Is it likely that there are people upvoting their own posts? Of course. It's likely because of the risk/reward ratio, which the post points out, likely reflects the reality: ie, easy to do, hard to detect, big pay off.
That argument is convincing.
I don't see any evidence in the supplied charts for cheats. The features identified in the charts could have been produced by many other factors, and even if they were produced by cheats, the features themselves are not that convincing: "the rough shape of a graph". It reminds me of people who have systems for Forex investing based on the shape of a graph. Compelling indeed, and yet probably doesn't work. More extensive study and testing would be required to validate the method.
My impression is this argument is fitting data to conclusion already assumed, rather than discovering a conclusion from the data.
If the proposal is then to administer a sort of "justice" via this approach, it's really no different than a kangaroo court.
Far more importantly, if some kind of analysis from rank and votes were able to identify cheats, I think those capable of doing so should really think through, go slowly, and work with HN/YC before deciding to "out" them.
Self-anointed vigilantes outside of the system almost never work as a way to create justice, because they end up wielding a power for destruction, which the community itself has not organically decided to give them. By holding power to decide who is a member of the community and who is not, without also shouldering the responsibility for building and constructing that community, they have an outsized ability to reshape the community away from the forces that built it.
In some cases, like when a community’s culture is so toxic to people’s common good, this kind of dramatic shift from equilibrium works as a way to restore an order more suited to people.
I don’t feel HN is in need of such treatment.
Sure, there's likely pain if a post doesn’t get upvoted, and that doesn’t mean that there is anything that doesn’t work about HN. It also, as the article points out, doesn’t meant there is anything wrong with that post. I haven’t noticed a limit on the number of times someone can submit new stories, so if a post doesn’t succeed at first, it can always be tried again.
The response to the upvote-pain could be to say, "What can I learn?" or it could be to say, "Who can I blame?" I think the implicit assumption that “the HN voting system is flawed” is incorrect. For a poster it’s a compelling idea to “explain” a post's lack of desired upvotes, because then I doesn’t have to take responsibility for ways they I have iterated and improved my post. I choose to blame an external system, in this case “the voting system.”
It’s a very compelling story because then whenever I post something that doesn’t get upvoted as much I want, I can say, “It’s not my responsibility, the system is flawed, and other’s successes are fake.” That can make me feel better. Except, it’s fake because I’m not actually feeling better because I created something of value, only by diminishing the value of something in comparison, and thereby elevating something which otherwise may not warrant elevation outside such straw-man comparisons. Secondly, it doesn’t work for me because then instead of learning from that experience, and iterating, I effectively reduced my learning to zero. Instead of saying, “what can I do that works?” I said, “The system doesn’t work. Now I will prove that is true.”
If we put all that aside and assume that the system really is flawed and a method really can identify fraudulent post elevation, then I still disagree that “outing” HN members is something which will work. Trust amongst HN members, the sort-of anonymous quality of it, even the unspoken “sacredness” of the voting system as a way of bringing us content, all contribute to social harmony here. This place has been shaped by the forces on HN and YC that sweated to build and unite it. If suddenly an asymmetric power arises outside this system that can identify cheats, and publicly outs them, this divisive force, used without the same care for the sacred harmony of community that the people who built it have invested in it, there will be a lack of harmony. This vigilante power would erode the trust and bonds between HN members.
It doesn't work trying to simultaneously mend the social fabric by breaking it. The threads are the laws and the people who protect and respect them. An entity which, outside of the fragile fabric of justice, decides which members are "good" and which are "bad" is eroding the lawful links that connect people, which were decided by the gestalt. Those laws, and unspoken codes and culture, were not created by one fell swoop of a powerful tech. They were created by sedimentary deposit of 1000s of man years of collaborative time and interaction, whether on HN or in other communities. This is how these bonds are shaped and the harmony they create is precious. It works that the ones who create this order in a community, with the members themselves, are the same ones who should wield the power to exclude members to protect some values, rather than an asymmetric power self-granted to some external self-anointed "authority". This seems unlikely to create harmony, because it is an external force imposed rather than one that arose and was guided skilfully by highly-invested and integrated community maintainers.
In this case the values to be protected are the credibility of the karma system, which is certainly an important aim. I don’t think public naming and shaming is the way that works to do so. And if it turns out it is, then it works to be based on a solid evidentiary process.
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.
As someone involved in news delivery, I find this to be an opportunity. I frequently dig up stuff with only 1 or 2 votes here that I can share on and look like I found it - ha! :)
As a long timer HNer, I think the main problem is there's little reason or motivation to monitor /newest so it's mostly visited by people being asked to vote up other people's posts. It'd be very cool if on the bottom of the homepage, you got a few items from /newest you were invited to vote on without actually going there.
