Every major article spawns countless knockoffs as media outlets compete for views. The HN guidelines ask you to post the original source. When you don't, moderators change it. When we don't, users ask us to.
This is routine moderation that we do every day. Tracking down better sources is one way that the community and moderators keep up quality here. We'd prefer it, of course, if submitters would do this before posting.
When I change a url, I often post a comment saying so. The other moderators can't, because they're not public. It isn't clear to me yet what to do about that.
Story submissions on HN are not the property of the submitter. They are pointers, not values—the content belongs to the source, and the HN slot belongs to the site and to the community. If changing the address in that slot makes the site better, we do.
(Comments are different. Story titles are not supposed to be your own words; comments are. We never change those, except on rare occasions when users ask us to.)
I know it sounds subjective and arbitrary to say "make the site better", as if by fiat, but it's less arbitrary than it sounds. The community nearly always agrees about one source being better than another. (I'd be surprised if there were much disagreement in the present case.) And on the relatively few occasions when users think we've made the wrong call, we usually change it.
1. High-end blogspam might be a more accurate way of putting it.
Could you comment about how the item was buried shortly after the changes?
If it wasn't buried by a mod, then either the penalty that's applied to surveillance stories is incredibly enormous, or the story was flagged into oblivion by users. Did flagging only start after the title and link were changed? Isn't that quite suspicious? It really looks like someone's doing title/domain/url monitoring for stories to flag, and this story is a good candidate for examination.
I guess I'd just like confirmation that 1) mods didn't bury the story, and 2) it doesn't look like flagging suddenly increased after the change. If flagging did suddenly increase, it would be strong evidence of flagging abuse. There's no explanation for flagging the second story and not the first, other than failure to detect it automatically. In fact, you would expect the exact opposite pattern if the story were being flagged legitimately.
The penalty applied to NSA stories --- if it still exists at all, since 'dang said last week he was considering ending it --- is tiny. The way I know that is: 'dang publically said it's tiny; the second smallest on the site, which has many penalties (almost all of them, as I understand it, quite boring).
I didn't flag the GCHQ online poll story, but I would have a very hard time arguing that someone who did was "abusing" the flag button. It's a marginal story, at best.
You've asked this several times, so I'm not sure where to put the answer. Let's try here.
(By the way, one reason we ask people to send moderation questions to hn@ycombinator.com is that then we're sure to see them. We can't possibly read every post to the site. We miss most of them, in fact.)
The story fell off the front page because of a moderator penalty, though user flags caused the fall to be more dramatic.
We're not trying to suppress surveillance stories [1]. We're trying to balance it so the major ones get attention but the me-toos don't crowd out other things. A moderator thought this was a me-too rather than a major. I might have made the opposite call, but it was probably a borderline case. For one thing, the article was shorter, and that's usually but not always correlated with lesser substance. Since we can't come close to reading everything, we're forced to rely on such heuristics. They're miserable, but they mostly work.
We moderate this way not because of personal opinion (I'm personally more inclined the other way, for what that's worth), but because if we didn't, the site would skew heavily toward these stories and there would be many more complaints (remember "Why has Hacker News turned into NSA News"?) than there currently are. The HN community is in disagreement here. Our job is to mediate that. Which of course is impossible. So we try to fail less badly.
We're open to suggestions about how to get the balance better. We do need to balance it somehow. The idea that votes can just determine everything—without irrevocably changing the character of the site—is sadly, utterly wrong.
Some of the most vocal feedback we're getting on the surveillance stories is from people who seem only to care about this one issue and would happily have it dominate HN. That feedback isn't so helpful, because it's at odds with what this site is. But if those of you who follow this issue closely would help us make better editorial distinctions between the major stories and the pilers-on, we'd be grateful. We're not domain experts on any of this.
We're open to suggestions about how to get the balance better. We do need to balance it somehow.
At the risk of suggesting the site become more like reddit, the site should maybe become a scosh more like reddit. Since the HN community has become more diverse in its interests than the structure of the site was perhaps intended for, it needs to be possible to group posts not just by votes, but contextually, either with tags or through sub-boards. This wouldn't necessarily have to affect the way the site appears to function[0], for users who like things the way they are now - but users who want to see more NSA related content could choose (maybe through their account page) to have NSA related content bubble up for them, regardless of their actual gravity.
