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

Analyzing HN moderation and censorship

drewdevault.com

Post reply on HN
Print view View original

291–300 of 409 posts

Re: Analyzing HN moderation and censorship

#291

Post by dang »

gnicholas wrote: post #276

I was actually going to build a tool that scrapes HN every couple minutes and looks for stories that were hot (on the front page) and then drops by 20+ slots. I've happened to be involved in a couple threads where this inexplicably happened, and I was unhappy to see the discussion being muted (people previously involved continued commenting, but no new people saw/joined the thread). I was going to show the buried pos…

Why don't you just look at https://news.ycombinator.com/active? (via https://news.ycombinator.com/lists). Most if not almost all of those stories show up there.

Btw since you brought up the point, we moderate stories less, not more, when they're about YC or YC-funded companies. That's literally the first rule of HN moderation:

https://hn.algolia.com/?sort=byDate&prefix&page=0&dateRange=...

That doesn't mean we don't moderate them at all, since the internet would drive a truck through a cheap ploy like that. But we do err on that side, since keeping the good faith of the community is more important than any particular story.

Re: Analyzing HN moderation and censorship

#292

Post by tptacek »

anigbrowl wrote: post #287
tptacek wrote: post #200

Earlier quoted context omitted.

No. That doesn't work.

Yes it does, it's just more labor-intensive.

No, it doesn't. I've been doing it for something close to a decade now --- and, in fact, now have it pretty close to automated. I programmatically follow the people who write the worst racist and misogynistic and homophobic crap on this site. This is a problem I've studied, and one I think you know I've spent real, measurable, expensive time on here. I don't think you can dismiss what I'm saying so easily.

Individual flagging does not work to recover toxic threads. Flags are too slow and too invisible. By the time the worst stuff gets knocked down, its authors --- almost all of whom have no meaningful reputation on this site --- have reached the win condition of having guided the thread into hellish nerd litigation in which even the reasonable people grappling with the issues are made to write comments that serve trollish ends.

Flagging is what we have, and I'll continue to do it just like you will. But it does not work. It is inadequate to the task. This problem requires sitewide intervention, and until that happens, it is a rational, reasonable, moral choice to flag this stuff off the site as a harm mitigation mechanism.

Re: Analyzing HN moderation and censorship

#293

Post by dhimes »

aaron-lebo wrote: post #170
hsod wrote: post #164

Earlier quoted context omitted.

The first day or two of what?

Christmas of course. What I mean is if you read the comments on an article about James Damore the day it broke and read the comments were the article to be posted today, they haven't really evolved. It's the same canards over and over. Damore being an example of a story that users were indignant about getting knocked off the frontpage.

It must have worked. I had to duckduckgo James Damore.

Re: Analyzing HN moderation and censorship

#294

Post by ocdtrekkie »

krapp wrote: post #136
mnglkhn2 wrote: post #125

Earlier quoted context omitted.

You seem quite sure! Are you part of the HN executive group?

He's got six digits of karma. He's obviously a Big Deal.

More likely just the fact that I can't recall an active HN conversation which tptacek didn't post in. It's kinda easier to rack up a high karma when you have a LOT of comments to upvote. His account is also about to hit it's tenth anniversary of it's creation in about a week and a half. He actually only picks up about 78 karma a day average.

Re: Analyzing HN moderation and censorship

#295

Post by tptacek »

ocdtrekkie wrote: post #294
krapp wrote: post #136

Earlier quoted context omitted.

He's got six digits of karma. He's obviously a Big Deal.

More likely just the fact that I can't recall an active HN conversation which tptacek didn't post in. It's kinda easier to rack up a high karma when you have a LOT of comments to upvote. His account is also about to hit it's tenth anniversary of it's creation in about a week and a half. He actually only picks up about 78 karma a day average.

Almost all my karma comes from simple name recognition. None of it is merit. I think we should get rid of the dumb scores.

Re: Analyzing HN moderation and censorship

#296

Post by dhimes »

dang wrote: post #288
ploggingdev wrote: post #67

The voting ring detector is in need of a serious upgrade. There are so many cases of vote solicitation by people with meaningful social media following and their posts even make the front page. It's mostly the Product Hunt style "Hey we launched on HN, check us out" which is code for "upvote my post". People have stopped directly linking to the post submission for fear of tripping the voting ring detector, but they e…

> many cases of vote solicitation [...] and their posts even make the front page That last bit is the critical one. Vote solicitation that doesn't go anywhere is an annoyance, but fake upvotes getting otherwise lame content on the front page is a threat. If you (or anyone) see cases of this, I wish you would send them to hn@ycombinator.com. We investigate, and can usually figure something out one way or the other. Fr…

Often people think that's what they're supposed to do.

