I think there could be an aspect of social cooling contributing to decreased per-user engagement. I find myself considering my online comments more carefully now than I did 10 years ago across all platforms.
Hard agree- a lot of people just opt not to at all- that's what I usually do (other than HN). And when you do decide to go ahead and participate, the energy it takes for each post is a lot higher- you have to consider every contingency you can think of and make sure they all check out. And none of us know if we're going to be wrong when they unearth these posts and pin them on us later, unfortunately.
> I find myself considering my online comments more carefully now than I did 10 years ago across all platforms.
I find myself doing this, but I chalked it up to aging and the reality that I may eventually have no choice but to work for one of the companies I have ridiculed online.
I know that participating on HN has made my comments on all forums more thoughtful, less hyperbolic, more serious, and generally better considered. The reasonable rules, excellent moderation and the community culture here have been excellent teachers. The voting mechanism helps me gauge if my comments are reasonable (I don't care about the magnitude but more the negative/positive score telling me if I'm off the rails).
Perhaps a similar effect eventually takes hold on many of the contributors, and for the ones who can't moderate themselves to the culture here, I guess they'll recess into lurkers or simply go away elsewhere.
> The voting mechanism helps me gauge if my comments are reasonable (I don't care about the magnitude but more the negative/positive score telling me if I'm off the rails).
This is my sense of why pg picked a balance number, not a + and - number separately.
The "decreased per-user engagement" evidence in this post is a bit thin. Mostly based on a kind of strange apparent outlier of accounts going idle (1+ year) in 2023 but the binning is also a full year and the 'idle year' is counted in a weird clippy (i.e. looking at calendar year rather than elapsed year) way. So it's one (aggregate) data point at the very end of the data. It might not be wrong but it feels somewhat iffy to draw conclusions from.
> Mostly based on a kind of strange apparent outlier of accounts going idle (1+ year) in 2023
Well, it is just outliers in 2023. This is an upward trend since 2020.
> but the binning is also a full year and the 'idle year' is counted in a weird clippy (i.e. looking at calendar year rather than elapsed year) way
Granted, and I acknowledge this limitation. My idea, however, is that when studying many users in the same manner, this will even out. Why? Because a full calendar year implies somewhere between 0-2 elapsed years. So the average elapsed year, over many users, is 1 year.
> I find myself considering my online comments more carefully now than I did 10 years ago across all platforms.
I am curious, why is that? Is the medium less safe than before (e.g., unpleasant interactions)? Or is it a high effort to engage in a discussion, and perhaps there are other priorities now?
1. I use my real name, which has a strong cooling effect on the kinds of stuff I write. I'm not going to get into a flame war or say career-ending things under my real name. I do this deliberately to force me to keep my responses as high quality as I can.
2. As I temper what I write based on today's norms and taboos, I also have to think about (guess) what might be taboo in 10, 20, 30 years. Assume HN will be searchable forever. People today have a habit of drudging up things that their opponents wrote decades ago, and measuring them against today's (more restrictive) yardstick. I've told jokes back in the 90s that were benign and funny then, but would get me fired today. Imagine people decades from now reading the innocent things I am writing today and how they will be offended by it!
The "users ceasing to post stories after one year" metric seems a bit wonky to me. I mean, I don't doubt it, but it's probably skewed by accounts created for self-promotion or spam. If you look at front-page stories, I think that most of them are from long-time members.
It would probably be useful to separate the churn of throwaway accounts and the ones that cross some engagement threshold - 50 comments or something like that.
We humans are emotional. Users' submissions didn't get the desired reach, and comments were thrashed to oblivion- they got emotional and left/lurked without participating anymore. Besides the usual promotional accounts that crop up, this is one of the reasons people do not participate.
Leave the emotion out of the equation, and HackerNews becomes way calmer for you.
The same story you submitted that failed gets talked about/discussed when submitted by another -- eh! All the smart people shred your comment -- ah, let's take a walk.
i mostly lurk HN, and don't post unless i feel like my contribution is useful in some way.
from my perspective, it's relieving to see that the number of users has remained mostly constant in recent years. of course it's selfish to think in this way, but almost no "social media" or UGC-based platform that i've used has actually become better or more useful to me as it became (much) larger.
this kind of fast growth in users (beyond some size) often leads to a shift in the culture that made preexisting users participate in the first place, leading to a loss in overall quality of the platform as a whole. if the growth is gradual enough, then new users eventually figure out how to fit into the culture or leave.
i guess i've said nothing that isn't obvious to people who have used a computer before, basically "yay no eternal september for HN yet", but i digress.