Nice idea. This is HN's biggest problem I think - there is no real incentive to visit /newest and the vast majority of good stories slide off it without getting more than a few votes. Adding some new stories to the bottom of home would be an excellent way to get votes.
The other problem of course is that what is popular is not always what is good.
I really thought you'd have automated 99% of the information gathering for your newsletters based on something like that. :)
The value of gaming the frontpage could be reduced by having random items from below the fold bubble up from time to time, and perhaps by also having popular items bubble down. Not only would this encourage worthy but not necessarily popular stories to be read, it would (more importantly to me) encourage users to notice that there is more than one page and those thirty slots to read. I believe lobste.rs does this.
It also occurs to me that some people find stories through /comments, and this page can also have an overcrowding problem when multiple comments for the same story take up most of the page. More stories would be discoverable there if comments were grouped by story instead of individually.
I think /active should replace or sit alongside /new. It would provide a nice contrast between the stories getting upvotes and stories getting comments.
Focusing on Karma as the outcome is vastly underselling the power of sites like HN (and especially reddit). Those who want to warp the system don't just want numbers on this site. They want power over crowds or money for their business.
I'd like to add something: replace the down-vote button with a "report spam" link.
The one thing that makes me rage on HN is seeing people getting down-voted for having a disagreeable opinion. It's supposed to be to suppress trolling or spam, but people just can't help but use it to oppress others' opinion. Just on principle, it makes me angry, but for practical reasons it's irritating as well because it becomes harder to read often more interesting comments.
I don't think users gaming the system can be spotted so easily. Let's take an example: let's say both of us like Astronomy and post sometimes articles about spatial exploration. It is only natural that we would end up upvoting each others' submission, especially if we both like to check the new submissions and tend to read HN approximately at the same time. Users like I described could easily be mistaken by a voting ring detector.
By giving harsh penalties like you suggest to users who are gaming the system, you take a big risk to alienate people who simply have shared interests. I think the ranking algorithm of HN takes voting ring into account, so if you and I are part of a voting ring, our upvotes will impact less the standing of a page - I think I read about that somewhere but can't find the exact source.
Having articles appear on the front page briefly, to seed views and see if they stick could be helpful to give then a chance. A report button would quickly stem any spam.
5. Use of better colors: For following keywords or topics & colors for users by active | top | influencers etc.
6. Verified accounts.
7. Use of on Demand Badges like: Trustworthy User Badge, Looking for(Badge), Looking for co-founder, Hiring, Remote jobs etc.
All these can be grouped under a Paid plan? I would definitely pay for value & additionals, even a good mobile app experience is missing.
8. There is lack of networking and peer discovery in HN.
They should add community managers & make this evolve. Take it to the next level.
9. Follow options: Follow this discussion, Alert-me's, Follow Posts by keywords.
10. Finally, there is lot of scope to Improve.
HN team & YC should Rise up to the expectations and deliver richer experiences. What are they doing?
11. Add a separate Top Link: Voice-UP your concern.
This is for: Alerting, finding loopholes in the system(gaming the system etc), User Site feedback, new interface ideas.
Creating a vibrant HN ecosystem is the way looking forward. Lets do it.
The simplicity of HN is something innegociable IMHO. As I understand you are proposing to evolve HN into a mix of Reddit-social network for hackers/entrepreneurs.
---
Sam I think you have great power to give /newest more views if you'd consider minor redesigns.
Perhaps you could dither stories so the front-page list goes top/new/top/new/top/new, etc, but that is potentially very messy.
Or perhaps instead, the front-page can show 30 top stories, then a line break, then 30 new stories, all on the front page. Long-scrolling pages are in fashion, after all.
These aren't well-developed ideas, but the point remains that the inertia of being one click away means that a huge percentage of people will never even look at /newest, never-mind up-vote interesting stories.
Please consider changes to modify the median behavior. It's within your power and would do us all good. It's worth an experiment, isn't it?
---
Sam replied: "good ideas here and in the responses. we will consider", but I guess they never got around to testing any of them, or nixed the idea, or mere inertia, etcetc.
I've had front page stories on HN maybe five or six times over the last couple of years. I don't feel anything has changed over that time. If you look at my submission history, you'll see about 90% of what I submit gets no votes. I once tried submitting a story a couple of times when I felt that it was getting less traction than expected - it made no difference.