I recall a discussion here many many years ago where we agreed that we wanted that- it was a cool part of the site. I didn't realize it changed.

Re: Analyzing HN moderation and censorship

#297

Post by ballenf »

tptacek wrote: post #117
ballenf wrote: post #115

Earlier quoted context omitted.

I have zero problem with someone announcing on social media that they launched on HN. If they have a following it's because they have (or had) something interesting to say there. If that's the case and they have launched something, then there's a >0 chance that it is also interesting. If nothing else, it's interesting solely for the purpose of being more plugged in to what the masses think is or should be popular, ev…

Then, to be straightforward about it, you have zero problem with people who have large Twitter followings controlling the front page of the site, because that's what you'd get if Twitter canvassing wasn't heavily penalized.

No, but I'd rather have sophisticated vote ring detection instead of mods having to police the verbiage used in a social media post on another platform. Or having to worry myself about being too excited about my product's launch.

The policing of verbiage is so hard. If I posted the exact same thing to my followers that some paid influencer did, the effect would me markedly different. I would likely get 0-2 votes from my post, because I have real world friends as followers few of whom have accounts here. My point is that the same words used by different people can have very different implied meaning and outcomes.

It just doesn't make sense to me to judge a post on the words used(1) instead of looking at the voting patterns. For example, what if someone not involved with the project just loves it and spontaneously posts about it? Should the project creators be punished for that independent act? I'd say, yes, but only if it looks like a vote ring.

(1) You could make an exception for blatant vote begging, but even that would be caught by a good vote ring algorithm.

Re: Analyzing HN moderation and censorship

#298

Post by dhimes »

tptacek wrote: post #250
mbesto wrote: post #236

Earlier quoted context omitted.

So how exactly do they get recruitment posts on the frontpage?

All the YC startups get access to a queue of new job postings that run on the site.

This. To my knowledge it's been a perq of being YC since the beginning- or at least close enough to the beginning that the physics was just becoming stable.

Re: Analyzing HN moderation and censorship

#299

Post by tptacek »

ballenf wrote: post #297
tptacek wrote: post #117

Earlier quoted context omitted.

Then, to be straightforward about it, you have zero problem with people who have large Twitter followings controlling the front page of the site, because that's what you'd get if Twitter canvassing wasn't heavily penalized.

No, but I'd rather have sophisticated vote ring detection instead of mods having to police the verbiage used in a social media post on another platform. Or having to worry myself about being too excited about my product's launch. The policing of verbiage is so hard. If I posted the exact same thing to my followers that some paid influencer did, the effect would me markedly different. I would likely get 0-2 votes from…

Part of the problem is the mindset that you're being "punished" when a story about your project is kept off the front page.

In any case, you have to look at the whole thing as a system. If you allow people to point to their HN submissions on Twitter, the site will be unduly influenced by Twitter, rather than by the community that actually uses and contributes to the site. It's not a matter of intention --- I'm annoyed too that I can't casually link to things from Twitter! --- but of empirical fact.

Re: Analyzing HN moderation and censorship

#300

Post by anigbrowl »

tptacek wrote: post #292
anigbrowl wrote: post #287

Earlier quoted context omitted.

Yes it does, it's just more labor-intensive.

No, it doesn't. I've been doing it for something close to a decade now --- and, in fact, now have it pretty close to automated. I programmatically follow the people who write the worst racist and misogynistic and homophobic crap on this site. This is a problem I've studied, and one I think you know I've spent real, measurable, expensive time on here. I don't think you can dismiss what I'm saying so easily. Individual…

Well, I was thinking of different moderation policies rather than user-driven flagging of the kind you describe, so in that sense we agree.

My basic disagreement is that 'just ignore it' and 'don't do things that attract their attention' are not viable strategies for target populations, and it's unjust to systematically shift the burden of responsibility from persistent antagonists to their victims.

I'm interested in your programmatic approach, but let's talk about that offsite.

Post reply on HN