Growth assumes that older posters continue to post, which both according to TFA and my own analysis isn't true.
In my own analysis, comments in a thread have a recent-user bias and roughly 30-40% of an article's comments will come from the last 3-4 years of users. I found this to hold true for many large threads over the past 5 years, though I haven't yet exhaustively demonstrated this yet simply because I don't do HN analysis that often.
What that means practically is that the folks who post on this site are constantly changing. I also, generally, find that the most contentious threads on this site tend to have a relatively stronger recency bias among its posters.
By that regard, eternal September is ever present: the posts on this site weigh toward recent posters. I'd be curious to see if that effect can be explained by throwaways and sock puppet posters but I'm not sure of any reliable way to identify those especially as historic karma counts aren't kept for users.
While I have more robust models that work a lot better, a very simple method I've found is that the more recent the upper percentiles of posters are on a thread, the less I will like it: to me Eternal September is here. Of course my user here is ancient from 2009. The numbers just quantify that there's been a change in audience since I joined, a wholly unsurprising fact.
i did read the article but didn't come to the conclusion that "Eternal September" is here. although i don't argue that the quality of posts has gone up recently, i do believe that many of the new users of 2022/2023 have become - or will become - the high quality posters that attracted them here initially. as i said in the first post, i think this is true because there hasn't been an unprecedented spike in new users.
basically, quality(new user + time) = quality(old user), as long as the proportion of new users/old users remains small. of course it's subjective as to where you put this value though. i just think it hasn't been reached yet here.
I was watching The Synanon Fix on HBO, this weekend, and towards the end of the first episode, we see the original addicts, being replaced by the more lucrative "lifestylers," and I suspect that the next episodes will be downhill, from here.
The app I just wrote and released, is picking up steam, but slowly. That's deliberate. We're not doing any promotion, and it's giving us a chance to make course corrections. By the time people start piling in (and they will, but it will never be more than a rounding error to many social media apps), I think it will be in extremely good shape.
I have long wondered about doing an analysis of HN comments to find correlations of opinions, instances where people have changed their mind over time, topics where the balance of opinion is out of proportion to the balance of comments due to a vocal minority.
I have made some casual observations that it would be nice to know whether the broader data supports what I have seen. For instance, one weird thing that I have noticed, is that antagonistic comments appear to me to be posted more often by users with PGP keys in their profile. It may just be by imagination.
Possibly more importantly, but also something should be approached with caution and sensitivity would be analysis of comments to identify people in mental distress. I know of a few HN users that have taken their own lives and have wondered if there was enough information in their comments to allow someone to provide pre-emptive support.
I would be really curious to explore such correlations.
Tbh, while I was working on this, I did struggle with forming hypotheses to test, or even a suitable manifold for the different items/users (e.g., I didn't know some of the users share the PGP keys).
Unsupervised methods were not very satisfying in uncovering such hypotheses as well.
There is also the point you raised about how to share (or what to do with) some of these correlations. For example, concerning mental illness, as much as I would love to uncover that due to my sheer curiosity, I am really concerned about what I would see.
> antagonistic comments appear to me to be posted more often by users with PGP keys in their profile
While it's definitely preferable to encrypt private communications among all people including the "nothing to hide" types who generally behave well within societal norms, the types who don't behave as well feel a greater need to encrypt.
> Even more interestingly, there was negative karma. I don’t even know how this came to be.
Brand new account commenting with very low value but not worth flagging. For example "I read the article yesterday" or "Why is this article even on HN?". I've seen a couple that had negative karma on the first days. Often though it's simply spam and the account soon gets shadow-banned though I don't understand the process that does that. Example user: https://news.ycombinator.com/user?id=gabriella4151
The karma analysis completely ignores (after mentioning it) the karma gained from comments. I post almost no stories to HN, but have a karma of over 18k. Any sane, full analysis of HN culture needs to take this sort of thing into account.
I think of it as the easy way and the hard way to gain karma, but I'm biased as a "commenter" user. In reality, some people work hard to supply HN with interesting stories and that's important, but it's pretty easy to have sizeable karma without really ever engaging in conversation.
I think the tradeoff there is that the HN submitter crowd are very fast to the draw. I'll often encounter a story elsewhere that I want to hear from HN about, so I'll do a search and find it was submitted 15 minutes ago, or even days ago (and I wonder what RSS fueled craze people must go through to find and post things so fast). Of course, for me those are the best cases, because I can go through and read what people think and reply without having to wait for the uptick. But no karma for the slow.
