HN Frontpage ranked using only votes from accounts over a year old

http://news.ycombinator.com/classic

by pg • 17 years ago

300 107 17 years ago

news.ycombinator.com

Very fascinating! Not drastically different but shows an interesting difference in voting. Perhaps you'll provide some sort of basic controls for us to tweak our personal homepages along these lines?

Zak• 17 years ago

What would be really cool is a form that took at arbitrary[0] Arc expression as an argument to the sort function.

[0] but restricted to a certain set of allowable symbols

eru2• 17 years ago

The function should also be provable fast.

pg3• 17 years ago

I wanted to see if there had been any visible decrease in quality. Doesn't look like it. There's surprisingly little difference between this and the regular frontpage.

DanielBMarkham• 17 years ago

I'd be more curious to see the difference on a highly-emotional news day -- election, war, disease -- that kind of thing. Would the old guard stick to tech-related material? Or are they as vulnerable to highly-charged stories as the noobs?

In all honesty, there really isn't much going on in the way of external-to-hacking events in the world right now.

einarvollset2• 17 years ago

You mean like Obama deciding to fight the release of pictures of prisoner abuse by US Forces?

Sri Lanka's government shelling a refugee camp containing rebel fighters despite a no artillery pledge?

geuis• 17 years ago

I'm not sure why, but my feeling is that I am personally less happy with the content a) being submitted and b) making it to the front page than I was even 6 months ago.

Maybe since I'm not a newbie on HN anymore I'm remembering through rose-tinted glasses and yelling at the kids to get off my lawn.

The things I care about are programming tips, new software, javascript/python/ruby/etc specific articles, new startup companies. There seems to be a lot less of this kind of content and more general tech news, which I can read at any number of other sites.

I don't care about business stories(other than those related to startups) and rumors Techcrunch has started about something.

On the plus side, the discussions that happen in Comments are definitely still good. I would be great if there was a way to get easier access to the comments than the ity-bity link we have now.

joseakle2• 17 years ago

Perhaps some other kind of weighing system might work.

Votes from high karma users could boost stories more than low karma users. Is there any way to downvote a story?

timr• 17 years ago

Looking at stories at one moment, on one day, is probably not capturing the whole story. It would be interesting to see something like the RMS difference in story ranking between the two menthods over a long period of time. If the two methods are truly equivalent, that number should tend to zero.

rgoddard• 17 years ago

I think this is mostly due to perception. I have experienced a similar change in perspective, but I still come here just as often and spend about the same amount of time reading the articles. But I now use this site as a baseline for quality comparisons. Rather then being wowed by all the cool articles here, I have come to expect this level of quality.Other sites have become less interesting to me as a result.

edw519• 17 years ago

I've been around long enough to have noticed several of these cycles. I could go for weeks seeing nothing interesting, then go nuts for a couple of days reading and posting. I don't know if it's me or hn, but I simply describe it as the natural ebb and flow of an organic entity.

Come with an open mind and have fun if there's interesting stuff that day. Get back to work if there's not. Either way, you win.

iamwil• 17 years ago

Like all news, you get lulls. Sometimes, there really isn't much going on, but we will it with things all the same.

anc2020• 17 years ago

> On the plus side, the discussions that happen in Comments are definitely still good.

Maybe its me, but I've found a lot more comments within the past fortnight or so grating on me than previously. Remarkably, every time I checked the submitter, they'd been on Hacker News for about a year and a half at least. And by grating I basically mean rude and unhelpful.

10ren6• 17 years ago

A few times I've compared the same programming article here and on proggit - and the later always had deeper comments. Maybe it's not surprising, since that's the focus of proggit, but only one of several here.

I've also twice noticed the same usernames on reddit and HN - and their comments on reddit are more helpful and intelligent than here... Though not a conclusive sample size, it does make me think that reddit's concentration of communities (with subreddits), and the ability to easily skip bad comments (with collapsing [-]) seems to be working.

But I find the stories submitted here are better than on reddit.

10ren6• 17 years ago

I find that there's usually 1-2 stories per day that are significant for what I am seeking on that day. So, for me, it's not surprising if that doesn't always happen, since there are so many factors involved.

randomwalker2• 17 years ago

In fact, the two views are almost identical -- I wonder if it is simply because there aren't that many votes from accounts less than a year old. Could you share what fraction of front-page votes came from new voters? Thanks.

