Hilarious that the original article was flagged off the front page, but this one isn't...
I find it very disheartening that the negative voices are being given so much weight. Everything that's worth doing will have detractors, and when it's something really worth doing it will have vocal detractors. Back when I had comments on my blog, every article I wrote that was any good had at least one person commenting that I was a moron or some equivalent statement.
Great things arouse passion - on both sides.
Giving 10x the power to the people on the negative side just creates an environment where new ideas are discouraged, where important but difficult discourse is pushed aside, where things of true import are penalised out of the group's attention by a few detractors.
There does need to be a system for flagging and removing spam articles, but if this system can (as it plainly regularly is) be co-opted to remove articles from sight just based on not liking them much, then it is broken. The people who have flagging powers are not responsible enough to use them wisely, perhaps.
I see at least one simple solution: lift the flagging privileges so it only becomes available to a much smaller segment of the population. Perhaps making the limit 10'000 instead of 500 would do that. That would still include hundreds of people, based on a quick extrapolation from https://news.ycombinator.com/leaders ). An even better model would be to make it dynamic - perhaps the top 200 commenters...
We don't let users abuse flagging. We have software that identifies users who flag excessive numbers of stories, and we take away their ability to flag.
Is the option visibly removed, or are the 'flags' silently ignored? When I post something I try go through a few pages of 'newest' and flag spam, but this makes me worry I'm wasting my time.
Just an FYI, but accidental flagging happens to me all the time on my iphone, would be nice to have a flag-confirm on mobile (I'd bet mobile flagging correlates much worse with controversy if others are having that problem too)
Company driven sabotage does not require excessive flagging, it requires a number of real-world accounts usually used in a correct way that are one-spot abused to avoid that the weakness of their product stays in the home page for too much time.
Yeah, I figure I'd probably get it back pretty easily if I asked Paul and apologised - but I really don't miss the "opportunity to be negative". So long as he's happy that there are sufficient people down-flagging the stuff that requires down-flagging, the site will happily exist without _my_ opinions of what's worthwhile and what isn't.
Apparently I was caught by that, yet I don't believe I flagged an excessive number of stories. I seldom used the flag at all unless it was for egregiously political or lightweight stories on the front page, or for spam comments a few years ago.
I doubt I used more than double-digit flags in the 6+ years I've been here.
I mentioned that I flag pretty much all of the climate change stories, and got my flag powers revoked. C'est la vie, I guess. I still don't think those stories add anything to this site besides some of the same old tired flame wars, so I'd probably continue flagging those and politics if I got the flag link back.
I'm sorry if this is blunt but in my opinion it is good that your flagging rights were removed. For every person who personally doesn't like climate change stories, there is a person who doesn't like Ruby on Rails stories, a person who doesn't like patent stories, a person who doesn't like NSA stories, and another person who doesn't like Node.js stories. Because flagging has such a strong effect on ranking, it should be reserved for highly inappropriate posts. With great power comes great responsibility.
Hardly any of the climate change stories are "interesting", but basically just "LOOK SEE I AM RIGHT IN MY BELIEFS AND THIS PROVES IT" sorts of articles. Those are poisonous to a site like this - they just beget a lot of useless discussion without much substance in it.
In other words, they are, IMO, highly inappropriate posts, not just stuff I happen to find uninteresting or don't like.
Climate change articles may not be interesting to you, but they are interesting to plenty of people. Might it just be your already formed belief that they are uninteresting that makes you find most of the climate change articles uninteresting? The comment you wrote here could equally well have been written by a person who is flagging any of the other categories I mentioned. The rules say:
"On-Topic: Anything that good hackers would find interesting. That includes more than hacking and startups. If you had to reduce it to a sentence, the answer might be: anything that gratifies one's intellectual curiosity."
Climate change definitely fits that description. Lets search for climate change: https://www.hnsearch.com/search#request/submissions&q=climat... None of those articles look any more inappropriate for HN to me, compared to what you'd find for any of the other categories I mentioned.
If you think a comment in a discussion is inappropriate, you shouldn't flag the story, you should flag the comment.
Do you really think most of those climate change articles are being posted by people who find them "intellectually gratifying"? I think most of them are pointing to stuff they think supports their own point of view, and they wish to share as widely as possible with the world.