LOL! No RSS, no nothing for me. For me, it is just consistency at a specific timeslot -- wind-down evening for me (Europe's in Office, West Coast wakes up). I tend to fire and forget. I think the hit is less than 5%, but the consistency pays off in the long run. The best I remember is that 4 of my submissions were on the front page over a couple of hours, which is 13.33% of the front page stories.
I do agree with your point, but unfortunately I was limited by what is available from the APIs: the upvotes for the comments were not provided. I couldn't find a proxy for it either. Thus, I couldn't include it in the analysis
You can build an analysis based on partial ordering. The Axios API gives you relative ranking of comments on a given level. Based on partial order you can build lattices and make assertions but that is really difficult and the juice probably isn't worth the squeeze.
Yes, this is the observation -- comments (when they are meaningful and upvoted) accelerate one's karma. This is also hard, consumes much of your core time, and is more challenging to deal with hyperboles.
If what they call 'consistency' is just the total number of stories posted, it's not really surprising to see a strong correlation with total karma points, is it? The more stories posted, the more karma gathered. Anything else would be a real head scratcher.
You could say, there might be people who keep posting stories even though the don't get upvoted, but that would be kind of irrational. If the community doesn't seem to be interested in what they post, they will stop doing so sooner or later.
> You could say, there might be people who keep posting stories even though the don't get upvoted, but that would be kind of irrational. If the community doesn't seem to be interested in what they post, they will stop doing so sooner or later.
Probably many people stop sharing in this case indeed. But for those who don't, I guess the idea is: eventually, if you keep sharing, you can collect upvotes here and there, and eventually the karma will go up
I didn't realize that the active user tenure is this low (one year) since I frequently come across veteran users.
On a similar note, sometimes i wonder where the high profile users who leave the platform move into (if anything), there doesn't seem to be a lot of options. Do they switch to alt accounts? turn into lurkers? maybe go off the grid?
Readit News
I find myself doing this, but I chalked it up to aging and the reality that I may eventually have no choice but to work for one of the companies I have ridiculed online.
Perhaps a similar effect eventually takes hold on many of the contributors, and for the ones who can't moderate themselves to the culture here, I guess they'll recess into lurkers or simply go away elsewhere.
This is my sense of why pg picked a balance number, not a + and - number separately.
Well, it is just outliers in 2023. This is an upward trend since 2020.
> but the binning is also a full year and the 'idle year' is counted in a weird clippy (i.e. looking at calendar year rather than elapsed year) way
Granted, and I acknowledge this limitation. My idea, however, is that when studying many users in the same manner, this will even out. Why? Because a full calendar year implies somewhere between 0-2 elapsed years. So the average elapsed year, over many users, is 1 year.
I am curious, why is that? Is the medium less safe than before (e.g., unpleasant interactions)? Or is it a high effort to engage in a discussion, and perhaps there are other priorities now?
Perhaps something else
1. I use my real name, which has a strong cooling effect on the kinds of stuff I write. I'm not going to get into a flame war or say career-ending things under my real name. I do this deliberately to force me to keep my responses as high quality as I can.
2. As I temper what I write based on today's norms and taboos, I also have to think about (guess) what might be taboo in 10, 20, 30 years. Assume HN will be searchable forever. People today have a habit of drudging up things that their opponents wrote decades ago, and measuring them against today's (more restrictive) yardstick. I've told jokes back in the 90s that were benign and funny then, but would get me fired today. Imagine people decades from now reading the innocent things I am writing today and how they will be offended by it!
It would probably be useful to separate the churn of throwaway accounts and the ones that cross some engagement threshold - 50 comments or something like that.
I will look into that. Cheers
Leave the emotion out of the equation, and HackerNews becomes way calmer for you.
The same story you submitted that failed gets talked about/discussed when submitted by another -- eh! All the smart people shred your comment -- ah, let's take a walk.
from my perspective, it's relieving to see that the number of users has remained mostly constant in recent years. of course it's selfish to think in this way, but almost no "social media" or UGC-based platform that i've used has actually become better or more useful to me as it became (much) larger.
this kind of fast growth in users (beyond some size) often leads to a shift in the culture that made preexisting users participate in the first place, leading to a loss in overall quality of the platform as a whole. if the growth is gradual enough, then new users eventually figure out how to fit into the culture or leave.
i guess i've said nothing that isn't obvious to people who have used a computer before, basically "yay no eternal september for HN yet", but i digress.
