Ask PG: Please stop the opaque voting policies
11 comments
If nothing is done, there will be problems. If different things are experimented with, there will still be problems. There will always be problems -- no matter what -- for this sort of site when voting is involved. Nothing wrong with not liking the changes, but unless you're going to recommend alternatives, the post doesn't help much. If he does nothing, there are just different problems that different people will be unhappy about. I think experimenting is better than doing nothing, even if it means learning the hard way along the way.
I already replied, but I think this warrants a second reply:
I would already be much happier if the voting policies were _known_, and I'll tell you why:
For example, I've frequently seen the idea that voting power should be based on some kind of average score of some recent comments. In the current bandwagon voting climate, this would mean that the guy who "wants to put bullets into the bad guys" (recent thread, +40 points) has more voting power than most other people here (a lot of _very_ informative comments linger at 1 or 2 points).
If the voting policies were known, I could make an educated decision whether I still want to be part of this site.
I would already be much happier if the voting policies were _known_, and I'll tell you why:
For example, I've frequently seen the idea that voting power should be based on some kind of average score of some recent comments. In the current bandwagon voting climate, this would mean that the guy who "wants to put bullets into the bad guys" (recent thread, +40 points) has more voting power than most other people here (a lot of _very_ informative comments linger at 1 or 2 points).
If the voting policies were known, I could make an educated decision whether I still want to be part of this site.
On the other hand, if the voting policies were known, people could more efficiently game the system.
If the voting policy is unknown (and PG is committed to keeping it unknown), perhaps fewer people will worry about it's functional form, fewer people will care about what their vote multiplier happens to be, and more people would get on with the incredibly simple business of submitting interesting links and voting up what they find interesting. Perhaps.
You should leave if the quality of the content deteriorates, not based on some background metric for an arbitrary scrip!
If the voting policy is unknown (and PG is committed to keeping it unknown), perhaps fewer people will worry about it's functional form, fewer people will care about what their vote multiplier happens to be, and more people would get on with the incredibly simple business of submitting interesting links and voting up what they find interesting. Perhaps.
You should leave if the quality of the content deteriorates, not based on some background metric for an arbitrary scrip!
As I recall from past discussions, pg has indicated that his algorithm discounts outliers. So someone who has one 40+ comment and a string of comments rated 1 or 0 will not see the benefit of the outlier.
A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3).
A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3).
> A solution to piling on, if it persists, may simply be placing a hard limit on the maximum score a comment can garner. Slashdot and Kuro5hin both max out comments (Slashdot at 5, Kuro5hin used to max out at 5 but that was later changed to 3).
The problem being that sometimes there are posts that deserve well above the limit for their insight/knowledge. A logarithmic scale might be better. Up to 5 points there is a 1-vote-to-1-point correlation, after that it takes an increasing number of actual votes to increment the comment to the next point (i.e. 10 votes to get from 5 to 6 points, 15 more votes to get to 7 point, etc)
That way a comment that is able to get a massive amount of votes gets 'rewarded' for it above a comment that is only able to get 5 votes.
The problem being that sometimes there are posts that deserve well above the limit for their insight/knowledge. A logarithmic scale might be better. Up to 5 points there is a 1-vote-to-1-point correlation, after that it takes an increasing number of actual votes to increment the comment to the next point (i.e. 10 votes to get from 5 to 6 points, 15 more votes to get to 7 point, etc)
That way a comment that is able to get a massive amount of votes gets 'rewarded' for it above a comment that is only able to get 5 votes.
I agree that ideally I should provide alternatives. However, I honestly thing that the kind of tweaking that is currently happening has no visible benefits but a lot of drawbacks. I simply do not /see/ any alternatives.
Also, I'm of the opinion that voting generally should not be tampered with. Either count every vote or drop voting altogether.
Also, I'm of the opinion that voting generally should not be tampered with. Either count every vote or drop voting altogether.
