back
▲ 35 points

Ask PG, HN: Is someone tampering with article positions?

by stagas·13y ago·24 comments·view on hn ↗
Can someone explain to me how the same algorithm can classify a highly controversial topic like the Monsanto story[1] in place 128 while another story[2] of the same publication time with less points and much less comments be on place 29 on the frontpage at the same time?

[1]: http://imgur.com/2IU0JCD

[2]: http://imgur.com/XspkTjs

24 comments
Flagging is taken into account with article position. People are flagging the monsanto article.

HN could be a useful place to discuss monsanto, but that thread is an excellent demonstration of why topics like that fail on HN.

There's a bunch of people who've gone into that thread with fixed opinions. They're not going to change their opinions; they're going to cherry pick articles to support their view points. Discussion is limited and polarised.

That makes sense and I understand it. However it seems way too artificial for me, because I was watching the Monsanto thread and it went from place 2 in the 1st page to the 3rd page in the span of a couple of minutes before anyone had a chance to express their opinion so as you see it now it's because only those of us that were in the thread actually continued discussion.
Advanced users can flag articles. Flagged articles sink. Controversial topics are more likely to get flagged.
Not just controversial. Some of us get our political news elsewhere and would rather see articles promoted on HN if they bear some relation to the theme. Flagging is the only way for the community to control what others post.
I had no idea that flagging was essentially a down-vote. Thanks.
It's worth noting that if you actually use it that way on any kind of a regular basis you'll very quickly have the ability to flag articles removed (I did, and it took less than a month and no more than a couple dozen flags)
There must be a bit more to it than that; I flag pretty aggressively (though admittedly only articles that I think are inappropriate rather than simply bad) and the link is still there.
I think there's some flagging-ring detection, I heavily flagged a group of spam accounts over and over again until I lost the ability which might look suspicious without examining the targeted accounts.

It never comes back either.

that seems to fit with me as well, i was flagging a spammy subset of articles on a subject that were being relentlessly promoted by a rather small group of people.
I flagged all the submissions by an obvious spam-bot once, and had my IP banned from accessing HN for a couple of hours, IIRC.
It seems much stronger than one downvote. Personally I reserve its use for blogspam, particularly toxic discussion, or if it's just beating a dead horse.
I believe someone is tampering with the position of links submitted. I say so because last week or so, I submitted our startup to HN which was on the first page for most of the day and receiving some positive comments/feedback. At some point in the day, we were no where to be found (unless I utilized the search functionality), only to appear again on the first page after a while. Later on in the evening, it dissapeared again.
That sounds a lot like flagging. cf. recent article on Margaret Thatcher that bounced in and out of existence. I actually assumed it was a moderator decision but info@ explained it got flagged dead, then put back by mods - which is almost a description of your events.

Oddly enough I have never flagged an article (except with a fat finger) - usually because I simply dont know what it does.

Yep, this has happened to me a couple of times as well. My post would be in the Top 10 for a few hours and then suddenly sink below 30.
Clickable:

[1]: http://imgur.com/2IU0JCD

[2]: http://imgur.com/XspkTjs

given that this is a meta thread, could somebody explain why all my posts end up on the News page rather than the Ask page. Is my karma too low to post the Ask page? I'm new

see this for example https://news.ycombinator.com/item?id=5534167

Does the rate that up-votes come in change the rankings?
There are a few ways that an article can appear lower than it should.

1) Flagging by people with karma over the threshold (currently set at 500).

2) For some domains, only 1/3rd of the upvotes count, e.g. Reddit, XKCD, images, tweets.

3) There is a "controversial" score which can bring down a story. Not sure how it is computed but perhaps it's based on number of comments versus number of points.

4) Voting ring detector. This can bring down a post pretty hard if it's triggered.

5) I suspect the HN supermoderators can cause a post to sink to the second page instantly.

Most of the time it's #1 that causes posts to go down.

I've usually seen it happen to posts that are anti-Apple, anti-Google or posts related to Microsoft that are not negative towards it. It's like the moderators flag posts that they don't want others to be seeing.

For example, the Surface Pro review

http://i.imgur.com/uFPTSqR.png

http://i.imgur.com/ADMcanz.png

5 is a big part too, a moderator flag appears to instantly drop a post about 60 ranks (from first to third page). I think when you see it only get to the second page that's user flagging.
That's interesting. Can we have background information for these "moderators" that are given these god like abilities? We need to know who's deciding what we should be reading or not.
Nobody's deciding what you should be reading. They're just deciding what gets top billing on Hacker News. I'm pretty sure none of the moderators believe the top stories on Hacker News are the only things worth reading. Hacker News is meant to be tightly focused on certain areas, so other things (e.g. politics) get buried whether you should read them or not.
pretty sure it's anyone thats a yc alum
The HN Rankings graph does show a slower decline, so my best guess is that it was flagging.

http://hnrankings.info/5523672/

Awesome tool! I had the impression it was less than that, but our mind can play these tricks. It seems a bit steep though, I'll compare it with other stories and see if I can find a correlation.