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

Hacker News metrics (first rough approach)

mjg59.dreamwidth.org

Post reply on HN
Print view View original

11–20 of 72 posts

Re: Hacker News metrics (first rough approach)

#11

Post by zorpner »

dang wrote: post #9
zorpner wrote: post #6

Earlier quoted context omitted.

In fact, it was: https://twitter.com/mjg59/status/527890332206510080

According to HN's software, this story has been heavily ring-voted. That's what has affected its rank. Call that "penalized" if you want, but it has nothing to do with the content, so that tweet is misleading—and ironic, too, given what provoked the "penalizing". (Also, in case anyone's wondering, moderators haven't touched the post.)

it has precisely nothing to do with the content

Do certain types of content set off the ring-voting detection more frequently? The type of content likely influences how it's shared and how people respond to it, which would affect how it's voted (and/or flagged). Algorithms are not unbiased, particularly when no one's bothered to study their confounding factors.

Re: Hacker News metrics (first rough approach)

#12

Post by grimtrigger »

I think there's plenty of evidence that humans by nature are uninterested in "politics", except when it might effect them. Then it becomes an object of great importance.

As for the role of the algorithm, I think its just an extension of "garbage in, garbage out".

Stories like this are a nice reminder that every community is also a bubble.

Re: Hacker News metrics (first rough approach)

#13

Post by JoshTriplett »

dang wrote: post #9
zorpner wrote: post #6

Earlier quoted context omitted.

In fact, it was: https://twitter.com/mjg59/status/527890332206510080

According to HN's software, this story has been heavily ring-voted. That's what has affected its rank. Call that "penalized" if you want, but it has nothing to do with the content, so that tweet is misleading—and ironic, too, given what provoked the "penalizing". (Also, in case anyone's wondering, moderators haven't touched the post.)

While I realize you can't go into too much detail about ring-voting detection to prevent people from finding ways around it, can you elaborate a bit on what ring-voting means to HN, beyond the obvious?

The obvious definition would be a clique of accounts that all upvote each other's stories, but that seems rather unlikely in this case; is there a secondary criteria? Are you looking to prohibit what Wikipedia calls "meatpuppetry"? Are you also looking to prevent upvotes from people who reach an HN discussion via Twitter or other sites?

Re: Hacker News metrics (first rough approach)

#14

Post by JoshTriplett »

zorpner wrote: post #11
dang wrote: post #9

Earlier quoted context omitted.

According to HN's software, this story has been heavily ring-voted. That's what has affected its rank. Call that "penalized" if you want, but it has nothing to do with the content, so that tweet is misleading—and ironic, too, given what provoked the "penalizing". (Also, in case anyone's wondering, moderators haven't touched the post.)

it has precisely nothing to do with the content Do certain types of content set off the ring-voting detection more frequently? The type of content likely influences how it's shared and how people respond to it, which would affect how it's voted (and/or flagged). Algorithms are not unbiased, particularly when no one's bothered to study their confounding factors.

I'm wondering that as well. In particular, it seems like content that's more likely to be shared via various social networks would be more likely to set off the ring-voting detector.

Re: Hacker News metrics (first rough approach)

#15

Post by dang »

zorpner wrote: post #11
dang wrote: post #9

Earlier quoted context omitted.

According to HN's software, this story has been heavily ring-voted. That's what has affected its rank. Call that "penalized" if you want, but it has nothing to do with the content, so that tweet is misleading—and ironic, too, given what provoked the "penalizing". (Also, in case anyone's wondering, moderators haven't touched the post.)

it has precisely nothing to do with the content Do certain types of content set off the ring-voting detection more frequently? The type of content likely influences how it's shared and how people respond to it, which would affect how it's voted (and/or flagged). Algorithms are not unbiased, particularly when no one's bothered to study their confounding factors.

> Do certain types of content set off the ring-voting detection more frequently?

Sure—people posting their startups, for example, set off the ring detector more frequently than the typical news article. Friends voting for friends' blog posts, too. But is that a bias of the algorithm? That strikes me as a stretch.

Re: Hacker News metrics (first rough approach)

#16

Post by somehow »

dang wrote: post #10

The Hacker News ranking algorithm tends to penalize stories that address social issues. The Hacker News ranking algorithm does not penalize stories that address social issues. It doesn't consider social issues at all, apart from a couple of special cases that we turn off once a major wave subsides [1]. The effects the OP is observing are mostly caused by user flags and the voting ring detector. 1. The intention there…

I would be more fair to address the broader point of the post rather than just one sentence. It's good that the moderator(s) specify why something is penalized[0], but it doesn't really change the overall result.

[0] As defined by the post in question.

Re: Hacker News metrics (first rough approach)

#17

Post by minimaxir »

JoshTriplett wrote: post #14
zorpner wrote: post #11

Earlier quoted context omitted.

it has precisely nothing to do with the content Do certain types of content set off the ring-voting detection more frequently? The type of content likely influences how it's shared and how people respond to it, which would affect how it's voted (and/or flagged). Algorithms are not unbiased, particularly when no one's bothered to study their confounding factors.

I'm wondering that as well. In particular, it seems like content that's more likely to be shared via various social networks would be more likely to set off the ring-voting detector.

You're confusing correlation with causation.

Sure, many article at the top of HN after often posted elsewhere (e.g. today) because they are incredibly newsworthy.

However, a HN post that people share on social media network for the purpose of upvotes is a different story.

Re: Hacker News metrics (first rough approach)

#18

Post by JoshTriplett »

minimaxir wrote: post #17
JoshTriplett wrote: post #14

Earlier quoted context omitted.

I'm wondering that as well. In particular, it seems like content that's more likely to be shared via various social networks would be more likely to set off the ring-voting detector.

You're confusing correlation with causation. Sure, many article at the top of HN after often posted elsewhere (e.g. today) because they are incredibly newsworthy. However, a HN post that people share on social media network for the purpose of upvotes is a different story.

I'm not assuming either correlation or causation; I'm suggesting a hypothesis that seems worth testing and evaluating.

There's a difference in motivation between "check this interesting article out" and "please upvote me", but the ring-voting detector can't necessarily tell the difference.

Re: Hacker News metrics (first rough approach)

#20

Post by minimaxir »

JoshTriplett wrote: post #18
minimaxir wrote: post #17

Earlier quoted context omitted.

You're confusing correlation with causation. Sure, many article at the top of HN after often posted elsewhere (e.g. today) because they are incredibly newsworthy. However, a HN post that people share on social media network for the purpose of upvotes is a different story.

I'm not assuming either correlation or causation; I'm suggesting a hypothesis that seems worth testing and evaluating. There's a difference in motivation between "check this interesting article out" and "please upvote me", but the ring-voting detector can't necessarily tell the difference.

> There's a difference in motivation between "check this interesting article out" and "please upvote me", but the ring-voting detector can't necessarily tell the difference.'

In the former case, why would they be linking to HN in the first place, instead of the article itself?

HN has a basic voting ring heuristic to ignore votes from direct links to HN submissions from a non-HN referrer for this reason.

Post reply on HN