Hacker News Reader

Comment by Hitchhiker | Hacker News Reader

Classically, honeypots are put into play to avert malicious activity / users.

My gut instinct about it being applied in this format, while extremely creative ( wow - very cool thought experiment ), may have undesirable consequences. The intuition is merely from the fact that you are creating an adversarial premise for a wide-band community with varying maturity and motivation.

n.b. have personal / professional experience wiring such efforts and am certain you know top blokes who're ninjas in the game

View on HN

> you are creating an adversarial premise for a wide-band community with varying maturity and motivation.

I really have to disagree here. There is this misconception in these comments that somehow my system would let people get these massive egos and encourage them to harm others. That is simply not true, for several reasons:

1) You will never know if you are a super flagger, normal user, or ignored/penalized user.

2) Super flaggers gain no power to move things up or down for a specific article. They are merely there as a proxy for detecting if anyone is consistently upvoting improper articles. If the top 10% are super flaggers, the next 80% are normal users, and the bottom 10% are ignored, then that means the super flaggers will only account for 1/9th of the upvotes on average. And if they flag an article, it will not get removed faster than if a normal user flags it; rather, it will simply increase the chance that the article will be used as a honeypot in the future.

3) A single super flagger has little leverage, assuming you choose a large enough pool of super flaggers. One person will not do much to push the honeypot threshold over the top.

4) It's a moving target. So even if you were to ascertain that you are a super flagger and you decided to try to flag articles inappropriately (in the ever-so-small amount that you can do damage that way), you won't be a super flagger for long. Rather, you'll quickly be drowned out by your own noise and you'll fall off the super flagger list when the next update is performed.

That's not to say this system is perfect. I suppose one could manipulate it if:

1) You were somehow able to determine that you were a super flagger (non-trivial).

2) You were able to get a bunch of evil buddies together who also were super flaggers.

3) Your group is a sufficiently large portion of the total super flagger population, say 30%.

4) The admins did not include some oversight to periodically check up on what was being made into honeypots.

Then, yes, you could go to town flagging things for a while. It's certainly not fool proof, but if you had that large of a coordinated group on HN, you could wreak havoc in much more efficient and straight-forward ways.

It seems to me that most of the issues are caused by too few ideal HNers watching the "New" page and up-voting quality content. Instead, primarily controversial submissions manage to garner the necessary number of up-votes to make it to the front page before falling off of the "New" page.

I don't believe that the proposed honeypot solution would address this.

Random articles from the "New" page could be occasionally inserted into the main page to see if they catch any up votes.

I was not even thinking about the mechanics of your system, which as you've reasoned above and in the article may work beautifully.. merely the fact that HN would start deploying such a methodology opens up a road that may have interesting ramifications down the track.

Am sorry if my response sounds less precise or more philosophical than you want.. but it is well intentioned.

In a democracy , one common problem is that you have to respect others that you think are voting wrongly and put up with bad content. HN as it stands now is a wide-band place.. Eternal September is always going to be a risk.

Another way to address some of these concerns would be to have sub-sections ( much like a normal web-board ) where people are encouraged to discuss some common subsets / topics .. or even have a special section for newer folks.

HN already does things like deadpooling people who are consistently offtopic, trolling or just nuts. I think that unless steps are taken to keep bad content under control, any forum is going to go under.

Thanks for redirecting me here.

>3) Your group is a sufficiently large portion of the total super flagger population, say 30%. //

There will be populations that aren't interested in the original mix. These populations could swamp the site under such a system.

So if HN becomes popular with a particular niche who're not interested in the original mix and they become a large proportion of the population - for example: There are many ebay sellers ("ebayers") that like the site for occasional link they don't find elsewhere. These ebayers are usually inactive. But they start to vote against nerdy tech stuff and always upvote ebay related articles. The site focus will drift and the feedback loop will attract more users that want ebay stuff and put off others from voting (as their votes are getting ignored because non-ebay stuff starts to become a honeypot).

...

Anyway. Would love to play with such a system and see where it goes. Like I said before I'd love it if somehow the site could let you implement this whilst at the same time allowing me to ignore your honeypot system and just have displayed voting (+clickthroughs and saves). That is we'd be able to establish our own metrics. Then people could try different filter algos and choose which gives them the nicest site.

The real problem your honey-pot suggestion has is that there is a diversity of opinions and expertise across the site, and it's not the case that they're easily separable.

Should you discount someone's technical opinion because they are vocal about their political opinion? Because that's essentially what your system would do. And it would do so without notifying them that this was occurring.

Reply on news.ycombinator.com