_cbsz• 17 years ago

Maybe the new accounts hate almost everything that shows up. Remember, we can't downvote, so lack of vote is the way to express disapproval. (60-odd days old here, would be on a downmodding rampage if allowed)

EDIT: By the way, the lack of downvoting means that posting any article that isn't so off-topic as to get flagged and removed is free karma. I've been working hard to get to the 100 karma so I can downvote and I'm still only at 78, and I'm thinking maybe it's because I don't submit articles.

icey2• 17 years ago

For things like this, it'd be cool to have a checkbox on our profile pages that lets us opt to use the classic or standard view as the default list.

It's not that different a list today, but over time it might be nice to have the option.

bouncingsoul• 17 years ago

Doesn't surprise me.

I don't know why some people here rag on new users so hard: everytime I've encountered a particularly low quality submission or comment it was from a 100+ day account.

amichail• 17 years ago

How do you measure quality?

stcredzero4• 17 years ago

http://en.wikipedia.org/wiki/Zen_and_the_Art_of_Motorcycle_M...

prakash• 17 years ago

What inspired you to do this?

As a quick & dirty test, can I see "HN Frontpage ranked using only votes from accounts over 2 years old"?

Can you turn this into a feature, with a user-specified time period? -- would be an awesome filter to view HN by and to see, if, over a period of time there is any difference in quality?

Thanks very much!

colins_pride2• 17 years ago

Yeah, when you think about it the general form is also true: it doesn't make a lot of sense to hard-code the sorting algos

andreyf• 17 years ago

I wanted to see if there had been any visible decrease in quality. Doesn't look like it.

I'm not sure if this measure means much... with no downvotes, if quality goes down, "classic users" will simply vote less often. That would be a decently interesting statistic: (page views)/(story votes) over the lifetime of older accounts. Not sure if that data is still available, though.

rw• 17 years ago

I've been here long enough... since February or so I've drastically decreased my voting participation. I perceive that my votes won't counteract the effects of the (well-intentioned) flood of newcomers.

pg3• 17 years ago

Actually one of the surprising things is the huge percentage of votes cast by old-timers. For frontpage articles it is never less than 40% and sometimes as much as 80%.

Tichy• 17 years ago

But it seems as if the new users vote on similar things as the old users, that is the interesting thing?

Although it might be fun to create HN/postmodern where only votes from accounts less than a year old are counted, and compare that to classic?

comatose_kid• 17 years ago

This makes the assumption that 'senior' members' voting habits haven't degraded over time. Say a bad post hits the front page - would this subtly influence future votes (regardless of how old someone's account is), resulting in more bad posts making the front page?

On a semi-related note, it would be neat to keep a history of the front page at regularly sampled intervals (say every year, starting at year 1) and see how the two measurements diverge over time.

dfranke• 17 years ago

How does it look if you set the cutoff to before the move from "Startup News" to "Hacker News"?

ftse• 17 years ago

Seems to me there are more Reddit-type comments submitted lately. It is something I have been very conscious of, much stronger than a hunch or gut feeling. I'm curious as to why you tried to measure this - do you still suspect a drop in the mean quality of the HN clientele despite what the /classic page might suggest?

epi0Bauqu• 17 years ago

The decrease in quality could be due to less good stories on the new page and not newer people upvoting. That is, newer people may be just as good at upvoting, but the real issue is that so few people upvote from the new page. If true, this change would further the problem because even less real votes from the new page would be occurring.

To test this possibility, randomly reserve some of the places on the front page for new stories (perhaps in the bottom half). Then more people would be looking at and potentially upvoting the good stories from the new page.

mlLK4• 17 years ago

Considering, [http://news.ycombinator.com/item?id=231209] are you still evaluating what makes the front-page according to this algorithm?

   (p - 1) / (t + 2)^1.5

It reminds me of the distance formula in a vacuum that evaluates [or rates] a submission as a ratio according to it's position out of 210 [basically any story with a non-zero karma value (p - 1) that evaluates <= 210 has made the front-page] I guess the gist of what I'm getting at is, how are you generating these lists? Are they all extended from this algorithm and filtered according to a condition? Have you ever generated a list that doesn't extend this algorithm?

[EDIT] Ok, so while I'm here, in this meta-HN of a thread, I'm going to ask a question that might seem like absolute bat-shit:

What is the running-total (summation) of all our Karma as users of HN?

mlLK4• 17 years ago

http://news.ycombinator.com/item?id=609200

greendestiny2• 17 years ago

What about articles submitted by over 1 year olds? A quick look of the front page shows we'd lose articles by: vorador, blasdel, kqr2, lrm242, ontilt, vijayr, markup, charlesmount and figured. On the whole those are pretty good articles, but a classic submits would be interesting to compare with.

10ren6• 17 years ago

I think it's the recent comments - their shallow attacks reduce the perceived quality of the submissions.