[0] although to prevent certain stories from overwhelming the site anyway, it might be necessary to sort the frontpage by trending tags/groups when multiple stories of the same type exist... I don't know.
Furthermore, I think it might help to alleviate the suspicions of people like discostrings if there were some automatic feedback to actually show the effect of user/mod flags on a thread over time... surely "mostly killed by user flags" is something that could be graphed somewhere.
Thank you for the detailed explanation of what happened to the post and your thoughts on the topic in general. I'm happy to hear that what's been going on with some of the bigger surveillance stories is the result of the difficulties of moderation and not external manipulation. I now have a greater understanding of why efforts to stop the topic from dominating the site end up supressing it.
As it seems you noticed, my main frustration is that surveillance stories with new information, and other particularly insightful pieces about the topic, often aren't staying on the front page, which essentially means that no one sees or discusses them at any level of depth. HN is generally so good at what it does that for most topics of interest to the community, it can be one's daily news source for an overview of what's going on. Readers come to expect that topics that interest HN will be covered well and that important developments will be posted. So it's particularly jarring after reading tech-related news only from HN for a few days to see you missed a major surveillance story, especially one that deals with the tech or tools. When you search HN and see you missed it because it was submitted but probably buried before anyone discussed it, you start feeling that moderators are parting ways with the community or that something sinister is going on.
I think a number of other difficulties arise as a result of the major stories not showing up. I imagine users start getting the sense that an important topic isn't being discussed, so they submit and upvote even more "me-too" stories. On the stories that do make it, people don't bother writing long and substantive comments because of the risk of the post dropping to oblivion at any moment and no one ever seeing the comment. Those who know a lot about the topic become less likely to write about it insightfully here, increasing the noise further and giving the topic a worse reputation for quality.
How to improve things is a difficult questions, as this topic's scenario really doesn't play to the site's strengths. It seems there's really no feature for "best-of-a-type" other than penalties.
Never-the-less, here are a few ideas (some of which are assuredly overly vague or simplistic, but since I don't know which might be helpful...):
- Consideration could be given to how many surveillance stories are currently on the front page. Very few people want an entire front page filled with these stories every day, but given the level of interest and the continuing developments one would still expect to see a number of stories every day or two. Maybe only resort to giving a penalty to an article that isn't obviously low quality / blogspam if there are already surveillance stories on the front page?
- Using heuristics concerning novelty, and whether the particular novel issue has been discussed on HN, could supplement length
----- A quick check of whether the article claims it's new information, the date of the article, and a general search for the main claim of the article
----- Less scrutiny of publications with a history of breaking stories or that have access to documents, like theguardian.com, firstlook.org, washingtonpost.com, and spiegel.de
- A way to know a topic isn't going to quickly disappear after investing time in writing a comment would be nice. Some indicator that a topic has been "blessed" by a mod and won't receive a penalty would be helpful for this.
- Having a moderator who follows the topic might be helpful. I definitely understand that not every moderator does, and that it's probably really hard to make judgments if you're not following it. Someone a bit more familiar with the topic would be a good resource and could keep an eye out for sub-optimal penalties being given out.
- In general, visible transparency about what's going on would be helpful. I'm not sure how to achieve that because the site's not really designed for it. Talking to you here is great. Having other people with the same concerns get the information is important too.
- The root issue isn't really that interest in the topic is too high--it's that there's a low signal-to-noise ratio in the stories that are submitted and the passion for the topic is so high that even the noise is upvoted. Some users probably want to make sure everyone knows what's going on. Once a normal level of the best stories are making the front page and there's an outlet for that passion, perhaps the submission problem will decrease and a virtuous cycle will begin.
I'll email first in the future for anything I'd like a response about or if I have more suggestions--I took the comment approach because in the case of a forum or many other online communities, it would be better to have the conversation publicly, but I agree that HN's not really set up well for that (as evidenced by my multiple requests for a response).
Again, I'm grateful for your response, and I greatly appreciate the work you're doing here and your commitment to addressing users' concerns.
Thanks for the thoughtful analysis. I agree with almost all of it.
