Live data from Hacker News

HN Cheaters: Catch me if you can

porter.io

11–20 of 176 posts

Re: HN Cheaters: Catch me if you can

#11
Sounds like a fun project.

I did some digging in the upvote data of ProductHunt a while back, and although submitting a product needed special privileges, I easily found lots of dubious upvote behaviour. Unfortunately its hard to distinguish between people asking their friends to help upvote (eg via twitter) and 'spamming' upvotes.

If the actual content is good, do we really care how it got in the top ranks?

Re: HN Cheaters: Catch me if you can

#14
post #4

So how do you fix it? I don't really see another way to rank large amounts of stories other than points. Sure, you can try to detect cheating, but you want a minimal amount of false positives, too. Edit: missing word

It's a game. Cheating the system without being caught is win in YC's universe. But now, this is hunters' turn to take one step ahead of the game!

Re: HN Cheaters: Catch me if you can

#15
post #5

Trying to login to Porter.io via Github gives a Cloudflare error suggesting your server is down: Error 502 Ray ID: 1d351e9c19a519b6 • 2015-04-07 10:41:24 UTC Bad gateway

Scaled to a large instance with more instance. Should be stable for now.

Re: HN Cheaters: Catch me if you can

#16
This is really interesting, but not all examples are convincing. For example, this one:

> With huge front page traffic, it’s hard to believe that a good story would only get 4 or 5 up votes in its 3 hour window on the front page, therefore, it makes the first 3 upvotes look very suspicious.

Sounds like a typical 5-10 point story that didn't go anywhere? I feel like I see lots of them on HN. Is the described story _that_ abnormal?

The front page and /new are very different, and I'd say that when browse /new (which happens rarely), I focus on giving upvotes to stories that deserve them.

Re: HN Cheaters: Catch me if you can

#17
post #4

So how do you fix it? I don't really see another way to rank large amounts of stories other than points. Sure, you can try to detect cheating, but you want a minimal amount of false positives, too. Edit: missing word

One Idea would be to increase the "gravity" effect on the frontpage: If it doesn't get the expected influx of votes in a short timeframe, drop it. Obviously still cheatable but would require more (i.e. easier to detect) votes.

What a great way of penalising posts that take a long time to read.

Re: HN Cheaters: Catch me if you can

#18
Most of these can be explained by the fact that submitting the same article as someone else counts as upvoting the first submission.

It makes sense to me that submissions comes in waves (as several HN'ers could find articles they find HN worthy through other channels at the same time)

I am sure that some manipulation happens (I imagine a lot of people use their network to ask for upvotes to their submissions if it's something important to them), it's just not that clear cut

Re: HN Cheaters: Catch me if you can

#20
Good that someone raised this topic.

1. Many good articles & posts don't get upvoted.

2. There are pre-formed HN groups who game the system often.

My Feedback

1. Some special users can be given special Front-Page-Post Button permission.

Say, 3 Top Users have voted this Post for FP(Front Page)

2. Frontpage should not be the highlighting part of HN. At least it looks like that as of now.

The other good parts of HN: new|Ask|Show be shown side by side or given equal weightage.

3. Giving Better experience is the challenge.

The Front page feed can be a variable of Votes, Interest based.

There is lot of scope for improvements.

4. Users Gaming the system, be strictly banned for X Days and shown in different colors.

Post reply on HN