Recent comments tend not to collaborate with the article - they aren't filling in gaps, adding further instances, extending the concept; discussing where the article is clear or unclear; correct or incorrect. And they are neither respectful nor kindly. They don't build on, or with, the article to create something better.

Instead, comments compete with the article - they find spelling errors, logic errors, or denounce the article in a shallow, general, global and vague way, often labeling it with connotational language. They aren't trying to build value, but destroy it.

The only criticism I enjoy reading is what you'd get from a kind and wise mentor. If someone hates an article, I'd prefer it if they just ignored it.

But I can't make them do that - so, recently I've taken to ignoring the comments here, and going direct to the article to make up my own mind.

stcredzero4• 17 years ago

I think I have recently been wrongly profiled by someone who shares this view. I mentioned that certain patterns of behavior in internet consumers resemble those encountered in dating sites and used car dealerships. I know there have been economic studies of this. (Couldn't find it with Google, though.) However, I was attacked as if I was merely trying to associate Apple Computer with dating sites and used car salesmen.

In my two decades online, I've found that those who attack others to try and "clean up" a news group actually end up trashing it with noise and junk. (In fact, I have been guilty of this as well.) Much better for everyone to act respectfully.

herdrick• 17 years ago

But front page voters influence each other. How about doing the same thing for the 'new' page? That's a cleaner test.

sayrer• 17 years ago

the n00bs are making us dumber

scorpion032• 17 years ago

When U [said](http://www.paulgraham.com/hackernews.html), need to "weigh the votes" I thought you said considering votes from people with more than a certain points as more, rather than what now seems to be "older peoples votes only"

pg3• 17 years ago

If we're going to keep this around, the threshold ought to be set at a particular time instead of 1 year back from whenever the request is made. So I changed it to use a fixed threshold set just before the infamous TechCrunch mention in March 2008.

sant0sk12• 17 years ago

Love the fact that the current "review our app" submission is notably higher on the "classic" view. These are some of the most interesting/important posts imo, and it appears other "older" HN users feel the same.

trickjarrett3• 17 years ago

I'm a huge supporter of the 'review' posts, it's one of my favorite things to do on here :)

kbrower• 17 years ago

ditto, these posts are the main reason I visit the new section

randomwalker2• 17 years ago

Am I missing something? There are two review our app submissions; one of them is #3 in both views, and the other is #17 in regular and #16 in classic. Perhaps you confused the second review submission in regular with the first in classic?

sant0sk12• 17 years ago

Your analysis was 17 minutes after mine. Things change.

kineticac• 17 years ago

I posted a review our app post, and within a few minutes it was pushed back to the next page, we only got a few comments while it was on the front, and then it's lost. Kind of sucks =( It could also just be that the app didn't have that much to talk about

johns• 17 years ago

Can we get an RSS feed for this filter?

weegee• 17 years ago

as an aside, how many points does a user need to be able to vote an article or coment down? (Interesting that this comment will be voted down by people who have that ability)

trickjarrett3• 17 years ago

I believe it is 50.

alexfarran• 17 years ago

It was 50. Then it went up to 100.

icey2• 17 years ago

You can only flag an article. I'm not sure what the current threshold is for down-voting a comment; it changes over time.

I'm not sure why you believe your comment would get voted down.

tdonia• 17 years ago

100

evilneanderthal• 17 years ago

101, actually.

At 100 you can't, at 101 you can.

I know because I just hit 101 =]

spkthed• 17 years ago

It's more than 56 :)

lucumo• 17 years ago

Another aside, what karma limits are there overall?

tokenadult• 17 years ago

Definitely worth adding the "classic" link to the lists page.

http://news.ycombinator.com/lists

And, yes, as a member of the less-than-one-year-old group, I'm encouraged to see that classic tastes from when I first started lurking here are still reflected in current voting up to the main page.

quizbiz2• 17 years ago

I had no idea that page existed. Would be great if it could be added to the navigation. Please...

ivank• 17 years ago

It's at the bottom-left, "Lists"

mlLK4• 17 years ago

I'm definitely off-topic but I think this raises an interesting idea for generating different views of HN's front-page in statistical terms. I think it would be cool to generate a front-page that evaluates according to a simple comparison like, select stories that users have upvoted where karma-accumulated to days-alive is greater than 1. Even more interesting, would be to generate a front-page according to some interval of values [or range of karma].

thalur• 17 years ago

How about ranking by how other users who have a similar voting pattern to the current user vote on an article? So, for example, when the page is rendered for user x, each user who has upvoted an article has their vote multiplied by the similarity between their voting pattern and user x's voting pattern before sorting with the current algorithm. Each user then gets a semi-personalised view which, hopefully, is to their taste.

jnovek• 17 years ago

It seems likely to me that this is just because "old timers" are seeing the same data as newbies. How many folks look beyond the first hundred posts, or even the front page for stuff to vote up? It makes sense that accounts older than a year will just slightly reshuffle the top 20 posts or so.