I'm not sure how practical these particular suggestions are--for example, we do already try to identify which issues have already been discussed on HN; it's just hard to get right every time. But I do think the pendulum has perhaps swung too far the other way, so we've provisionally turned off the automatic weight on NSA stories. Going forward, I hope the community will help us distinguish major articles from minor ones. People can do that by flagging minor ones and commenting here or (better) emailing hn@ycombinator.com if they see a major one that they think might be weighted unfairly. We can turn off the flags when that happens.
Thanks for your reply. I wanted to respond sooner (while the thread had a little life left), but was traveling.
In any case, I think a major issue is the continued attribution without notice to the submitter or publicly on the actual submission itself. You clearly appreciate this because, as you've indicated, you generally try to note when you've made changes, however not all moderators do because they are not "public". It seems a simple solution would be to have the system automatically generate a public change note on the submission when any mod makes a change. The actual mod can certainly be anonymous if that's desirable for some reason. An additional automated email to the submitter would be good as well.
Not only is this a courtesy to the submitter, but it improves the experience for those who may be actively involved in discussions on the thread. And, frankly, it is simply the right thing to do where the principle of proper attribution is concerned, so it is worth ensuring the consistency that automation would provide.
The idea that people are submitting a "slot" seems contradictory to me. Does HN want topics or actual content? The UI and submission process (allowing URLs, asking for title retention, etc.) are also inconsistent with this notion of slots. Most importantly, I think most people associate attribution with actual content versus a "slot".
My apologies for not using the email address to present the question that is the subject of this submission. I was sincerely unaware. However (and not to qualify that apology), I will separately add that a lack of consistency from HN mods invites confusion and guideline violations (inadvertent or otherwise). That is, given that there is an acknowledged value in HN publicly notifying the submitter/community when a moderation action has been taken, then on the occasions when it fails to do so it leaves users open to confusion and frustration. Hence, more questions will be asked and, statistically, some of these will--inadvertently or otherwise--not be via email, thus contributing to an increase in the almost comically hated meta-noise.
This is actually consistent with UI design. Just as any good UI provides consistent feedback and teaches the user in order to prevent errors and frustration, HN can teach consistency by demonstrating it and by answering questions before they are even asked via providing reliable feedback.
Speaking of "meta", it seems worthwhile to have some forum for public discussion of HN operation/moderation, which can obviously be apart from standard content that winds up on the frontpage. Because, as this thread alone indicates, there is obviously a great deal of interest on the topic, and that interest is on both sides. That is, ironically enough, those who seem to most hate "meta" also become vocal about their hatred for it, thus contributing to it themselves. So, they really do want to talk about it.
Most importantly, however, those folks who advocate maintaining the status quo without further discussion have plenty of means for sinking content, including flagging and simply ignoring/not upvoting. And, of course, they are free to use comments to openly bash that with which they disagree. Since they are supporting the status-quo, their comments are encouraged and welcomed.
Yet, the only means by which users can offer suggestions or pose questions of concern is privately via email. There can be no discussion among the community as a whole without bringing out the anti-meta zealots and being modded/flagged into oblivion. If HN is OK with its approach because it believes it to be the best way to ensure quality, then OK. What else can we say? But, it opens itself to a bit of a confirmation bias problem, wherein only those who agree with the status quo are heard at significant volume.
It also seems that the number of factors that contribute to "quality" creates an unnaturally supressive effect on content and appears to lead to a bit of over-modding (or an overly robust negative toolset). How is it that a story can have more interest than any other story and, at the same time, be flagged or modded into obscurity? This effectively says that, while there is nothing more interesting than this submission to the community as a whole, there are nonetheless enough people who dislike it (if even for technicalities) to kill it, thus stifling discussion for others. By definition, it will then be replaced with less compelling content. I hear you that upvoting alone may not be sufficient to surface content. But, it does seem to be a serious bikeshedding problem when a submission that rockets to the top spot and lingers for a few hours is suddenly sunk on a "technicality". The net effect is that a minority of HN killed a community-relevant story that garnered much interest. Is that result in the best interest of the HN community? Or has some purism crept in?
At the end of the day, HN can do what it wants and part of the problem may be that its appearance is democratic, when it actually is not. I mean this with regard to both how it is operated and how content is surfaced. So, that may lead to disappointment for a number of users, which I believe we are seeing on this thread.
Comments
You submitted http://www.slate.com/blogs/future_tense/2014/07/15/documents.... That's just a digest [1] of the firstlook.org article [2], so a moderator correctly changed it.