How many articles are there about uncontroversial, but interesting aspects of climate science? Do any of them ever get upvoted, ever?
For instance, in the search that I provided, there are several articles that I find intellectually gratifying, for example "Decoding Climate Change with Perl, gnuplot and Google Earth" and "Climate Change Authority Admits Mistake" from the MIT technology review and "Nasa.gov: Evidence of Climate Change". Do you think these articles are less intellectually gratifying as articles about the NSA or articles discussing the merit or insanity of Node.js? If so, then you are applying a much higher standard to climate change articles than those other articles.
But whether or not YOU should find these articles intellectually gratifying is beside the point. We already have a feature for those kind of articles: don't upvote them. Flagging is not for those kind of stories. Again, if you find a discussion inappropriate, flag the comments in the discussion. Anyway, I don't think this discussion is going to be productive any more, so this will be my last comment.
Political stories are off-topic and should not be posted here. Most of the climate change articles seem to me to be pretty much about people's political beliefs, rather than an inherent interest in climate science or meteorology or something. That's why I flagged them. What other sorts of climate science do you find interesting? I find it odd that so many people are so very "intellectually gratified" by stuff that revolves around the very political "climate change" stuff, but don't seem to care too much for other kinds of more mundane science.
I probably would not have flagged the perl/gnuplot one, I'll grant you that one.
Node.js articles are on topic, even if one or the other happens to be boring. So I would not flag them, even the most uninteresting ones.
That sounds like an impressive way to create an echo chamber of opinions. "Looks like what you said isn't what other people believe so you shall be punished!"
Yes, it seems so. If you get caught on the wrong side of some upvote mob you lose. Whatever; this is what the oweners want and it's a silly thing in the grand scheme of things.
PG has said in the past that there is a flamewar detector, and that it suppresses discussions. In the past, this has been triggered on certain Microsoft related articles, which has caused several people to claim that there is some sort of ring of HN users who strategically flag articles about Microsoft off the page (and depending on who you ask, they also vote up Google articles). Ironically, it seems that at least some of the people making these claims were sock-puppeting themselves.
Votes made from an article's direct link (e.g. https://news.ycombinator.com/item?id=6799854) don't count towards its rank and too many of them will set of the voting ring detector (this is why it's a bad idea to link directly to a HN submission from your blog or from Twitter/Facebook). This is probably what happened here.
Wait...what?
I am not suppose to upvote from the article page but the main pages? I know I'm new here, but I find that counter-intuitive and disappointing that I've been doing wrong.
Wouldn't it make sense to remove the upvote option on the article page, surely I'm not the only one who got this wrong (I hope).
I think it's based on referrer, not where you actually do the upvote.
I.e. if you get to the article page from the main page, then that's fine. But if you arrive there with no referrer or from somewhere else, then it isn't counted, since quite possibly someone sent you the specific article just to upvote it.
On the topic of flamewars causing removal of articles, the difficulty, IMO, is that an entry on HN is composed of both an article and comments. Flamewars have terrible comment quality, so even if the article is upvote-worthy, it can mean the spot on HN is better occupied by something with a less interesting story but better comments.
As to flagging, I think a bigger problem is activism-upvoting, in other words upvoting so others see a story and not because it's interesting.
If you don't trust the average user with flagging powers because of how they affect the front page, why do you trust them to upvote good articles onto the front page? Clearly only select users should be able to upvote articles.
I think it's also funny that some people are under the delusion that sites like this and reddit, are actually controlled by the 'voters', and that everyone has the same power.
They're not. They're controlled by editors who will place items where they want to.
For me, the worst part of hacker news is the silent banning if you are critical of any YC funded startup. Censorship is ugly, but it happens routinely.
It pulls https://www.hnsearch.com/bigrss and re-sorts it with the basic formula (but without penalties). If people like it/return to it I'll clean up the UI for better readability/mobile use (and display comments there instead of linking back to HN).
EDIT: Interestingly, without penalties it looks like this thread would no longer be on the front page.