I'm actually surprised at how well HN surfaces decent content (I'm quite happy to accept that my stuff which got no upvotes just wasn't of interest.) Whenever I put something up it gets about twenty eyeballs off the 'New' page so real people are looking at it.
I can say for sure that I don't have any friends with HN accounts (much to my disappointment!) and I've never tried to game it.
Approaching 3000 days on HN. The things I notice most:
* increase in users has increased the churn rate of new submissions.
* the new page is dead because of the submission rate. good stories disappear off the stack quickly.
* the up-click is not a good indicator of story quality. I now see stories with up to 80 votes sans one comment. I used to read the comments BEFORE the post. This has now changed because of limited comments.
* the rate of high vote submission decay, means some good stories are lost as the HN crowd chew over some interesting post(s).
* also interesting to see users comment/post ratio. A lot of ppl have limited or no posts yet they comment. If you look at my post profile, in the last 60 days I've posted 100 posts. Approx 1:25 hit > 100. It would be interesting to see how this has changed over time. cf: https://news.ycombinator.com/submitted?id=bootload&next=9025...
The biggest problem is the dilution of good posts/actual posts. Slow the rate of posts and might influence the quality of posts.
On the comments:upvotes ratio - I think there's two types of submission.
Blog or newsy stuff invites discussion and comments. But other things, like Show HNs, for instance, might be upvoted for being useful but without anyone commenting much. Both are fine.
Better enforcement of the deeply interesting guideline would help but stories that are intensely interesting get many upvotes so it's unlikely to change.
In my experience so few people visit the /newest page, and so few in general vote, that there is a disturbing amount of chance involved.
That was long enough ago though, don't know how much has changed since - and I submit stories very rarely.
However, I'm sometime quite disappointed that what I submit does not get traction because I really think it is interesting to the HN community and the discussion here would be even more interesting. I guess the time of posting and luck plays a big part in what is upvoted.
I sometimes link my newly created posts to friends or IRC (some of them have HN accounts and may upvote) but it doesn't seem to make any difference.
The first chart goes to over 1,000 votes. The other ones stay one or two orders of magnitude lower. Cheating or not, it's absolutely normal that a stochastic phenomenon looks smooth when there are many events, and not smooth when there are few of them.
This comparison is statistically on very shaky ground at best.
Sidenote: HN page rank is not solely based on upvotes (for example the recurring recruitment thread). So analyzing it like it was will not provide reasonable results.
1) Yes, submissions are manipulated, but due to the flagging mechanic, any bad submission with vote manipulation will be shot down. It's worth noting that public vote manipulation on HN (https://twitter.com/search?q=https%3A%2F%2Fnews.ycombinator....) is far, far less worse than the meritocracy known as Product Hunt (https://twitter.com/search?f=realtime&q=product%20hunt%20upv...), which doesn't penalize spamming people for upvotes.
2) Using startup-esque techniques like Recommended Articles and Verified users won't work, as it will kill the simplicity of HN (note that Reddit has tried similar systems not too much success). Content is the most important factor to determining upvotes.
3) Yes, there are a few articles by YC alumni over the years which receive suspicious amounts of upvotes, and I'm disappointed by that. (case in point, see the cofounder of ReelSurfer's submission history: https://news.ycombinator.com/submitted?id=njoglekar)
Also, there's no correlation between negativity-in-comments and the number of upvotes an article receives.
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
(I might be wrong on this, but it definitely should not be taken as a sign of cheating. It happens way to often, with links that have no obvious commercial motive).
Also, the first abnormal result the OP was talking about 3 votes. I don't think we can worry about a handful of people voting for their own startup submission. The first, "normal" graph had hundreds of votes.
I think my takeaway from all of this is that, like other statistics, with small numbers the values have little meaning. Now if we had seen evidence of, say, 25 or 50 people working together (a few people with multiple accounts each, for instance), then that would be more actionable evidence. But my guess is that pg has already dealt with that problem, and apparently pretty effectively, because it's not showing up here.
This isn't to say others don't do it to, they do, just that it's another "benefit" of going through YC.
I've thought about this a lot and believe that the way for us to help YC startups make HN's front page is by giving them our best advice about what the community finds interesting vs. what tends not to work—and then sharing the same advice with everybody. In other words, do for HN what PG did with advice about startups. Everything he tells YC startups to do and not do, he published in essays that anyone can read.
That approach makes even more sense for HN, since it's in all our interests to have better stories, regardless of who posts them. I hate to see startups (YC or not) put a ton of effort into content that is unlikely to resonate with the community. There is much to say about this. Unfortunately, I am a slow writer.