Every major article spawns countless knockoffs as media outlets compete for views. The HN guidelines ask you to post the original source. When you don't, moderators change it. When we don't, users ask us to.
This is routine moderation that we do every day. Tracking down better sources is one way that the community and moderators keep up quality here. We'd prefer it, of course, if submitters would do this before posting.
When I change a url, I often post a comment saying so. The other moderators can't, because they're not public. It isn't clear to me yet what to do about that.
Story submissions on HN are not the property of the submitter. They are pointers, not values—the content belongs to the source, and the HN slot belongs to the site and to the community. If changing the address in that slot makes the site better, we do.
(Comments are different. Story titles are not supposed to be your own words; comments are. We never change those, except on rare occasions when users ask us to.)
I know it sounds subjective and arbitrary to say "make the site better", as if by fiat, but it's less arbitrary than it sounds. The community nearly always agrees about one source being better than another. (I'd be surprised if there were much disagreement in the present case.) And on the relatively few occasions when users think we've made the wrong call, we usually change it.
1. High-end blogspam might be a more accurate way of putting it.
2. https://firstlook.org/theintercept/2014/07/14/manipulating-o...
Could you comment about how the item was buried shortly after the changes?
If it wasn't buried by a mod, then either the penalty that's applied to surveillance stories is incredibly enormous, or the story was flagged into oblivion by users. Did flagging only start after the title and link were changed? Isn't that quite suspicious? It really looks like someone's doing title/domain/url monitoring for stories to flag, and this story is a good candidate for examination.
I guess I'd just like confirmation that 1) mods didn't bury the story, and 2) it doesn't look like flagging suddenly increased after the change. If flagging did suddenly increase, it would be strong evidence of flagging abuse. There's no explanation for flagging the second story and not the first, other than failure to detect it automatically. In fact, you would expect the exact opposite pattern if the story were being flagged legitimately.
Thanks.
The penalty applied to NSA stories --- if it still exists at all, since 'dang said last week he was considering ending it --- is tiny. The way I know that is: 'dang publically said it's tiny; the second smallest on the site, which has many penalties (almost all of them, as I understand it, quite boring).
I didn't flag the GCHQ online poll story, but I would have a very hard time arguing that someone who did was "abusing" the flag button. It's a marginal story, at best.
You've asked this several times, so I'm not sure where to put the answer. Let's try here.
(By the way, one reason we ask people to send moderation questions to hn@ycombinator.com is that then we're sure to see them. We can't possibly read every post to the site. We miss most of them, in fact.)
The story fell off the front page because of a moderator penalty, though user flags caused the fall to be more dramatic.
We're not trying to suppress surveillance stories [1]. We're trying to balance it so the major ones get attention but the me-toos don't crowd out other things. A moderator thought this was a me-too rather than a major. I might have made the opposite call, but it was probably a borderline case. For one thing, the article was shorter, and that's usually but not always correlated with lesser substance. Since we can't come close to reading everything, we're forced to rely on such heuristics. They're miserable, but they mostly work.
We moderate this way not because of personal opinion (I'm personally more inclined the other way, for what that's worth), but because if we didn't, the site would skew heavily toward these stories and there would be many more complaints (remember "Why has Hacker News turned into NSA News"?) than there currently are. The HN community is in disagreement here. Our job is to mediate that. Which of course is impossible. So we try to fail less badly.
We're open to suggestions about how to get the balance better. We do need to balance it somehow. The idea that votes can just determine everything—without irrevocably changing the character of the site—is sadly, utterly wrong.
Some of the most vocal feedback we're getting on the surveillance stories is from people who seem only to care about this one issue and would happily have it dominate HN. That feedback isn't so helpful, because it's at odds with what this site is. But if those of you who follow this issue closely would help us make better editorial distinctions between the major stories and the pilers-on, we'd be grateful. We're not domain experts on any of this.
1. To pick an obvious example, https://news.ycombinator.com/item?id=8058362 has been on the front page all day.
At the risk of suggesting the site become more like reddit, the site should maybe become a scosh more like reddit. Since the HN community has become more diverse in its interests than the structure of the site was perhaps intended for, it needs to be possible to group posts not just by votes, but contextually, either with tags or through sub-boards. This wouldn't necessarily have to affect the way the site appears to function[0], for users who like things the way they are now - but users who want to see more NSA related content could choose (maybe through their account page) to have NSA related content bubble up for them, regardless of their actual gravity.
