Ask PG: Sub-HNs as a solution to cluttering of HN front-page?

The reason I resist doing something like this is that it's sort of like dealing with getting fat by buying bigger pants. I'd be willing to introduce categories to make HN a better site, but I wouldn't want to do it as a way of solving the problem of the frontpage getting filled up with crap. I'd rather attack that problem directly.

Also, it seems to me the problems with crap posts on the frontpage are not as old as 6 months. In fact they're so recent that I'm by no means sure that it's not just random variation. The frontpage was ok today.

I think as HN has become more mainstream (in spite of its focus), I surmise the diversity of the audience has increased too. I like the idea of a self-correcting community (just as the markets should correct themselves in a capitalist economy) but from time to time outside intervention is needed when they stray too far.

Perhaps I should rephrase my OP. Sub-HNs not to solve the front-page problem but as a way to segregate related posts.

Ask HN is already a sub-HN. Jobs is the next logical one. I don't think we need create-your-own like Reddit, but a few major categories would be awesome. The "ask" page has already become my favorite part of HN.

I'd be willing to introduce categories to make HN a better site, but I wouldn't want to do it as a way of solving the problem of the frontpage getting filled up with crap.

I think separating temporal from non-temporal would be useful. The latest news about the iPhone and a wikipedia article about deep-sea crustaceans are both interesting, but the former is likely to be irrelevant after a few days while the latter is likely to be just as interesting next year as it is right now.

If you take the non-temporal articles off /news and take the temporal articles off /best, it would probably clean up both quite significantly.

If there was one simple thing that could improve the quality of HN, it would be a new section of all recent moderator actions whether they be automatic or manual: bans, privilege removal, etc.

People would know the right things to do and why (by seeing the wrong things), and they also see the implementation of the guidelines.

The only issue is that protests may be raised, but the benefit is transparency. Also, something good may come of the protests anyhow.

I think thats the wrong direction, the community should be in a self-policing equilibrium with a demphasis on moderator action. If moderator action has to become frequent enough to have a news feed on it, then the community isn't being given enough opportunities to police itself.

Downvoting above a certain karma on comments has worked wonders as Hacker news has great discussions. It gives an easy way for the community to state they value insight over snarky observation or trolling. I think the same feature needs to simply be added to submissions as well. It will allow the community to inform members that they do not value resubmissions of things already on the front page or linkbaiting to old topics, without requiring moderator action.

It'd be interesting to see an up/down tally on karma for any given comment score, like at http://www.urbandictionary.com/

You'd see the controversy or variability around a comment.

On Digg, I think submissions reaching some threshold of buries get killed, even of already obtaining lots of digs.

>a way of solving the problem of the frontpage getting filled up with crap

add downvoting to submissions and the community will take care of it

When I used Reddit, some people said they downvoted submissions on niche topics they weren't personally interested in, e.g. a Rubyist downvoting Haskell articles.

There are enough Haskell enthusiasts here that I don't think Haskell submissions would suffer, but who knows what niche topics would?

I used to be on Reddit in the Lisp days (before it had many users). The day that they introduced the `my.reddit` was when the quality of posts massively fell. The recommended behaviour at that point was to try and train your recommendation engine by voting for things you liked and down vote things you didn't. The `my.reddit` then showed your own personal homepage. From that point votes no longer became a quality control as people would vote without thinking based on a headline.

I am all for some way of adding meta-data to links so a user can filter based on them. But you have to be really careful to not turn votes in to a meaningless action and keep them as a way of voting on quality rather than opinion.

(ps. I stopped using Reddit shortly after that change and as such haven't seen/used it recently and don't know how the sub-reddits are doing to manage that situation. Also it was one of many changes that negatively effected signal-noise but one I always feel was the turning point)

Nowadays your best bet is to find a smaller subreddit with a tight focus, essentially starting a new community. The main community has major issues with quality control, but the splintered subgroups are generally much better overall. Really, it does seem like small size is the best quality control, unfortunately.

I don''t think HN would become another reddit. Downvoting would be only available above a certain karma threshold and its worked well for comments, which havent become reddit like.