The interesting thing about the flag button is that it's more impactful than an upvote, which means that if the system is gamed and an article is vote-rigged, it makes the article more prone to be flagged and it immediately corrects itself. It's an unstable equilibrium.
I recall Digg had a massive "digg brigade" problem where certain groups of users and particular power users like MrBabyMan would make the homepage on what felt like a daily basis. Solving these kinds of problems is a lot harder than just analysing data and looking for patterns, because sometimes innocent users can get caught up in data.
If solving cheating on social link submission sites was easy, it would have been solved already. Even Reddit suffers from the same issues.
> 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.
1- Some source sites are penalized heavily. Sites with a history of low quality articles have built in negative modifiers (ie Gawker)
2- Posts with more comments than upvotes are penalized, on the theory that someone is flame warring in the comments. This actually seems to work out more accurately than you'd expect.
3- Flagging, which you don't see as a visible indicator until a story is dead. However, there's some threshhold beneath which it affects a story's ranking, but it hasn't yet killed the story.
And the mods sometimes manually tinker with stories they feel are / are not what they want on the site.
... one mans flamewar seems to be another mans: it takes a few tries to explain this.
I'm find HN is very civilized. On other forums I see people talking about nuking entire counties, calling each other names etc all while posting under a full, real life names.
The thing that bugs me at HN (except interesting content disappearing like this post describes) is anonymous downvotes, - and mostly not because they hit me (usually they don't do : ) but because so often they are so obviously a result of political correctness and groupthink that it hurts.
I lump these types of posts in with those that decry how unfair Google's hiring process it and how they reject highly qualified engineers. Again true, but there goal is not to hire every qualified engineer. There goal is to make sure every engineer that they do hire is highly qualified.
If my contributions are valuable. I will get upvotes. If they are not, then I won't. Without gaming the system, I believe this was by design (PG also shared this in a few of his earlier posts).
In reality. I'm always sharing links with the community of things I find interesting. My experience is that rewards for sharing links (upvote karma) seems to be very random, with stories I think are genuinely fresh, conceptually challenging and interesting getting far fewer upvotes then stories which I consider to be relatively dull.
Subcategorisation does not need to be user-driven like reddit though and I'd encourage the admins of Hacker News to control that themselves with the type of content posts they would most like to see us post around.
If they are open to suggestion, I'd at least consider categories such as: Startup Advice (like Ask HN), Innovative tech, Show HN (as its own category), VC
The vast majority of submissions that hit the front page cannot be explicitly linked to one company, or a product, which would indicate that the vast majority of that which is on the front page is there because users liked the content and upvoted it.
If some parties are occasionally abusing the system to promote their own content then I don't think the answer lies in a dramatic overhaul to everything including throwing in flair pieces to submissions and giving people special accounts.
Perhaps a solution would be to disable the upvote button until a user had a certain amount of karma points, as is done with the downvote button? Or maybe add weighting to upvotes based on the amount of karma points a user already has? Or maybe monitor submissions for actual link clicks and use that as a weighting i.e. to prevent people upvoting content they themselves haven't even looked at?
All of that is game-able, of course, but not without significantly more effort than is currently required, and the abuse of the current system appears to be infrequent as it is.
I don't think HN needs any radical changes.
I think it would be possible to build review queues for user by listing stories positively correlated to stories you up-voted and even more correlated to stories you commented.
Why does it indicate cheating? Couldn't it just mean that there is a variable that you are not accounting for in the ranking algorithm?
Posters would self-organise and voting gangs would talk to themselves. Gaming it would mean that voting gangs had to have convincing interests etc.
An old writeup on the idea: http://williamedwardscoder.tumblr.com/post/15581427232/self-...
Hence there are some front page posts that have nothing to do with cheating. Just to add some balance here.
I had a submission for three days in the FP and that was the only time I didn't ask any friend to upvote me over IM.
> Getting on the front page is too random.
So in other words, getting to the front page is hard because you don't know your delta-V and TWR before launch, and the cheaters are the ones who are able to secure an estimate for those parameters.
#ObscureSpaceflightReference
For example a magic button for admins or moderators that will boost a story to the front page irrespective of the votes or time the stories have been submitted.
Because some older stories with little number of votes come up to the front page where new stories with same number of votes doesn't end up in the front page.
What do you guys think can there be a power for a selected few?
This article makes a BIG assumption about what constitutes a good story and about how HN users actually use the upvote feature. The least subjective way I could define a good story on Hacker News is by basing it on the only visible metric - number of upvotes. This doesn't mean I agree that the article is good from a subjective perspective.