EDIT: It would be interesting to see this comparison in an alternate reality where we could downvote articles. I would bet that you would see more variation between newer and older accounts.

bradgessler• 17 years ago

It would be interesting if all of the "new" stuff was scoped to showing "new" stuff submitted by other people who have accounts approximately as old as your own (maybe within a month?).

This would limit "new" items to your hacker news "start class", which may make the content more relevant.

SapphireSun2• 17 years ago

I'm not so sure about that. It depends on how it is implemented. If the older users see the classic page by default, then you will see much more of a divergence.

Xichekolas• 17 years ago

What would be really interesting is if the submission's current rank (at the time a vote was cast) was recorded with the vote itself.

Of course, that would change the entire way votes are recorded (most likely), but it would allow us to see how far past the front page people actually look.

frosty• 17 years ago

i actually wanted to use HN from RSS reader but so many stories get pushed that it was practically useless. So i created a filter which pushes stories to rss only when they gain 20 votes or more before being pushed out of front-page:

http://hackernewsrss.appspot.com/downloadrss

it has around 60% less articles. add to ur feed readers if you find it useful.

TweedHeads• 17 years ago

Most voting systems can be gamed. Interesting to see if this approach will reduce the PR submarines on the front page.

Payperpost, astroturfing and mob-voting are the worse diseases of news aggregators.

sosuke• 17 years ago

Anyone else happily surprised their vote counts on the "classic" version? :)

einarvollset2• 17 years ago

I wonder what would happen if you did a diff between the two - a front page of only the stuff new people liked..

wallflower• 17 years ago

I personally like the diversification from pure software hacking to hardware and personal/mind hacking.

arjunlall• 17 years ago

Im curious, what percent of user accounts are over a year old vs not?

Also, do accounts with higher karma boost up a story more?

Im wondering if the reason why the two lists are very similar is that the old users inherently have a high weight (in terms of number of accounts and karma power) on the normal page already.

kleevr2• 17 years ago

As a simple heuristic, it's an elegant solution.

Could the front page 'merit' be more or less ~accurately~ described via continuous function?

for (User user : item.up_voters) merit += user.life_span_in_months/12;

for (User user : item.down_voters) merit -= user.life_span_in_months/12;

kleevr2• 17 years ago

Or, perhaps you could define 'merit' as continously compounding interest with their average karma density (votes/items).

Taking the interest forumula "Pe^(tr)", and substituting might yield something like:

P = 1, -1, it is the single up/down vote

t = user life span in years

r = karma/submissions (karma_density)

for (user:item.voters) item.merit += (sign_of_vote)e^((user.life_span/12)(user.karma/user.submissions))

rythie• 17 years ago

I can vote in 40 days time! not that's it's a big deal I mostly vote on comments not stories.

sah• 17 years ago

The differences are hard to notice, but to me the /classic page is subtly better. Things I'm personally interested in ("Fibers & Cooperative Scheduling in Ruby") are mostly voted a couple places higher, and things that bored me ("Steve Blank: Gravity Will be Turned Off") are a few places lower.

justlearning• 17 years ago

@pg: I understand the need for this.

But here's my standing.I have been "loitering" here since the early days. I must have registered with atleast four userids..AND used always mess up my password.

Most recent one - even the openid auth - didn't work for a while(it works now!). I created the one I am using now couple of months back only to be treated as an outcast.

Is there a way to synch my accounts? (even a forget password feature would be helpful)

SapphireSun2• 17 years ago

I have a feeling that if you open up a way to sync accounts, trolls would find ways to exploit it very quickly. Perhaps it would work if there was not automated, but required an admin to authorize the change. However, I'm pretty sure that unless this is rarely used, they would not want to deal with the number of requests.

quizbiz2• 17 years ago

It's interesting or me to note the changes in rankings that have occurred over the day while this article just remains leaps and bounds above anything else.

triplefox• 17 years ago

If this view persists, the entire site's direction is going to change because some proportion will upvote only from "classic" and others only from "main."

volida• 17 years ago

personally I rarely vote. Maybe other older members act similarly.

adrianwaj• 17 years ago

You can view it as bookmarking an item with the upvote button instead.

Skeuomorph• 17 years ago

"Konami Code Sites", ranked 6th on current home as I write this, isn't even on the first page in the 'classic' view.

rapind• 17 years ago

This is great. Thank you.

HN Frontpage ranked using only votes from accounts over a year old

Related Stories