The alternative could be make flagged have a displayable number that acts as downvotes for position but doesnt effect karma. Display the number of people who have flagged an article. It's like downvoting but more explicitly clear only for things that are objectionable on some other ground than you disagree.

I use flagging as downvoting on posts I see as off-topic. It's the tool I have...

Add a displayable number of how many people have flagged something and have this negatively impact page ranking and I think thats a solution. Down-voting with a more explicit label discouraging down-voting for disagreeing with a topic.

I'd like to see tagging, with a page and corresponding RSS feed for each tag. I usually read HN via RSS, so maybe I don't have a pony in this fight (is that the expression?), but I suspect most folks who visit the front page are there for things that could plausibly be tagged "news", and those complaining about clutter want to avoid things that would be tagged "humor" or "rant".

I think having non-exclusive, optional tags would be much less of a leap than having to assign each post to a single category. Perhaps all users with karma above a cerrtain threshold could have authority to remove a tag from a post, ensuring that the "news" page would only contain things that are clearly considered to be news. Or there could be a "tagflag" link under each tag.

The biggest problem I'd expect to see with this scheme is that editors would have to add tags to items missing them. But important news items are typically submitted multiple times. So perhaps the tags from multiple submissions could be aggregated.

Those don't seem like categories to me. It looks more like you're talking about two continuous dimensions upon which all posts fall: ephemerality (or, conversely, timelessness), and community-orientation (or, conversely, xenophilia.)

A "learn" post is just a post that will still be good a while from now. A "discussion" post is just a post closely related to the goings-on around here, to the point where all the information related to it can be included and it doesn't need a link.

Couldn't these be separate voting axes? Voting on ephemerality would affect a post's time-factor in its calculated ranking (ephemeral posts would drop off the front page quickly, while timeless ones would last a while); and community-orientation would be a setting each user could have in their profile (a slider for "I like more/less community in my HN.")

The problem isn't the organization, it's the depth. This site is based around news items and self-submissions that act like news items. 25-48 hours after submission, no one will bother commenting on an item.

There's no room for depth of discussion in the current format, and so it gets repetitive in both news items and comments. This leads to the perception HN quality is degrading.

I think there's room to innovate in the management and threading of discussions, but it's more complicated than splitting into categories.

I usually start my visits to HN with my threads page. I frequently answer comments well after articles leave the first page but only on threads I got involved. There are times I go back to the root discussion to read more, if I see some reason to assume much has changed after my last visit.

Only when I see nothing new there is that I move to the home, and then to the "new" page, where I frequently do some flagging.

If something has led me to perceive the quality as degrading, it has been a surge in articles that are much more "news" than "hacker".

The most obvious problem is that the "gratifies one's intellectual curiosity" clause in the guidelines is too broad and, therefore, impossible for certain people to internalize. That's why there are even long-time users who somehow believe that pictures of girls in t-shirts are appropriate (http://news.ycombinator.com/item?id=1367047). That misconception creates a broken windows environment where people start racing downwards.

I don't have a precise solution. In my experience, zero tolerance for off-topic subjects like political discussions is the best way to keep online communities healthy. As the saying goes, give an inch and they'll take a mile. Somehow making the "gratifies one's intellectual curiosity" clause stricter would go a long way to improving quality overall.

I like the idea, but the reality is that I'd just read all three out of bordeom or inability to not click stuff. So it would end up being the same.

If you want to make a new site with categories, you are welcome to, of course. If people like it, they will find it. (Remember when there was only Slashdot? And then Reddit? I don't read either of those anymore...)

Making a new site is the solution. There are too many people who like things the way they are here, for better or for worse. HN is frozen; it doesn't adapt.

I myself may be guilty of "diluting" the site as I'm not from computer science or engineering by background, I'm just interested as a hobby, and I joined around a year ago. Having said that, may I ask whether the point hurdle to upvoting is static? If so, an easy way to keep the quality constant may be to mark it up every year in order to account for point cumulation over time or to introduce a scheme similar to the rating of articles (i.e. points divided by time). The "natives" will have been around for a long time and/or will have a higher point/time ratio while interested bystanders/observers like myself will not. Why am I harming myself in this way? Because a few interested bystanders, not synchronized, will do no harm but add more and possibly synchronization and they may even harm themselves unintentionally (not to mention the "natives").