[0] although to prevent certain stories from overwhelming the site anyway, it might be necessary to sort the frontpage by trending tags/groups when multiple stories of the same type exist... I don't know.
Furthermore, I think it might help to alleviate the suspicions of people like discostrings if there were some automatic feedback to actually show the effect of user/mod flags on a thread over time... surely "mostly killed by user flags" is something that could be graphed somewhere.
Thank you for the detailed explanation of what happened to the post and your thoughts on the topic in general. I'm happy to hear that what's been going on with some of the bigger surveillance stories is the result of the difficulties of moderation and not external manipulation. I now have a greater understanding of why efforts to stop the topic from dominating the site end up supressing it.
As it seems you noticed, my main frustration is that surveillance stories with new information, and other particularly insightful pieces about the topic, often aren't staying on the front page, which essentially means that no one sees or discusses them at any level of depth. HN is generally so good at what it does that for most topics of interest to the community, it can be one's daily news source for an overview of what's going on. Readers come to expect that topics that interest HN will be covered well and that important developments will be posted. So it's particularly jarring after reading tech-related news only from HN for a few days to see you missed a major surveillance story, especially one that deals with the tech or tools. When you search HN and see you missed it because it was submitted but probably buried before anyone discussed it, you start feeling that moderators are parting ways with the community or that something sinister is going on.
I think a number of other difficulties arise as a result of the major stories not showing up. I imagine users start getting the sense that an important topic isn't being discussed, so they submit and upvote even more "me-too" stories. On the stories that do make it, people don't bother writing long and substantive comments because of the risk of the post dropping to oblivion at any moment and no one ever seeing the comment. Those who know a lot about the topic become less likely to write about it insightfully here, increasing the noise further and giving the topic a worse reputation for quality.
How to improve things is a difficult questions, as this topic's scenario really doesn't play to the site's strengths. It seems there's really no feature for "best-of-a-type" other than penalties.
Never-the-less, here are a few ideas (some of which are assuredly overly vague or simplistic, but since I don't know which might be helpful...): - Consideration could be given to how many surveillance stories are currently on the front page. Very few people want an entire front page filled with these stories every day, but given the level of interest and the continuing developments one would still expect to see a number of stories every day or two. Maybe only resort to giving a penalty to an article that isn't obviously low quality / blogspam if there are already surveillance stories on the front page?
- Using heuristics concerning novelty, and whether the particular novel issue has been discussed on HN, could supplement length
----- A quick check of whether the article claims it's new information, the date of the article, and a general search for the main claim of the article
----- Less scrutiny of publications with a history of breaking stories or that have access to documents, like theguardian.com, firstlook.org, washingtonpost.com, and spiegel.de
- A way to know a topic isn't going to quickly disappear after investing time in writing a comment would be nice. Some indicator that a topic has been "blessed" by a mod and won't receive a penalty would be helpful for this.
- Having a moderator who follows the topic might be helpful. I definitely understand that not every moderator does, and that it's probably really hard to make judgments if you're not following it. Someone a bit more familiar with the topic would be a good resource and could keep an eye out for sub-optimal penalties being given out.
- In general, visible transparency about what's going on would be helpful. I'm not sure how to achieve that because the site's not really designed for it. Talking to you here is great. Having other people with the same concerns get the information is important too.
- The root issue isn't really that interest in the topic is too high--it's that there's a low signal-to-noise ratio in the stories that are submitted and the passion for the topic is so high that even the noise is upvoted. Some users probably want to make sure everyone knows what's going on. Once a normal level of the best stories are making the front page and there's an outlet for that passion, perhaps the submission problem will decrease and a virtuous cycle will begin.
I'll email first in the future for anything I'd like a response about or if I have more suggestions--I took the comment approach because in the case of a forum or many other online communities, it would be better to have the conversation publicly, but I agree that HN's not really set up well for that (as evidenced by my multiple requests for a response).
Again, I'm grateful for your response, and I greatly appreciate the work you're doing here and your commitment to addressing users' concerns.
Thanks for the thoughtful analysis. I agree with almost all of it.
