that is bad for the platform. That isn't a problem. It would actually be worse for the platform overall if you did anything to appease that group.

Re: Twitter's Recommendation Algorithm

#680

Post by myko »

Auracle wrote: post #374
nonethewiser wrote: post #352

Earlier quoted context omitted.

Seems dumb of them. Cost is trivial and their competition that isn’t so politically motivated will have a much further reach. The smart move would be silent on the policy change, pay, and support rival platforms as they can. Instead they will eventually pay and look like they lost.

We’re in a time where ideology trumps revenue for some companies. You know, “Get woke, go broke.”

The phrase is "Better Broke Than Woke", we'll see how it works out for folks
Post reply on HN