FOLLOW-UP: My note in the edit was due to a mistake in the implementation, kindly pointed out by OP via email. It is now fixed, and this thread appears in roughly the same place on both sites (there is some lag between the RSS feed that I use as my source and HN proper, so I wouldn't expect an exact match even if I did no re-ranking at all). It is, however, easy and interesting to spot which articles have been penalized; there's an NSA article at the bottom of the front page right now that's at the top of my penalty-free page.
I'll be modifying my page to highlight penalized items and calculate their penalty when I have a bit of time.
Thanks, very nice. How do you get story point scores from the RSS feed? The last time I looked at it, it was missing that crucial bit of information, and I don't see it now either. Or are you using parsing or another data source to get at them?
I tend to flag any article that is about analyzing, optimizing, or gaming Hacker News.
It's just a website for people to talk to each other and share links. Trying to out-think the algorithms is, to me, a smell for cynical PR or at best misplaced priorities.
I've often wondered about the possibility of gaming HN through it's participants. For example, pre-emptively down-voting yourself from a second account early on in a busy article, possibly with a reply using that second account noting you don't know why anyone would feel the need to down-vote that...
At some point over the past 2 years HN has stopped being my friend. The folks here? Great people. Very happy to have gotten to know many of them. But the system itself? Not so much.
People expect machines they interact with to behave in some kind of logical manner. After 2 or 3 times of submitting an article that HN has traditionally liked -- and watching it tank -- just not that motivated to submit more. After submitting my own articles, having people stop me in the hall and tell me they liked it and voted up for it on HN, only to see it have no votes? Not so motivated to submit more. After the tenth conversation about how people expect HN to act one way and instead it acts another? Not so crazy about it.
I think the problem here is that PG wants folks to participate, but only to a certain extent. People want to interact with the system, but on some kind of mutually-fair terms. I'm not sure PG's goals line up with the average user any more. There are good reasons for this, and I'm not trying to trash the entire effort. It's just that this is a tough problem. I don't think you can code your way out of dealing with messy human issues at scale. If you could, we'd all be managed by computers in 50 years, and that's not a future I would wish for my children.
"I think the problem here is that PG wants folks to participate, but only to a certain extent."
Short answer is HN is basically deal flow for YC. To that goal it obviously works very well.
PG and YC need "everybody" but yet they need "nobody".
As a business HN works to achieve the greater goals of YC. I don't think it is about anything else and I haven't been around long enough to know if things were ever different. They may have been back at the start because I guess if you were to arbitrary you would never build up an audience or fan base.
On a personal level I've always thought that people who get up to the karma level you are at actually deserve some special treatment because to me the top commenters are some of the glue that makes a place like HN special. In any business I have ever run you always give special treatment to the better customers.
How HN page rankings really work: you vote stuff up, and then the flag mob and hidden moderators axe right off the front page whatever irritates the pro-capitalist internet-libertarian techno-optimist idelogical sensibilities of the
white male Californian HN hive mind even in the subtlest of ways.
Isn't that the point of this site? That's mainly why I come here, to peer into the collective unconcious of the californian ivy-league portfolio-owning hivemind. It's like a zoo but with tesla owners instead of chimps.
I don't imply an ideal of neutrality; no online community is ideologically neutral, and neutrality isn't a merit to strive for. It's just that most participants seem to have internalized a supposedly meritocratic "whatever is interesting to good hackers floats to the top" mental model regarding how HN works, when that simply isn't the case. Every online community turns into a self-censoring echo chamber with time, and given its origins and initial purpose, the precise kind of echo chamber that HN has been turned into shouldn't be a surprise to anyone.
The part that bothers me is that it's not self-censoring. It's being actively censored, and sometimes the motives being the penalization are obnoxious; everything that doesn't directly promote california imperialism gets shunted to the side.
I believe it to be uncontroversial that all communities need some moderation, but the lack of transparency or involved around these parts has always made me feel uncomfortable.
I distinctly recall some articles on gender inequality in the industry…
Well, duh. The actual purpose of HN is as an advertising and recruitment tool for YC. It's certainly not for the benefit of casual commenters like ourselves. I think of it as a more tech-oriented form of television. Possibly one of those high-hundreds cable channels.
Wait, everyone thinks its a clean pure meritocracy but it is actually a complicated system of structural penalties and personal bias that sees specific apparently meritorious posts prevented from reaching the top? That reminds me of something....
HN is far more coastal-liberal than any flavor of libertarian. With a few exceptions, of course: people favor some policies relating to tech on which the conventional left doesn't have a strong opinion.
Not sure -- to me it looks like at least the majority of commenters are quite strongly pro-regulation: e.g. look at all the discussions of AirBNB vs. zoning just a few weeks ago, or 23andme more recently. Sure, voter and commenter demographics could be different, but probably not that much.
The post this morning, "Ask HN: What kind of side projects are you working on?"* got a lot of great responses and was killed because of it. There was nothing "controversial" about the post; it was merely popular with comments of what people are working on. This algorithm needs to be tweaked or HN risks losing its base.
Yes, with 82 comments and 52 upvotes (currently), that post will get a fairly substantial penalty.
For the most part you don't need to be "afraid" of comments - an article is only penalized if it has more than 40 comments AND more comments than upvotes. Most articles have considerably more upvotes than comments, so commenting won't hurt them at all. But you might want check the upvote and comment numbers before adding a comment just in case...
The number of links within it probably drags it down. Though I am ambivalent whether it's good or bad, I sort of see the point...there's not much discussion or analysis in the comments.
It's bad because this is exactly the kind of thing that many on HN come here to see. I hate sifting through the "YC startup raises $50" news, but I love to see what people are working on, and I'm not alone.
I think the issue is that it's mostly just links, and not much meaningful feedback. People are mostly talking past one another and there was [when I looked] very little dialog.
It was more like the monthly job thread and HN might have legitimate reasons for discouraging analogous threads for everyone's side projects.
It is also possible that the thread's pattern of growth triggered some heuristic.
I find it really disheartening to learn that any article with "NSA" in the title is pretty severely penalized by HN's algos. This seems like one of the seminal issues of the decade, for this community in particular.
That's as much of a problem as it is IMHO because hacker news doesn't appear to be designed to handle a wide variety of topics, a broad userbase and complexity of discussion.
The first is stifled by the lack of metadata (tags, categories, subboards, pick your poison) so everything rises or sinks within the same channel.
A broader userbase means groups of people who want to see posts about x and others who think x is destroying the community, and the strife caused by what may be the inevitable fact that some users want variety and others want bubbles.
This leading to the third problem, complex threads which can in practice be composed of more than 40 comments and more comments than upvotes without necessarily being a flamewar. Hacker news appears to be set up to promote upvotes and comments directly to the OP post, and discourage discussion between users. If that's what their ideal model represents then they should just move to a flat commenting system which makes it more obvious, visually.
I disagree that limiting a website for discussion on a small set of topics - an "information bubble" - is necessarily a bad thing. A web site can be designed to become a platform for a wide variety of topics (like Reddit) - or it can be designed for a small subset.
The internet is the platform that provides the variety: Hacker News for some topics, and for example RCGroups, SpaceFlightNow, or hundreds of other topical forums for others.
While it is silly to complain about "community X is going downhill", I do miss the old "information bubble" that HN was many years ago. The YC alumnus Reddit has taken over the role of a general link-sharing and discussion on every possible topic platform for every topic - but there is no longer any website (that I know of) focused on the advice, thoughts and discussion of technology startups mostly by founders of technology startups.
But for that to work you'd need a lot more restriction on signup and posting rules than currently exist, practically a closed, invitation only network. I'm not arguing that it's necessarily bad, I agree with you - the programming and technology content from professionals is why I'm here, it is unique. But there are a lot of people here who aren't startup founders, as an inevitable consequence of the site's popularity and the practically nonexistent bar on entry.
Invitation-only schemes only delay the inevitable, yet make the early members even more insufferably vocal about the gilded days of yore, the comments even more thoroughly interleaved with links to the Eternal September wiki page.
Downsides would probably be that the site as a whole might appear slower, and people might tend towards isolating themselves in sub-communities. Also if this is making money for pg as it is there's always the problem of killing the goose that lays the golden egg I suppose.
Tags for posts might serve the same purpose but allow the community to remain more cohesive.
You can apply this logic to several other topics of the day as well. What about bitcoin posts? What about all the Linux/MogoDB posts?
I find it interesting only the NSA articles get penalized. When in reality, you can make the same argument for a host of other topics not getting penalized.
I think we have to take the referenced article with a grain of salt. If I read it correctly, there is nary a mention of the flagging function. So, to put it bluntly:
Correlation does not equal causation
NSA articles may be more often penalized...but they are likely more often to be flagged. It's not that people here are defending the NSA or anything, it's that people have often said that they are tired of reading NSA all the time. And so, their reaction may be to flag those stories, giving them a penalty.
The referenced article does not really account for that, AFAIK
The behavior I saw with NSA articles is they were already penalized by the time they hit the front page, while other articles were up for minutes or so before getting penalized. This implies (but of course doesn't prove) the penalty is automatically applied, not the result of flagging. (And I mention flagging three times in the article.) Also, there is code in the published source to automatically penalize articles based on "rallying cry" words in the title. It's possible people are flagging NSA articles right off the "new" page, but it really looks like they are getting automatically penalized.
Edit: Several people said I should have said more in the article about flagging. There's no way to distinguish between penalties applied by administrators and penalties due to flagging. So there's not much I can say.
Ah, my bad...I was reading on a tablet so didn't do the Ctrl-F-find-the-word to back up my assertion. I think I'm likely biased into thinking that flagging plays a larger role, and thus expected to see it more in your analysis, whether it's the reality or not. But yes, capturing the state of article ranking as they hit the front page would help control for independent flagging behavior. It's possible that people are flagging on the New section but I have to doubt that...those would be HNers with a bit too much time on their hands.
This remind me initial war with Gmail on trying to change colors of links in your HTML email body.
Google was replacing #000000 with blue links to make sure some won't deceive users that underscored black text is or is not a link. So of course people start using #000001 color, then this one got blocked they went into #000002. Gmail quickly realized they wont win this one :) so they revert it back to replacing only 000000.
My guess is that it isn't actually the "NSA" keyword that is penalizing those articles but maybe some other type of filter. For instance, what if it instead penalizes topics basic on keywords relative to other articles? Basically a way to keep the content of the homepage fresh and unique, instead of having 6 different topics at one time on the NSA, or 6 topics on bitcoin, it instead penalizes those popular keywords to prevent a flood of similar content, with this filter actively changing based on the popular keywords at that time, If this was the case- I would actively encourage this filter, the thing that I love about HN is that it doesn't beat a dead horse.
<hat type="tinfoil">
No, the NSA are MITMing and deep packet inspecting everyone's HN traffic, and preemptively clicking your [flag] link on NSA articles…
</hat>
I find it really disheartening that there is at least one person who isn't completely tired of NSA articles at this point and even wishes to read more on the same topic. Same with Bitcoins, Snowden and so on.
That depends on the content of "more". "More new and game-changing information" != "more of the same expressed in different words". Government surveillance or non-governmental money are very broad topics, which can have a lot of different aspects. OTOH, reading yet another story "NSA spies on you" is excessive, I agree - we already know that.
The avoidance of controversial topics when talking together is one of those things we Europeans are typically not so good at. I know from many Europeans who like me lived in the US for a while that they had to learn the art of talking without touching controversial subjects. At first it seemed superficial but then I realised that it makes discussions that are not controversial but nevertheless important possible and I came to appreciate it every now and then.
Anyways, it would be nice if we in the settings could apply our own penalizing to subjects that we don't care about or that we find controversial instead of having others decide for us. But that would mean that submissions ranked differently for different users, of couse...
I don't think it's an American phenomenon. Every culture has their taboo and sacrosanct subjects - you're just used to the European ones.
From an American perspective, you could argue that it's tough to have a candid conversation about the monarchy in England, World War II in Germany, abortion in Ireland, or entrepreneurship and wealth creation in France.
Snap, I often have to put up with responses like; "So you want us to be like France then, you traitor? The queen makes us British. You are just jealous of her money. Etc, etc, etc."
Unfortunately many British people have the idea of royalty tied up very tightly with their national identity, so any suggestion of getting rid of the monarchy is seen as being treasonous.
Umm, no? It might have been when Walsingham was running the secret service in the late 16th Century, but I think we've moved on.
It is an opinion that gets discriminated against a bit more than most (along with Anarchism) simply because the Media and the Police don't take either seriously as ideas.
Example: People planning republican protests during the Royal Wedding were arrested in advance and held during the event, to stop their protest being heard[1]. As far as I know, nobody was actually charged with anything, only arrested on suspicion of "conspiracy to cause a public nuisance" and then released when it was all over.
Nonetheless, writing about the idea is perfectly acceptable, and a protest in a less sensitive area / at a less sensitive time would be "tolerated" (not that this justifies the censorship it gets sometimes.)
This gives the impression that all British are pro-monarchy. While it's try that royalty is tied to national identity, the nation in this situation is England, Ireland, Scotland or Wales. You'll find very different levels of republicanism in each country.
Personally I'm pro-tradition. If I wasn't, worrying about the complicated but small-impact question of the monarchy would come after issues such as the de facto Christianity in our post-disestablishment country.
How does a reply from one British person who doesn't think monarchy is a good idea, to another British person who doesn't think monarchy is a good idea, possibly convey the impression that all British are pro-monarchy?
Those three were all pretty much direct quotes. I don't know if programming would make someone particularly immune to this kind of reasoning though, in my experience it is to do with how much attachment someone has to the trappings of group identity, which can be quite disconnected from how they reason about other subjects.
I like to think that programmers are better than most at seeing the assumptions behind ideas, and testing them where possible (or taking the ideas less seriously if not.)
Maybe it's rarer than I thought to apply this to ones own social ideas, though. Shame.
I like to think that programmers are better than most at seeing the assumptions behind ideas, and testing them where possible (or taking the ideas less seriously if not.)
Did you go out of your way to test the assumptions behind this idea before taking it seriously?
Although a lot of people are (IMO) intellectually dishonest in what they really consider intellectually gratifying, rather than simply find themselves strongly agreeing with, or angry about. For instance, US immigration/border policies are something that I feel strongly about, but are they interesting from an intellectual point of view? I don't think so, particularly.
I definitely think jurisdictional issues can be interesting from an intellectual point of view. Especially since the Internet has made many such concepts far less obvious. One could of course argue that discussions (especially outside of obscure technology) rarely reach a level where they become interesting.
Most people here have a lot of domain knowledge about some combination of startups, hacking, and design, meaning they are quite qualified to comment about those things in an interesting way. Domain knowledge about legal issues is limited to a much smaller subset of people here, so discussions have a lot of strong opinions, but not a lot of basis in established laws.
Domain knowledge about legal issues is limited to a much smaller subset of people here
This is true, but often times we learn quite a bit from conversations digging into eg IP law. The bigger (and unfortunately) more nebulous area is who writes the laws? That is a question of politics, why discussions have a lot of strong opinions.
I remember when I first heard of the concept that you should not discuss sex, politics or religion in polite company. My thought then, which I still hold now, is that polite company sounds pretty damn boring.
It isn't a question of whether it is possible to have interesting conversations that do not involve those topics, I enjoy discussing a wide range of things. It is more that having forbidden topics as an indication of correct manners enforces a weird kind of false sterility on human affairs and I feel cheapens and narrows the culture that these manners claim to protect.
Someone who has seen lots of good forums wrecked by political bullshit. And to be perfectly clear, I do not mean "politics that I happen to disagree with", but politics, in general.
If the self constraint applies everywhere it's probably unhealthy but in some forums discussions on politics, race etc. are just not very fruitful and bringing up these issues could stand in the way of other discussions that have a better chance of leading somewhere.
The thing about controversial subjects? They are so often a waste of time. People are set in their beliefs. Most of the time the best outcome I can hope for when the subject turns controversial, is that no bridges were burned and no one is particularly upset. Forget about actually changing opinions.
Interestingly, the best strategy for keeping an article you like up high is to upvote it and not comment (or, if you must comment, do so only once.)
If you do comment, however, you can be as verbose as you like (as long as you are bland enough not to provoke replies.)
I wonder if this will change the strategy some post authors have of "hosting comments on HN" (and replying to every comment, even just to say "thanks".)
EDIT: and to edit your posts instead of replying.
I think this is penalisation of comments is a shame - I certainly come to HN for the comments, not the articles (although they're interesting stimulus for discussion).
Interesting. So if you want to get rid of the stories on the front page, and see more stuff, you should comment a lot. Because on HN a lot of comments is a death sentence for an article.
On the other hand, if you are an article writer and add a "discuss this on HN" link in your articles, you should remove the link as soon as you get a good ranking. Or actually don't ask people to discuss at all, because it is harmful, just ask them to vote and have your own comment system for discussion.
HN basically reinvented "sage", the concept from 4chan and its Japanese origins where people sometimes comment on a thread just to get it closer to the comment limit before it would no longer be bumped up to the front page when replied to.
Articles with a lot of comments without a lot of upvotes are probably rightly labelled "controversial". I sometimes comment without upvoting the OP on something I don't like. So this behavior by the algorithm matches my use of the site.
This article was more interesting than I anticipated. While I admire the tinkering which goes on with moderation here in an attempt to keep discussion civil and interesting, sometimes it has counter-productive effects. In particular this rule doesn't seem to work very well:
In order to prevent flamewars on Hacker News, articles with too many comments will get heavily penalized as controversial. In the published code, the contro-factor function kicks in for any post with more than 20 comments and more comments than upvotes.
Is a vigorous discussion bad? Should everyone commenting also upvote?
I find this interesting as well. Flamewars are counterproductive, but if I had to pick between flamewar and no discussion at all, I'd probably pick flamewar. After all, it seems like flamewars sprout when ideologies clash, and shielding ourselves from ideological clashes feels like it encourages insularity of opinion. On the other hand, one could argue that heated arguments increase polarization, so I'm not really sure what should be done. I just think there needs to be more dialogue on topics that people disagree on, not less.
My personal heuristic for reading is "if more comments than upvotes, don't bother". Such posts usually have a lot of repetitive, obvious comments -- typically bike-shedding, people trading anecdotes, etc.
Some specific examples might be questions that touch on child-rearing, or college experiences, or city-of-residence questions. Everyone knows next to nothing (sample size 1) but feels compelled to share. It's not information-dense enough to bother with.
Its a wonderful analysis, kens if you would ever like to come work for me identifying robot search clients just give a shout :-). I chafed a bit though at calling it a 'penalty'. Isn't it really a 'moderation' ? The scoring is adjusted by the moderators to be more the site they want to have and so they moderate articles that they feel aren't appropriately more heavily than those that are appropriate?
I understand that for some people the moderation choices offend them, I think that is unavoidable, but the goal is, I believe, to make a 'better' collection not to shoot down particular articles.
Assuming this article is correct regarding the penalization of comments, I'm a bit surprised (maybe even disappointed) that it is assumed that discussion is a sign of controversy. And maybe it is, historically?
It's a shame for those articles sparking insightful discussion though.
It seems like a weighted penalization could be implemented, potentially looking for red-flag words like "pedantic", or "not to be *". Or maybe it already is.
According to this article, the link currently in #3 [Vote Now: Who Should Be TIME’s Person of the Year? Edward Snowden](https://news.ycombinator.com/item?id=6800145) must be the victim of a very harsh penalty. I'm guessing 'Snowden' is a heavily penalized term as well, and having less than twice as many upvotes as comments might not be helping either.
It would be interesting to know if there are 'flagging rings' in the same way there are 'voting rings' and whether HN actively detects the former as it does the latter.
Comments should matter more than votes. A thread with all votes but no comments is a meaningless number, but a thread with all comments but no votes is exactly the same as that thread with votes.
That particular weighting doesn't make sense to me. Clearly some people vote by commenting. Which is as it should be.
I disagree. If you like something, vote for it. If you don't like something, comment on why and don't vote for it. If something just doesn't belong here, flag.
Not everyone has the same criteria for what content they want to read. Nor does it really help the world for all HN visitors to read the same content.
Would love to see ideas that broke from the model of a single ranked list: let folks tune their personal penalty amounts and gravity; add random jitter to rankings and throw a couple random new stories onto each list; classify/cluster users by their votes, so people who vote for jokes or NSA articles or their neighbors' articles (automatically) see more of those things.