I think (and I can only guess) that part of the process of testing is just doing things based on what Paul wants to see, then determining if the results meet his goals or not, but without having to ask permission from the user community or tell them everything that's going on. Just the fact that he's testing at all would indicate that he's keeping an eye on whether his tests are giving him the desired results. If not, I'm sure he'll keep iterating and trying new things.
I can certainly appreciate your frustration, and I'm not just brushing that aside, but I think for anyone (yourself included) who has been here awhile, it's obvious that Paul is always trying new things, and ALWAYS to make the site better, so it's just a matter of dealing with the changes as he makes them with an understanding that everyone has the same ultimate goal. If he makes changes without announcing them, it's just a quirk that members of the community need to be aware of, and just learn to live with.
I can certainly appreciate your frustration, and I'm not just brushing that aside, but I think for anyone (yourself included) who has been here awhile, it's obvious that Paul is always trying new things, and ALWAYS to make the site better, so it's just a matter of dealing with the changes as he makes them with an understanding that everyone has the same ultimate goal. If he makes changes without announcing them, it's just a quirk that members of the community need to be aware of, and just learn to live with.
Apostasy!
Anyway, the problem is not the policies as such, it is the lack of explanation/feedback. Depending on the nature of the heuristic employed, feedback could perhaps be given in a users profile or during the action of voting (or even merely a single, central explanation of the heuristics employed).
(Personally I am neutral on the issue).
Anyway, the problem is not the policies as such, it is the lack of explanation/feedback. Depending on the nature of the heuristic employed, feedback could perhaps be given in a users profile or during the action of voting (or even merely a single, central explanation of the heuristics employed).
(Personally I am neutral on the issue).
Hm yeah. I second that. My votes didn't have an effect on anything for a couple of months now, which is very disappointing when coming across good stories and comments. Voting degraded for me just to keep bookmarks around in my saved stories. :(
Your recent average (discarding the two extremes) needs to be above 2.
You can check yours here http://news.ycombinator.com/user?id=flashingpumpkin
Your "26 point" comment would be discarded as an extreme. There's also a delay (a day?) til averages are recalculated, so your present "5 point" comment wouldn't be included yet.
You can check yours here http://news.ycombinator.com/user?id=flashingpumpkin
Your "26 point" comment would be discarded as an extreme. There's also a delay (a day?) til averages are recalculated, so your present "5 point" comment wouldn't be included yet.
Your recent average (discarding the two extremes) needs to be above 2.
You can check yours here http://news.ycombinator.com/user?id=flashingpumpkin
You can check yours here http://news.ycombinator.com/user?id=flashingpumpkin
Another level is the censoring.
Don't get me wrong, I love HN and the quality of posts here. But I was censored early on, (probably) for giving _one_ tongue-in-cheek comment about Erlang (it even got many upvotes).
At first I didn't notice that I was being censored, until I accessed HN from another computer, where I saw that all of my posts where not displayed, nor where my comments. Even the loading time of HN was significantly slowed down while being logged in with that old nick.
While I'm a guest here and must accept the hosts, I'm writing this, just to let you know that some stuff is happening in the background, which is perhaps not apparent to everyone. :)
Don't get me wrong, I love HN and the quality of posts here. But I was censored early on, (probably) for giving _one_ tongue-in-cheek comment about Erlang (it even got many upvotes).
At first I didn't notice that I was being censored, until I accessed HN from another computer, where I saw that all of my posts where not displayed, nor where my comments. Even the loading time of HN was significantly slowed down while being logged in with that old nick.
While I'm a guest here and must accept the hosts, I'm writing this, just to let you know that some stuff is happening in the background, which is perhaps not apparent to everyone. :)
I just can't keep up with Hacker News anymore. I used to read /newcomments a lot but that doesn't work so well for me these days. The opaque voting policies don't really bother me, but I probably understand them more than most people. I wish there was some way to shrink Hacker News in a Dunbar's number sense, but that's as of yet an unsolved problem in social news.