I'm not sure how practical these particular suggestions are--for example, we do already try to identify which issues have already been discussed on HN; it's just hard to get right every time. But I do think the pendulum has perhaps swung too far the other way, so we've provisionally turned off the automatic weight on NSA stories. Going forward, I hope the community will help us distinguish major articles from minor ones. People can do that by flagging minor ones and commenting here or (better) emailing hn@ycombinator.com if they see a major one that they think might be weighted unfairly. We can turn off the flags when that happens.
HN's goals here are balance and substance.
Thanks for your reply. I wanted to respond sooner (while the thread had a little life left), but was traveling.
In any case, I think a major issue is the continued attribution without notice to the submitter or publicly on the actual submission itself. You clearly appreciate this because, as you've indicated, you generally try to note when you've made changes, however not all moderators do because they are not "public". It seems a simple solution would be to have the system automatically generate a public change note on the submission when any mod makes a change. The actual mod can certainly be anonymous if that's desirable for some reason. An additional automated email to the submitter would be good as well.
Not only is this a courtesy to the submitter, but it improves the experience for those who may be actively involved in discussions on the thread. And, frankly, it is simply the right thing to do where the principle of proper attribution is concerned, so it is worth ensuring the consistency that automation would provide.
The idea that people are submitting a "slot" seems contradictory to me. Does HN want topics or actual content? The UI and submission process (allowing URLs, asking for title retention, etc.) are also inconsistent with this notion of slots. Most importantly, I think most people associate attribution with actual content versus a "slot".
My apologies for not using the email address to present the question that is the subject of this submission. I was sincerely unaware. However (and not to qualify that apology), I will separately add that a lack of consistency from HN mods invites confusion and guideline violations (inadvertent or otherwise). That is, given that there is an acknowledged value in HN publicly notifying the submitter/community when a moderation action has been taken, then on the occasions when it fails to do so it leaves users open to confusion and frustration. Hence, more questions will be asked and, statistically, some of these will--inadvertently or otherwise--not be via email, thus contributing to an increase in the almost comically hated meta-noise.
This is actually consistent with UI design. Just as any good UI provides consistent feedback and teaches the user in order to prevent errors and frustration, HN can teach consistency by demonstrating it and by answering questions before they are even asked via providing reliable feedback.
Speaking of "meta", it seems worthwhile to have some forum for public discussion of HN operation/moderation, which can obviously be apart from standard content that winds up on the frontpage. Because, as this thread alone indicates, there is obviously a great deal of interest on the topic, and that interest is on both sides. That is, ironically enough, those who seem to most hate "meta" also become vocal about their hatred for it, thus contributing to it themselves. So, they really do want to talk about it.
Most importantly, however, those folks who advocate maintaining the status quo without further discussion have plenty of means for sinking content, including flagging and simply ignoring/not upvoting. And, of course, they are free to use comments to openly bash that with which they disagree. Since they are supporting the status-quo, their comments are encouraged and welcomed.
Yet, the only means by which users can offer suggestions or pose questions of concern is privately via email. There can be no discussion among the community as a whole without bringing out the anti-meta zealots and being modded/flagged into oblivion. If HN is OK with its approach because it believes it to be the best way to ensure quality, then OK. What else can we say? But, it opens itself to a bit of a confirmation bias problem, wherein only those who agree with the status quo are heard at significant volume.
It also seems that the number of factors that contribute to "quality" creates an unnaturally supressive effect on content and appears to lead to a bit of over-modding (or an overly robust negative toolset). How is it that a story can have more interest than any other story and, at the same time, be flagged or modded into obscurity? This effectively says that, while there is nothing more interesting than this submission to the community as a whole, there are nonetheless enough people who dislike it (if even for technicalities) to kill it, thus stifling discussion for others. By definition, it will then be replaced with less compelling content. I hear you that upvoting alone may not be sufficient to surface content. But, it does seem to be a serious bikeshedding problem when a submission that rockets to the top spot and lingers for a few hours is suddenly sunk on a "technicality". The net effect is that a minority of HN killed a community-relevant story that garnered much interest. Is that result in the best interest of the HN community? Or has some purism crept in?
At the end of the day, HN can do what it wants and part of the problem may be that its appearance is democratic, when it actually is not. I mean this with regard to both how it is operated and how content is surfaced. So, that may lead to disappointment for a number of users, which I believe we are seeing on this thread.
Thanks for reading and, again, for your reply.