In my own analysis, comments in a thread have a recent-user bias and roughly 30-40% of an article's comments will come from the last 3-4 years of users. I found this to hold true for many large threads over the past 5 years, though I haven't yet exhaustively demonstrated this yet simply because I don't do HN analysis that often.
What that means practically is that the folks who post on this site are constantly changing. I also, generally, find that the most contentious threads on this site tend to have a relatively stronger recency bias among its posters.
By that regard, eternal September is ever present: the posts on this site weigh toward recent posters. I'd be curious to see if that effect can be explained by throwaways and sock puppet posters but I'm not sure of any reliable way to identify those especially as historic karma counts aren't kept for users.
While I have more robust models that work a lot better, a very simple method I've found is that the more recent the upper percentiles of posters are on a thread, the less I will like it: to me Eternal September is here. Of course my user here is ancient from 2009. The numbers just quantify that there's been a change in audience since I joined, a wholly unsurprising fact.
basically, quality(new user + time) = quality(old user), as long as the proportion of new users/old users remains small. of course it's subjective as to where you put this value though. i just think it hasn't been reached yet here.
I was watching The Synanon Fix on HBO, this weekend, and towards the end of the first episode, we see the original addicts, being replaced by the more lucrative "lifestylers," and I suspect that the next episodes will be downhill, from here.
The app I just wrote and released, is picking up steam, but slowly. That's deliberate. We're not doing any promotion, and it's giving us a chance to make course corrections. By the time people start piling in (and they will, but it will never be more than a rounding error to many social media apps), I think it will be in extremely good shape.
I have made some casual observations that it would be nice to know whether the broader data supports what I have seen. For instance, one weird thing that I have noticed, is that antagonistic comments appear to me to be posted more often by users with PGP keys in their profile. It may just be by imagination.
Possibly more importantly, but also something should be approached with caution and sensitivity would be analysis of comments to identify people in mental distress. I know of a few HN users that have taken their own lives and have wondered if there was enough information in their comments to allow someone to provide pre-emptive support.
Tbh, while I was working on this, I did struggle with forming hypotheses to test, or even a suitable manifold for the different items/users (e.g., I didn't know some of the users share the PGP keys). Unsupervised methods were not very satisfying in uncovering such hypotheses as well.
There is also the point you raised about how to share (or what to do with) some of these correlations. For example, concerning mental illness, as much as I would love to uncover that due to my sheer curiosity, I am really concerned about what I would see.
While it's definitely preferable to encrypt private communications among all people including the "nothing to hide" types who generally behave well within societal norms, the types who don't behave as well feel a greater need to encrypt.
My hypothesis is broadly similar to yours, that comments like this come from individuals with a more adversarial view of the world.
Brand new account commenting with very low value but not worth flagging. For example "I read the article yesterday" or "Why is this article even on HN?". I've seen a couple that had negative karma on the first days. Often though it's simply spam and the account soon gets shadow-banned though I don't understand the process that does that. Example user: https://news.ycombinator.com/user?id=gabriella4151
I think part of the confusion that I had was that I didn't know about the downvoting option at the beginning.
I think it will be really great of HN releases the scores of the comments.
I think the tradeoff there is that the HN submitter crowd are very fast to the draw. I'll often encounter a story elsewhere that I want to hear from HN about, so I'll do a search and find it was submitted 15 minutes ago, or even days ago (and I wonder what RSS fueled craze people must go through to find and post things so fast). Of course, for me those are the best cases, because I can go through and read what people think and reply without having to wait for the uptick. But no karma for the slow.
I do agree with your point, but unfortunately I was limited by what is available from the APIs: the upvotes for the comments were not provided. I couldn't find a proxy for it either. Thus, I couldn't include it in the analysis
You could say, there might be people who keep posting stories even though the don't get upvoted, but that would be kind of irrational. If the community doesn't seem to be interested in what they post, they will stop doing so sooner or later.
Probably many people stop sharing in this case indeed. But for those who don't, I guess the idea is: eventually, if you keep sharing, you can collect upvotes here and there, and eventually the karma will go up
On a similar note, sometimes i wonder where the high profile users who leave the platform move into (if anything), there doesn't seem to be a lot of options. Do they switch to alt accounts? turn into lurkers? maybe go off the grid?