Upvotes could be given for lots of reasons and the higher number of votes could just be because the article is about a subject that more people take an interest in (you also have to take an increase in casual visitors into account). An article about a niche subject, however "good" that article or product is, is less likely to get a higher number of upvotes because less people are interested in it. And posters need to ask themselves if their article or product really is of interest. Posters would also need to put effort into promoting their article elsewhere too (twitter, reddit, other social) and not just expect results from HN.
At the lower right corner, every minute or so, pop up a small box containing three or four newly submitted links. The box is only shown for five seconds, then it disappears by itself. Of course users can close the popup box with a click (and can turn off the feature altogether).
In this way a whole lot of new links are exposed to HN readers.
I don't mind this distraction at all, because 90% of time when I am on HN I am not working, have no need for intensive concentration. Frankly, I am often looking for distractions and unexpected/unusual contents. The more links I can see the merrier (come on, I know you agree with me). When I see an interesting link in the popup, I might click it right away, or just keep it in mind.
The links shown can be the same to everyone on HN at the moment, or better, tailored to the user depending on many factors generated by some testing algorithms.
The benefits are more exposure on newly submitted links and probably more additive and longer mindless browsing on HN :-)
IMHO, it's rare to see obviously gamed submissions, and the quality of front page articles has been pretty decent for something that rises out of hive behavior (although it's not as well targeted to my own interests as scanning-and-hand-picking-what-I-like, obviously).
My own theory is that with the high volume of medium-high quality submissions (i.e. a post can easily fall off the first page of /new in less than an hour), you end up with lottery-like dynamics since your post is likely not in the "breaking news" magnitude.
Edit: missing word
I am fan of adding randomization. It was previously proposed: show few random new items on a front page in random places, maybe with a little threshold.
I was thinking about community page destined to fork. After community reaches certain threshold (size, activity) there is automatic fork and 2nd generation community is created. Some algorithm adds accounts for certain users and blocks some other users from account creation. It's just a pipe dream...
We tested this idea (though we didn't roll it out for everyone) and the results were terrible. The median story is too low in quality for randomness to add value here. You just end up planting junk on the front page and annoying people. From that we concluded that there needs to be additional filtering, be it by algorithms, humans, or a combination.
Obviously still cheatable but would require more (i.e. easier to detect) votes.
I don't have any data but I would be interested to know if these companies are gaming the system. It sure would be in their best interest and I can't help but shake the feeling that they are after looking at the front page sometimes.
It would be good to get a 'hail corporate' button / profile setting that could toggle some of these domains from your front page.
Our current idea for the solution is twofold: create a new review mechanism for stories (not to replace /newest but to live alongside it), and reward users with karma for participating. But we're a fair way away from having anything worth rolling out.
Error 502 Ray ID: 1d351e9c19a519b6 • 2015-04-07 10:41:24 UTC Bad gateway
I'm already drowning in information overload. HN feeds me several good stories a day. If it got any better I would probably stop reading because it would take too much time and work.
Not to say I like cheaters...but short of adding the equivelant of subreddits...I'm fine with only enjoying a few top page stories a day.
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?
I wonder if there might also be a more nefarious problem where say YC alumni down-vote competitors?
I think that the lack of down votes for normal people means that you end up seeing a lot of support for some things that are popular with the people who CAN down vote.
Maybe I've got it wrong, but it seems that way.
Anyone with an account can flag a submission. Flagging should only be used for submissions that violate site guidelines. No-one can downvote a submission.
Anyone with an account can upvote comments. (Click the timestamp to reveal the flag link.) Anyone with an account can flag comments - again flagging should be restricted to comments that violate the guidelines.
People with karma over 750(?) 1000(?) can downvote comments. There's disagreement about when downvotig should be used. Some people feel it should only be used for posts that violate site guidelines[1]; others think that downvote to disagree is okay.
[1] that makes the flag redundant?
I'm not trying to be too negative here...I'm just saying. There are days when the number of Go stories is just crazy. It reminds me of the Linux advocacy of the 90's where people seem to think they are doing a great good by posting them.
I realize that it could also just be a very visible indicator of the overall zeitgeist of people using the site. Something tells me this isn't the case, though.
I recommend providing more detail about how it works, to gratify readers' curiosity.
Just look at the abnormal vote patterns on stupid stories concerning Google...they've mastered the art of getting their shit on social media.
Unfortunately the masses are too stupid to tell and Google continues to destroy freedom on the WWW by curating the WWW and controlling what is seen and what is not.
Google killed the free(libre) WWW and now it's their ad crap box.
