Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Ask PG: Publish story upvotes data to aid voting ring detection
11 points by malandrew on May 1, 2013 | hide | past | favorite | 13 comments
Would it be possible to add one link to every story submitted to HN that goes to a page that lists all the people who upvoted a story that has been submitted?

Gaming the front page of HN (and other sites) has always been an issue, but techniques have gotten good enough where it is harder to detect. With that in mind and the skillset and hobbies of HNers in mind, it would be great to provide transparency so that data-science minded members of community can detect voting rings and provide their proofs to you.

It would be great if this page included: username, karma, timestamp_account_created, timestamp_upvoted and maybe the first three decimal dotted numbers of the IP address (or some other data point to help determine geographical or network proximity without sacrificing privacy)



I'm not sure I like this idea, it would mean all our votes are no longer anonymous. I think there is something to be said for not having to publish what you are voting for. You could anonymize the data with hashing but to make it useful it would still have to be able you to see what other items were voted on by one account. It would then not be too hard to match people to a hashed account is based on low voted items with only a few comments as you could assume they had also been one of the votes, with enough points of data you can then make a fairly safe judgment on who someone is.

I beleve PG has implimented a fairly sofisticated voting ring detection tool, I think I remember him talking about it before.


Comments and submissions are already non-anonymous, what's the advantage of anonymous voting ?

The knowledge that votes/flags are public may well change incentives, but I suspect the overall effect will be positive. That is people will be more thoughtful about what they vote up/down, rather than just voting for something because it's posted by a friend or flagging an article about a competitor.


Maybe then limit the published data to upvotes without comments and hash the usernames so you can link votes from the same account. At the end of the day, voting rings are going to almost invariably involve voting without commenting, since the cost of manipulation becomes much higher when you need to write a thoughtful comment as well.


Please no...

Why should this become a place of witch-hunting?

You give no reason why it should be implemented other than that gaming sites has become more sophisticated.

That is not enough justification, especially since

-) i don't see clear signs for HN degrading in Quality

-) i am sure that pg is concerned about HN and has implemented some measurements to guarantee the high quality of this site for the future

No need for such drastic measurement just yet.


Maybe because there is witchcraft being practiced? It's not like we suspect voting ring craft is merely a belief like witchcraft was. Voting rings most certainly exist and there is no doubt that they are manipulating the site right now.

If the data is anonymized and there are no details published that can be used to identify the real user involved, it's no harm no foul. This wouldn't be about witch hunting but simply pattern matching.

We wouldn't be identifying witches but the act of witchery. We would never know who the witches were, only pg would. There would be no elements of shaming or individual people called out. This would be entirely ad hominem free.


Voting rings most certainly exist and there is no doubt that they are manipulating the site right now.

Where's the evidence that these rings, if they exist, are having a detrimental enough effect on the frontpage to warrant something like this? Why is it so bad that PG, his fiddly algorithms and his staff of moderators can't deal with it?


We know they've existed in the past (such as the voting ring around The Atlantic), so are you asking if we've succeeded in catching them all? I highly doubt it.


No, i'm just wondering if their effect beyond whatever efforts pg is putting in, is big enough to matter.


Personally my motivation isn't just to improve the front page, but just to do this in the interest of science. We probably have one of the best communities on the internet insofar as percentage of members interested in data science. Providing this data would probably result in several HN worthy write ups on how someone went around detecting voting rings from the data made available.

Science should be a good enough reason to do it ;)


In theory, it sounds like a cool idea but I think it will be an overkill and of not much use. Not to mention that now everyone will know who is voting on what. That could start other flame wars besides the point of HN.

From what I have seen being on HN, the front page is usually of good quality and i m sure the voting ring detection mechanism is already decent.


The list of upvoters could be anonymized. Assign a random and unique ID to each account, and display that.

It would be pretty neat to let the hackers on Hacker News see what kind of nefariousness can be sniffed out.


It seems you could learn a lot about a person by what stories they upvote. Is there some way of anonymizing this data enough to prevent deanonymization, without making it useless?


Wouldn't publishing this data enable those who "game" the system to learn more about the system they are trying to beat?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: