I had once suggested that the main page be woven out of a combination of popular + new, perhaps 50/50:
1. popular
2. new
3. popular
4. new
This injects a lot of noise into the front page, because most of new is noise, but it helps.
Another option is to front-load popular stories, then have 10-20 "new" items at the bottom of the page. That way the front page remains highly salient, but anyone who merely scrolls more will also see a small chunk of new. So 1-30 popular (as it is now), then 30-45 new (making the frontpage longer).
Most new post are just too bad. (I've seen them.) People would write angry comment complaining that they are in the front page.
There is something similar, the mods cherrypick some post and they have a second chance and are shown in the front page for a small time. More details by dang in https://news.ycombinator.com/item?id=11662380
Out of frustration of not seeing a lot of good blog posts that are submitted everyday on HN. I built a newsletter which gather all (EDIT : a lot of ) the personal blog posts which didnt reach the front page in a newsletter https://hnblogs.substack.com/ (that's curated manually every day from /newest)
From what I understand it's not every blog post, it looks to be still curated by you. Which is totally understandable, as otherwise there'd be way too many links in the email. Your newsletter helps (and I subscribe to it and am a fan), but I'd still say overall this is an unsolved problem on the modern web.
Yeah you are right, I edited my comment. It is not all of them. But I do think it is at least 50% of them. I apply an auto filter on the full list of links, and then I curate manually. In both the auto and the manual part some links are missed.
In spite it doesn't make too much sense, if you like the post an upvote is nice for the submitter.
Also an additional vote may be useful to highlight it to the mods that cherrypick some post and they have a second chance to be shown in the front page for a small time. More details by dang in https://news.ycombinator.com/item?id=11662380
I'd like something similar to 'rising' on reddit. Currently I check page 2/3 for something not yet on front page, but that depends on HN activity I guess. It is good enough right now, but about 12 hours from now (which is my preferred morning read in my time zone) only few of them come under not yet got enough attention category. By having a separate selection, they can be browsed more easily.
A two-column layout is exactly how I plan to address this in my reddit clone that I'll never, ever actually get around to writing. Not sure what to do about mobile layout, though.
I used to use an extension for Chrome that had a two-column layout as you describe and I rather liked it. However, I switched to back Firefox some years ago and I've never found a similar extension to replace it.
What about a clearly deliniated box with 5 items: 3 with 0 upvotes, 1 with <10 upvotes, and one with more than 10 but not enough to hit the frontpage.
That way there's always something interesting in the box, and everyone is encouraged to contribute to arbitrating the new posts a bit more.
edit: I guess posts that are upvoted should have a significant higher exposure, to reward them with a better chance of hitting FP, so if there's too many upvoted posts it could be 2-2-1. Oh and of course there should be a profile checkbox to turn the whole thing off, wouldn't want to impose ourselves.
I love that idea. I'm working on a HN clone in Prolog and will totally implement that. But, I think you could also make something like that for HN as a Firefox extension. Or perhaps write a simple web app that provides a RSS feed of popular and new interwoven.
Another idea would be to reuse the comment algorithm, which bumps comments straight to the top when they’re created for a very brief period to see if they get any traction.
I guess that it wouldn’t necessarily have to be the top for posts.
A decade ago my friend and I built our “better digg” with a box at the top that cycled through new submissions to help those new stories gain traction. You need a downvote or skip button.
You test out new links with very few people and expand or demote based on the results of the first couple of testers. Pretty sure a bunch of social media, youtube, etc do something similar.
I visit "New" occasionally but I find it often disappointing because it is full of flagged submissions. Many of them seem reasonable submissions (not on TV, not click bait), so I keep wondering why they are flagged. Not sure if some people flag duplicates, but finding out whether it is duplicate takes quite a while, especially on a phone.
Most of the [dead] post are just spam. Some are just off topic. If you see one that is not spam, you can vouch them, and if it is really great you can write an email to the mods. (It's a manual process, so only write about the very good post.)
There are some corner case, like people that make very simple technical post like "How to sort a list in Python using the standard sort function", and submit every day one of them. A good discussion about different sorting algorithms is nice, but a two lines post about [1 4 2 3].sort is not.
And anything political that doesn't meet the hivemind's right-leaning, libertarian political mindset. At least those who downvote and flag posts and comments, tend to effectively cancel particular views a lot.
Pretty sad state of affairs, as the hivemind remains a political bias, as new users don't have the same rights to manage visibility. Even on Reddit the Hivemind can't kill a comment completely. IMO comments/posts should never fall below 1 karma points, karma shouldn't be visible, just affect ranking, and comments only be removed by moderators, with an explanation, which community rule was broken.
This is an inaccurate view of HN. It has to be, because the people on the opposite side have an identical view, only with the high-order political bit flipped. That is, they're wrong in an exactly symmetrical way. Here's a taste—there are many more where these came from:
In reality, HN is divided on any topic that is divisive in society at large, including all the variations across the different countries represented here. That's a lot of variation.
Once you get over how both $X and anti-$X users pronounce HN to be an "echo chamber" (or "cesspool", an old favorite) controlled by their opponents, it's a bit of a puzzle how that is even possible. How can people come up with 100% opposite views of exactly the same thing?
I think these false feelings of generality are generated by the mind because the comments one dislikes make a much stronger (and more painful) impression on it than anything one likes or agrees with. The more one dislikes those, the more pain they cause, and the mind responds by creating an image, almost like an organ might secrete a membrane to protect itself. This explains both why these views are so identical (except for the high-order bit) and why they're most often posted by the people who have the strongest feelings for or against $X.
Unfortunately this creates a situation where everyone with strong passions feels like the site is dominated by their enemies and they're basically surrounded by demons when they spend time here. That's not so good for community.
In reality, HN is probably less hostile than similar places of comparable size on the internet. It's just that it's organized in a way that keeps everyone together, so you're much more likely to randomly bump into remarks that you passionately disagree with. On other sites people protect themselves from this by subscribing or following or connecting primarily with users of their own tribe, and come into contact with the other tribe only when going out on skirmishes. I wrote a mini-essay about that if anyone cares:
And yet these comments are not downvoted or dead. Just the claim of bias isn't evidence of bias. I would also confidently say, the nature of the comments you linked are very much different: I didn't merely stated "but the other side!1!!", but argued about the inherent power dynamics of HN's workings. Let me ask you, should my comment be flagged and downvoted?
The right has adapted a generalized "No U!" line of discourse, most prominent in the moral panic of "cancel culture". It's always based on the distortion of semantics and therefor false equivalency, ignoring context.
Please show me examples of actual arguments leaning politically left, which got upvoted and tolerated by the Hivemind. Please, try it yourself.
It is not both sides. There is a difference in intention and semantics.
The comments I linked to above were downvoted and dead. I unkilled them so that readers who clicked on the link (and don't have 'showdead' turned on in their profile) could read what they were linking to.
> Let me ask you, should my comment be flagged and downvoted?
If you mean https://news.ycombinator.com/item?id=25903933, of course it should! It breaks more than one of the site guidelines, e.g. the ones that ask you please not to use HN for ideological battle and please not to go on about downvotes.
I'm sorry, but the phenomenon I'm talking about could hardly be clearer: the hivemind, to use your term, has both right-leaning and left-leaning devotees, some of whom are extremely passionate; the deep irony of the situation is how much they resemble each other. Both show up a lot on HN—their presence in the comments is likely far greater than their numbers in the community, because passion makes one more likely to comment on the internet. Both deny that their side is as common or as influential as it really is, and both make out that the other side is the dominant one and that therefore HN is under its control or in its pocket.
All that is just empirical observation. Whether the psychological mechanism I'm proposing to explain it is true or not, I don't know, but the empirical observation is the important part. No one can look at these things objectively and not see the similarity, indeed the isomorphism, between them.
They probably auto-flag duplicates and known spam.
Whenever I look at the "new" page, it seems like a firehose of posts. I'd imagine there's luck involved with how many people happen to visit the "New" page before a post is pushed into the abyss. I'd also imagine there's a certain time-frame where a post is more likely to catch traction.
Flagging is entirely broken on Hacker News, which is a problem I've pointed out to the moderators before. It's basically a super-downvote and allows people to censor opinions that they don't like. Comparing with downvotes:
- Flags work on comments and posts, but you can only downvote comments.
- Flags work on replies to your comments; downvotes don't.
- A few flags will remove a comment entirely, downvotes can only make it grey.
- A few flags will remove a post entirely, and even with showdead enabled you have to see it in /new or at the last page of /hot, then choose to vouch for it to rescue it. This almost never happens. A flagged post is basically a death sentence for that post.
- You get flags at 31 karma, but can't downvote until 501.
So the answer is that flags are utterly broken and vulnerable to abuse, and /new is where it shows most strongly. Tons of posts are killed there. Controversial topics almost cannot be discussed on HN because it only takes a few people to vanish the post for good.
> Controversial topics almost cannot be discussed on HN because it only takes a few people to vanish the post for good.
Yes but that's a choice, because these topics tend to create flame war discussion, so they don't belong to HN. It doesn't mean, these topics should not be discussed. Hn is just not the right place for it.
> This almost never happens. A flagged post is basically a death sentence for that post.
I'm not sure for that. I've rescued post, and I've seen post being rescued. But it could be easier to rescue a post, I agree.
Hn works as a filter for submitted content, which works pretty well but tend to filter always the same kind of posts
This works for technical topics, but quickly falls apart in certain topics of conversation. When a user-base skews one way politically, this is a terrible feature. People are blind to the fact they're chiseling their own echo chamber.
Edit: how is downvoting political opinions (and therefore reducing the post's visibility) beneficial to anyone? It's a shallow way to make the down-voter feel good temporarily, and dissuades actual discussion of other philosophies. At the very least, political downvotes should require a response.
>Upvote to encourage more people to visit New Links on Hacker News
I experimented with the "/newest" link a while back and I decided that it was not a good use of my limited reading time. It was better to read the threads that the crowd voted up to the default HN front page or the "/best" algorithm for some older threads.
The issue is that the "/newest" would only work for me if it was a whitelist of the 50 HN users' opinions I trust.
(And nothing against you personally because I don't know you but I notice that /newest page has many more submissions by crazypython because you've submitted many articles in the last 24 hours. Do I need to see more crazypython submissions?!? I dunno.)
But I'm willing to be convinced that I should expend reading effort on the /newest page again.
heh just like search ads... "before we show you what you came here for, please review these items which benefit us"
Still a reasonable suggestion. I like the idea of making "new" items part of the default view so they get more review. If this is done, they should be visually separated from the popular stuff, just like ads need to be clear.
Is it really a problem though? When I submit a post I usually spend a little time voting on other submissions in the new queue. I'd guess that's a pretty common mode of usage.
My own submissions and the ones I see when voting on new seem to land on the front page in rough proportion to their quality.
I see more political (in some sense) posts than I really want, and not enough super-duper technical stuff - but I don't really get the impression that that arises from a lack of attention in voting stuff onto the front page.
I don't see it as a problem either. I visit HN a few times a day, when i need a break from work and want to avoid descending into the dark Twitter rabbit hole, and usually check out and vote on a couple of articles in the new links queue.
Only difference is that i come less for the technical stuff — i'm not a developer — and more for the (usually) intelligent discussions, which sometimes lead me to even better stories.
While it's true that the 'new' section is pretty under-visited, one of the reasons is definitely due to the content being much worse (on average) than the front page; it's a great way to see how effective the HN system is at showing us the better content that's been submitted.
But it's also quite often that some great posts don't manage to gain traction simply due to bad luck - I imagine the amount of people that read 'new' is around 1-3% of the userbase (I know if I'm going through pages of it, it's probably time to start focusing on work instead...), so it's nice this has been submitted.
The problem is I often don't think of clicking on the "new" link and I do understand why people may be reluctant to spend the effort of curating a couple of new entries.
So, idea:
How about everyone gets a random new submission shown on their front page, at a static spot where it's easy to ignore if you don't want to see it. This would still drastically increase the chance of something interesting catching your eye, without polluting the front page too much.
It's a randomized algorithm, but paradoxically, I think it would reduce the randomness of whether a submission gets picked up or not.
Maybe I’m crazy for suggesting this, but for a non mobile layout the single column is a waste of real estate. Maybe allow a the current UI for those who don’t want to change but have a 2 column layout for those with the space that has top and new side by side. For mobile I think the current UI is sufficient.
How about taking the current ranking formula and dividing it by the number of clicks? This should make the "new" page obsolete. I ran some simulations to analyze the new behavior: https://github.com/fdietze/downvote-scoring
Y'know what would be really easy to implement? A /random which just took you to a random post (not flagged/dead) that's been submitted in the past X hours.
Imgur's users once conducted an experiment, or a flashmob, where people hanging out at ‘new’ were supposed to upvote just anything, to show the mainpage users what it's like in ‘new’. In practice, weirdest things were floating up to main, and that night was by far the best experience I've had on the site.
There was also a regular feature called ‘Late Night Imgur’ where at night (on the US time) people upvoted horror stories, strange visions and such.
You'll pretty regularly see a story on the front page with a low double-digit score, especially during off-peak times. Point is, it doesn't take that many upvotes coming quickly to rocket a story to the front page.
I'm more frustrated at stories that get missed due to poor titles.
But that's kind of also true across all of life, so probably not something HN is going to solve anytime soon.
I'm sure it's easy to write a small script that spawns new tabs in the browser and opens several random sites from /newest. In-fact I am tempted to write a bookmarklet that does this. You could also code up something in Tampermonkey that automatically opens everything in /newest
Not sure how that will help but I can certainly say, much of the best content of HN as far as links go doesn't pass through the barrier of getting on the front page.
https://hckrnews.com sometimes catches those links but even then, it's nice to stroll through /new HN posts. I've been doing it for a while and it's around 600-1000 links (15 min time of the day)
I go through new submissions a few times a week. A couple things that make it undesirable:
1. No way to limit to certain topics/tags. That means I'm seeing philosophy, politics, news, etc.
2. No way to see a summary without clicking the link and waiting for the page to load. Reasons this is bad: Title length is limited (for good reason) so titles on HN are routinely uninformative. There's more likely to be annoying ads/paywalls/etc. inside these posts. That's a lot of overhead to devote to a single potential vote for a probably-not-good post. As a result, I just skip over most links, and that's the same as not visiting the New Links page.
It would be nice if there were some sort of advanced features, like downvoting or flagging, which could only be maintained by users who upvote at least dozen links off of the new page per week.
I’d suggest keeping the karma threshold required to earn these privileges, but just require a little bit of ongoing service to maintain them. No penalty if you are take a break from HN for a while; just spend a couple minutes voting on new and your advanced abilities will return.
hmmm.. It is always interesting to learn other's behavior. I hit the top links often. Maybe that just shows how often I'm here on HN :-) Gotta feed the dopamine monster
I don't understand why so many people are flagging this post, when it caused 8 different articles (which are ranking above it) to get the front page. If anything, it should rank above the articles it helped get to the front page.
You completely bullied the algorithm. There are like 20 new posts on the front page and no post stay at the top. Because there are too many posts being upvoted
HN should create 4 categories for posts(technical, political, non-technical & non-political, misc -- or something similar), then simultaneously display the top 10 for each category on the "New" page.
This happened to me on my phone and it was driving me crazy for a few days until I noticed the browser switched the HP for the New page in the Popular section above the bookmarks. Maybe a variant of this is happening for you.
Using https://hckrnews.com/ interface with top 20 filter. Works out pretty well, oc I'm kinda blind to less voted threads and thankful to your all voting the interesting stuff up :)
It is disappointing what gets picked up in New. Dang's second chance pool etc are a great way to address it, but the New tab upvotes prioritize controversy and "hot" topics over substantial ones. Here are my last 25 submissions, please guess which ones made it to the front page,
The Romeo File (1997) (newyorker.com)
1 point by areoform 17 hours ago | past | discuss
A Review of “So You Want Your Child to Be a Genius?” (plos.org)
2 points by areoform 2 days ago | past | discuss
Hollywood and hyper-surveillance: the story of Gorgon Stare (2019) (nature.com)
52 points by areoform 5 days ago | past | 17 comments
Should We Open Some Sealed Apollo Moon Samples? (space.com)
1 point by areoform 8 days ago | past | discuss
Flashback: Bob Dylan Performs at the 1963 March on Washington (rollingstone.com)
2 points by areoform 9 days ago | past | discuss
Onesimus (wikipedia.org)
1 point by areoform 10 days ago | past | discuss
NSFW: Someone at Osama Bin Laden's Compound Sure Loved Anime (gizmodo.com)
3 points by areoform 12 days ago | past | 1 comment
In a Murderous Frame of Mind (washingtonpost.com)
2 points by areoform 16 days ago | past
Did the NSA Help Russia Target Dudayev? (1997) (free.fr)
1 point by areoform 19 days ago | past
IBM and the Holocaust (wikipedia.org)
55 points by areoform 19 days ago | past | 11 comments
Kettering Bug (wikipedia.org)
2 points by areoform 22 days ago | past
Mason Jar (wikipedia.org)
3 points by areoform 22 days ago | past
An Economic Analysis of Liquidity-Saving Mechanisms [pdf] (newyorkfed.org)
1 point by areoform 22 days ago | past
Computers in Spaceflight: The NASA Experience. Simulations and Image Processing (nasa.gov)
1 point by areoform 24 days ago | past
[dupe] Nintendo's use of surveillance and private investigators to target modders (twitter.com/eclipse_tt)
264 points by areoform 26 days ago | past | 112 comments
Pan-American Highway (wikipedia.org)
4 points by areoform 26 days ago | past
Sonic Hedgehog (Genetics) (wikipedia.org)
1 point by areoform 28 days ago | past
When California Went to War over Eggs (smithsonianmag.com)
2 points by areoform 28 days ago | past
What Happens to All Those Letters Sent to Santa? (smithsonianmag.com)
4 points by areoform 29 days ago | past
Paul Baran and the Origins of the Internet (rand.org)
4 points by areoform 30 days ago | past
Elon Musk: The Architect of Tomorrow (rollingstone.com)
15 points by areoform 32 days ago | past | 17 comments
FSB Officer Inadvertently Confesses Murder Plot to Navalny (bellingcat.com)
1 point by areoform 34 days ago | past
Can You Hypnotize a Chicken? (smithsonianmag.com)
2 points by areoform 34 days ago | past
Designing with Fluid Type Scales (utopia.fyi)
1 point by areoform 35 days ago | past
I Stole Gram Parsons’ Body (loudersound.com)
1 point by areoform 35 days ago | past
Some of the most interesting submissions never saw the light of day. The "I Stole Gram Parsons' Body" story is bizarre in a very HN way. An economic analysis of liquidity saving mechanisms explains how the modern economic payment system works. The Romeo File goes into an interesting facet of espionage history where the stasi targeted secretaries with "Romeos" to send information to the USSR. A practice that has reverberations to this day. When California Went to War over Eggs is a bizarre story about protein at the early west.
However, none of these received any interest. Which is frustrating. I try to share interesting things I find on HN, but it feels horrible to find something cool that you wish to share with the world and watch it slip into the endless void.
Unless, the submission generates outrage. Such as the Nintendo story. Or, IBM and the holocaust. Or, the Gorgon Stare. That's what makes it to the front page. It is soul crushing to know that interesting stories about the time there was a very literal war over eggs won't get as much attention as these clickbaity things. Which leads to a natural conclusion, the reason why the internet is so crap in general is because people like things that make them rage or feel strong emotions, and they reward it.
Dang's inbox is always open, but it feels needy and depressing to email him repeatedly "Hey, I submitted something great, have a look?" I wish there was some way to fix this problem.
I'm sorry but HN has gotten really toxic over the years. Flagging and censorship is a big concern here as there is an asymmetric balance of power that is clearly being abused.
Somebody doesn't agree with me politically? flagged & downvoted.
Somebody is a potential competitor? flagged & downvoted.
Somebody drives a Ferrari and I don't? flagged, downvoted (recently, somebody on HN tried to doxx me with no moderator taking action)
I miss the days when there were really good discussions, while those are still around, HN increasingly feels like a competition of "I am smart because X so Y". There appears to be a cabal especially around specific topics like Bitcoin or their affiliated companies.
edit: heh case in point. now all of my other comments are being brigaded & downvoted. Lots of bitterness and negative vibes here that is only going to impede growth. I will pray for you.
Yes, there is sometimes an excessive pedantry that is used to attack an FOSS or some new project on Show HN or new thoughts. It's impossible to placate this small group, they decide what gets upvoted and what doesn't.
It is the same problem on Reddit too. Anybody can create multiple accounts or buy aged ones, use different proxies to downvote. On the server side, what can you really do to fight this?
Fingerprinting, IP address intelligence, account age, points, user activity that will be compared to the normal distribution of most legitimate user activity (I doubt this is used efficiently).
edit: Case example, within minutes of posting my parent comment, its downvoted to oblivion and they are probably going through all of my comment history to do the same. It's really churlish and parochial.
Maybe requiring a comment to downvote? Like you click the downvote and instead of it actually downvoting, it attaches the downvote to the comment box and requires you to leave a reply justifying your downvote.
1. popular
2. new
3. popular
4. new
This injects a lot of noise into the front page, because most of new is noise, but it helps.
Another option is to front-load popular stories, then have 10-20 "new" items at the bottom of the page. That way the front page remains highly salient, but anyone who merely scrolls more will also see a small chunk of new. So 1-30 popular (as it is now), then 30-45 new (making the frontpage longer).