It's maybe a bit much to ask PG and co. to architect radical alternatives to HN, because HN is a handful as it is and, besides, I hear they have day jobs. It could be cool to let a thousand flowers bloom: publish most of the now-hidden ranking data (maybe not all, because it can be useful to obscure how anti-spam algorithms work); let users opt in to publishing anonymous votestreams for clustering, etc.; then let other folks use all of this to make their own homebrew HN frontends within certain limits.
I suppose that, too, is kind of a pipe dream, because opening HN up for people to easily build their own frontpages is far-from-trivial for both tech and policy reasons. But it's a nice pipe dream.
If someone wants a 'weekend project': an interesting browser plugin would be one that undoes the effect of the penalties and reorders the page accordingly, it could even allow the user to choose the level of penalties enforced against stories in their view in case they did want some to disappear off the front page.
Using score adjustments as automatic community management makes a lot of sense; there are certain topics which are more likely to upvote/rank than others. Similarly, there's also a good bit of research that controversial articles generally do better on Reddit and other social news sites.
Applying an automatic penalty to certain topics / tactics which are likely to gather excessive upvotes, due to the nature of the content vs. it's quality, helps ensure you've got a diverse mix of content occupying the front page. Which is generally good for the overall user experience.
Otherwise, the front page will be a massive list of shock jock posts about the NSA.... [since controversial posts about those subjects will get sympathy votes, regardless of their actual contribution to the community...]
Instead of using ad hoc scoring rules like that, HN could use a machine-learning based system. This can also help solve another issue -- automatically determining the initial score of the new stories.
Hm, this post was kinda bad for me to read. I thought this was a completely community driven board. But I'm a fairly new member and some things never change, apparently.
What I find quite strange is that I have posted articles which show up in the new section, then later when they make the front page, it shows them posted as someone else. Either dupe detection isn't working (these are very simple URLs typically, like from the New York Times) or HN is rewarding the post to someone else after I've posted it. Weird.
As other folks have mentioned there is a raw feed if you want to see a non-penalized version.
It would be interesting if it somehow incorporated other elements to determine article "value":
- Open rate
- Ratio of comments to opens
- Time spent on article or comments
- Depth of comments
Sentiment analysis is becoming pretty accurate so maybe you could take into account the number of negativew/positive comments as well. Maybe that will eventually replace voting systems, who knows?
Interesting that github.com links are automatically weighted down, especially since almost every blog post they make hits the front page, and often projects hosted there show up fairly regularly as well.
This article is worse than the original, and the original article was crap because it never addressed flagging. I have on occasion flagged articles on a particular topic...sometimes I just get frustrated with recycled topics. It also doesn't address the possibility that penalties are applied when HN's heuristics suspect vote rings.
And given that the hackaday article is blossoms, don't be surprised to see it fall.
On average, about 20% of the articles on the front page have been penalized, while 38% of the articles on the second page have been penalized. (The front page rate is lower since penalized articles are less likely to be on the front page, kind of by definition.) There is a lot more penalization going on than you might expect.
I find it very disheartening that the negative voices are being given so much weight. Everything that's worth doing will have detractors, and when it's something really worth doing it will have vocal detractors. Back when I had comments on my blog, every article I wrote that was any good had at least one person commenting that I was a moron or some equivalent statement.
Great things arouse passion - on both sides.
Giving 10x the power to the people on the negative side just creates an environment where new ideas are discouraged, where important but difficult discourse is pushed aside, where things of true import are penalised out of the group's attention by a few detractors.
There does need to be a system for flagging and removing spam articles, but if this system can (as it plainly regularly is) be co-opted to remove articles from sight just based on not liking them much, then it is broken. The people who have flagging powers are not responsible enough to use them wisely, perhaps.
I see at least one simple solution: lift the flagging privileges so it only becomes available to a much smaller segment of the population. Perhaps making the limit 10'000 instead of 500 would do that. That would still include hundreds of people, based on a quick extrapolation from https://news.ycombinator.com/leaders ). An even better model would be to make it dynamic - perhaps the top 200 commenters...