Security through obscurity is inferior to security by design, and for that reason I agree that the voting policy should be transparent to all.
That said, I have not noticed the quality issues the original author laments.
That said, I have not noticed the quality issues the original author laments.
I think the problem is that most of the people on HN are of a relatively high quality. People who's votes don't count and people who's votes do count, all voting on something, doesn't necessarily lead to diminished results in the quality of what gets upvoted.
The problem is that the special few who get disenfranchised have no idea such a thing has happened, and apparently it's not happening because they are trolls or expressing some other type of undesirable behavior, it's happening purely based off of some arbitrary calculation of the week that's otherwise completely hidden from the users -- so in the end there's no reason at all for it to be part of the system.
The problem is that the special few who get disenfranchised have no idea such a thing has happened, and apparently it's not happening because they are trolls or expressing some other type of undesirable behavior, it's happening purely based off of some arbitrary calculation of the week that's otherwise completely hidden from the users -- so in the end there's no reason at all for it to be part of the system.
Not sure about the new voting system. HN front page has become weired those days. I made few front pages in the past, the link stay for few hours (at least). Few days later, I made a front page, the discussion stayed few minutes and got only 12 point although (IMO) I think it's interesting and should get more attention.
What makes me angry is that some stories quickly disappear from the front page, which is going to turn HN into another Digg.
What makes me angry is that some stories quickly disappear from the front page, which is going to turn HN into another Digg.
Could someone explain exactly what the problem is for those of us that missed it? Either my lurking is too occasional or I'm oblivious.
PG constantly fucks around with the voting process to do things other than the obvious time-decay voting system where fresh content and fresh votes get a temporary boost.
Recently it's seemed like someone on the backend is manually demoting posts from the front page. That's better than the old policy of deleting them and sometimes hell-banning the poster, but it's still pretty anti-community.
Something that's gone on for a long time is that you'll get your account disenfranchised -- the voting buttons all still show up, and your votes get registered (can't vote again), but the points don't get added: http://news.ycombinator.com/item?id=1029776
Recently it's seemed like someone on the backend is manually demoting posts from the front page. That's better than the old policy of deleting them and sometimes hell-banning the poster, but it's still pretty anti-community.
Something that's gone on for a long time is that you'll get your account disenfranchised -- the voting buttons all still show up, and your votes get registered (can't vote again), but the points don't get added: http://news.ycombinator.com/item?id=1029776
Dumb question. What exactly is the difference between a hell-banning and a normal banning?
When you're hell-banned, you aren't notified of it, you can continue to post links and comments, but you're the only one that sees them normally -- they're deleted for everyone except logged-in users that have [showdead] turned on.
It was pioneered by the goons at SomethingAwful for dealing with problem users in a particularly lulzy way. It is rather effective for drive-by trolls that would just create another account and amp things up if they knew. But even SomethingAwful, with the most deliberately lulz-stoking moderators on the web, uses other methods almost exclusively: probation, timeouts, and normal banning.
news.arc is uniquely capricious in that hell-banning is the only method in use
It was pioneered by the goons at SomethingAwful for dealing with problem users in a particularly lulzy way. It is rather effective for drive-by trolls that would just create another account and amp things up if they knew. But even SomethingAwful, with the most deliberately lulz-stoking moderators on the web, uses other methods almost exclusively: probation, timeouts, and normal banning.
news.arc is uniquely capricious in that hell-banning is the only method in use
Maybe you would be happier if you chose to use the site differently?
It seems to me that a lot of the complaints about the voting system revolve around how it impacts what gets on "the front page" and what stays there. I have been a member here for a few months and only very recently figured out how to intentionally pull up "the front page" (I'm quick like that). I typically click on the "new" option and refresh that several times during my visit. Then I read what gets my interest. I'm not much of a crowd-follower and never have been. It really doesn't mean much to me if something is "popular". So what? Lots of things that are popular are pure dreck -- which is apparently the basis for why the voting system is being tweaked to begin with.