What if randomly the front page changed to http://news.ycombinator.com/newest? PG may know better from statistics, but my guess is that most people never go to that page to upvote newest stories and the few people who does, are those who shape the front-page.

Maybe it's a dumb idea, but that's what I was thinking.

Or make it a split screen on option, that you can show hide new on the right side. Would fit nicely on a wide screen. Then you could see if anything catches your eye in new, without have to consciously know you were doing it and it would introduce it to people who haven't used it before.

I say an editor is enough to ban posts about facebook selling our mothers or apple killing its employees in vietnam.

Hackers don't like gossip and HN is not a yellow tabloid.

No need to diversify.

Better yet, give moderator powers to the top ten users (or those who volunteer) and watch them closely just in case they abuse their powers. If three out of ten consider some articles yellowish they can ban them even if they are in the front page already.

Too much work imho to constantly moderate/edit content.

I disagree. I really don't see what is broken with Hacker News (and I've been here for a while too).

I'm not saying it's broken. Just saying some of the cracks could be fixed in time.

To me, the thing that makes Hacker News, well, Hacker News is that the site is really simple, from a technology contraption point of view. It is controlled and shaped and molded (and whatever other verb you want to use) by us users (counting PG as, well, prime user?) and not shaped so much by arbitrary software constraints such as binning topics (putting in buckets). The touch to tuning this site has been light, with such adjustments as flagging and tuning the rise and fall of topics based on a variety of heuristics. Some tuning has been undone, such as color coding "most valued posters"

If anything, I would say if we were to change the site, the next experiment would be to put a greater collective downweight on titles or source sites if there is a strong local temporal grouping (glut within a timeframe) of those tags/keywords posted. Example: lots of facebook posts at one time then facebook posts drop off the front page a little faster than if there was only a small number of facebook posts. But even that, I'm not sold on this.

But I have to agree that if you are not interested in facebook and apple, half the post are useless. Plus, if you already know 5% of linux, 1/4 of the post are useless since it's usually a blog post explaining how cat works. The rest is interesting to me (of course, not the How to start a startup repeating what as already been said 10 millions of time, and why block are cool in ruby)

The interesting thing is that this site should be a reflection of those participating in it. That means that either the audience is growing in a direction that is not consistent with the original participants or that there is some other force at play trying to alter the content of the site.

The first thing to do would be to figure out which of those two scenarios is at play and address that. If the tastes of the users has changed (or an influx of new users that have different tastes has emerged) then I'm not really sure what the answer is.

Unlike some other people, I agree with what you're saying about how HN has changed. I guess the fundamental question is whether to adapt HN to the userbase changes or to try to keep HN from changing at all (probably with stronger moderation). My preference is for the latter (of course, I'm not a mod, so that's easy for me to say), and I do think that's kind of how it's been handled so far...but I think the former isn't a terrible option if the mods become overwhelmed.

I hate categories, hn is perfectly fine as a "surprise me" portal

where else I can post an article on Caligula's giant ships http://news.ycombinator.com/item?id=1370063

or piping cats http://news.ycombinator.com/item?id=1116085

Reddit.

thanks, i'll give it a try

I think it's a social phenomenon, not a technical one. Clay Shirky wrote the standard work on what happens when you try to fix social problems with technical ones: http://www.shirky.com/writings/group_enemy.html

A better solution may be a new form of Erlang flood. Perhaps with some added technical incentive, say, a new list of users most hospitable to "classic" content.

That's a shit idea, but I'd much rather see PG work to better the community, than try throwing tech at it and watching the social forces mitigate his efforts.

What about folder-like threads instead of sub-reddits that would group threads with identical keywords like Google, Apple, iPad, etc.

if this were done, I'd suggest separate "learn" sub-HNs for geek stuff and business stuff. (github vs marketing)

Anything start-up related should be in the main HN, all other items in a sub-HN.