Off the top of my head, I don't know of a comparable site with comparable size which has tackled a similar problem. It seems to me HN is a victim of its own success in that these problems apparently arise out of how much traffic/membership has grown. Either the people that run it will learn and grow and adapt or it will hit a fail wall and seriously degrade. I've left some sites which hit that wall and which I felt were no longer worth participating in. In the cases I know of, a huge limiting factor was the head honcho. My hope is that the experiments being done here will yield good results, resolve the problem(s) that inspired them and HN can go on to new, grander levels of success.
Oft quoted remark in some forum I used to go to: "Growth for the sake of growth is the ideology of the cancer cell." My hope is that HN continues to have goals other than just "growth". Presumably, these irritating experiments are being done in the name of keeping some of the qualities which got HN here to begin with and are a search for a means to keep the baby but dump the bath water.
Here is to hoping it all works out.
It seems to me that a lot of the complaints about the voting system revolve around how it impacts what gets on "the front page" and what stays there. I have been a member here for a few months and only very recently figured out how to intentionally pull up "the front page" (I'm quick like that). I typically click on the "new" option and refresh that several times during my visit. Then I read what gets my interest. I'm not much of a crowd-follower and never have been. It really doesn't mean much to me if something is "popular". So what? Lots of things that are popular are pure dreck -- which is apparently the basis for why the voting system is being tweaked to begin with.
Off the top of my head, I don't know of a comparable site with comparable size which has tackled a similar problem. It seems to me HN is a victim of its own success in that these problems apparently arise out of how much traffic/membership has grown. Either the people that run it will learn and grow and adapt or it will hit a fail wall and seriously degrade. I've left some sites which hit that wall and which I felt were no longer worth participating in. In the cases I know of, a huge limiting factor was the head honcho. My hope is that the experiments being done here will yield good results, resolve the problem(s) that inspired them and HN can go on to new, grander levels of success.
Oft quoted remark in some forum I used to go to: "Growth for the sake of growth is the ideology of the cancer cell." My hope is that HN continues to have goals other than just "growth". Presumably, these irritating experiments are being done in the name of keeping some of the qualities which got HN here to begin with and are a search for a means to keep the baby but dump the bath water.
Here is to hoping it all works out.
I don't think that pg's goal is growth for growth's sake:
"When we launched in February 2007, weekday traffic was around 1600 daily uniques. It's since grown to around 22,000. This growth rate is a bit higher than I'd like. I'd like the site to grow, since a site that isn't growing at least slowly is probably dead. But I wouldn't want it to grow as large as Digg or Reddit—mainly because that would dilute the character of the site, but also because I don't want to spend all my time dealing with scaling."
http://www.paulgraham.com/hackernews.html
"When we launched in February 2007, weekday traffic was around 1600 daily uniques. It's since grown to around 22,000. This growth rate is a bit higher than I'd like. I'd like the site to grow, since a site that isn't growing at least slowly is probably dead. But I wouldn't want it to grow as large as Digg or Reddit—mainly because that would dilute the character of the site, but also because I don't want to spend all my time dealing with scaling."
http://www.paulgraham.com/hackernews.html
I'd like to have the option to hide all karma at least for one's own comments (and total karma).
As much as I hate to admit, it affects the things I say and I don't want to see it anymore.
As much as I hate to admit, it affects the things I say and I don't want to see it anymore.
geez, seems more like "command PG" than "ask PG"
complain on your blog, not here...flagged
complain on your blog, not here...flagged
Whatever you are experimenting on, I do not think that it is having any positive effects. Bandwagon voting - which I think is more detrimental than unfair downvoting - is still rampant.
For me the current situation has the effect that scores cannot be trusted. Coupled with the fact that completely legitimate posts are sometimes auto-killed, for me this site's credibility is questionable.
I do believe that you have good intentions, but please consider the unintended consequences.