News.YC statistics - community cohesion(clipboarded.blogspot.com)
clipboarded.blogspot.com
News.YC statistics - community cohesion
http://clipboarded.blogspot.com/2008/02/newsyc-statistics.html
11 comments
Just looked (again) at that link. I recall this page and recalling myself not getting far through it. On my revisit I still couldn't read it through entirely. It looks like one of those handouts that presents an unspecific problem, to which it throws a bunch of unspecific waypoints that lead to an unspecific solution.
I suspect you see something that I didn't; how do you approach reading this?
Also, between reading and coming back, you have been downvoted once. I have no idea what is downvote-worthy in your post... sometimes I just don't know what I am missing -- or if there is anything to miss in the first place.
I suspect you see something that I didn't; how do you approach reading this?
Also, between reading and coming back, you have been downvoted once. I have no idea what is downvote-worthy in your post... sometimes I just don't know what I am missing -- or if there is anything to miss in the first place.
I've wondered about that too - what motivates people to vote something up or down? Is voting up more a measure of interest, agreement, surprise, appreciation...? Would you vote down because you don't agree with a comment or because you think it's irrelevant, self-serving, unoriginal, or poorly-stated? (I didn't see anything I thought worthy of downvote on the parent comment either; I vote it up as thought-provoking even though I didn't quite get the point of the thing he linked to.)
Correct. :)
No, seriously, all the reasons you present a perfectly legitimate reasons for up/downvoting and are reasons that I have considered at various times.
No, seriously, all the reasons you present a perfectly legitimate reasons for up/downvoting and are reasons that I have considered at various times.
Is it even possible to vote something down? I don't see any down arrows. Maybe I don't have enough karma?
You can't downvote until you have enough karma, and a comment cannot be downvoted after it's been around for more than 24 hours. This is to prevent someone from vindictively going through a person's "threads" and downvoting all of their comments in an effort to kill their karma.
You can never downvote article submissions; they can only be upvoted. Downvoting sumbmissions has merit in some cases (removing spam), but in practice it has been used on other sites to kill everyone's submissions except your own, rapidly leading to a downmodding war where the winner is whoever is more committed to their karma score, instead of the one who is submitting the best articles.
You can never downvote article submissions; they can only be upvoted. Downvoting sumbmissions has merit in some cases (removing spam), but in practice it has been used on other sites to kill everyone's submissions except your own, rapidly leading to a downmodding war where the winner is whoever is more committed to their karma score, instead of the one who is submitting the best articles.
Interesting, thanks!
"How do you approach reading this?"
Normally people solve problems using some variant of the rational decision making problem. That is, define the problem, define assessment criteria, brainstorm solutions, choose solution, implement solution, measure solution against assessment criteria.
A wicked problem, on the other hand, is a problem that can't be clearly defined and has no clear answer. For example, what is the best way to design a car? If you maximize mpg then the car isn't as safe. But if you maximize safety then the car is heavier so the mpg isn't as good. And both safety and mpg affect style.
The idea of the idiagram thing is that while the rational decision making model may be the best way to solve well-bounded problems, it isn't a very good way to solve wicked problems.
This is roughly where the idiagram page leaves off, but it footnotes the book Wicked Problems and Social Complexity, the first chapter of which is available online. This explains dialog mapping, which is a method of discussing wicked problems in group meetings using mindmaps.
Why is this important?
If you've been to a lot of discussion forums on the web, you probably realize that forums with nested comments tend to have more rational discussion than forums with flat comments. After all, it's just easier to reply to specific points of an argument with nested comments than it is to fisk one with flat comments. You can go back and forth on one point, and then go back up in the discussion later and talk about a completely different point raised earlier.
Trying to solve wicked problems in real life meetings without dialog mapping is sort of the equivalent of an flat discussion on the web. You are sitting around and someone comes up with a complicated proposal. The next person responds to whatever they see in it. And then the other points in the original comment become more or less buried. However if you mindmap all of the points of a person's dialog, then people are able to use the visualization to jump back and forth during the discussion and gain perspective on what's important and what's less important. This ultimately means better solutions to wicked problems in less time, with more focus on the problem and less focus on the personalities involved.
The idiagram then is a way of visualization the final output of the brainstorming via dialog mapping process so that all involved can be on the same page when in comes to understanding the solution and each person's responsibilities and how success will be measured. (The whole submission is kind of recursive, because it's a picture about the four steps of the WP solving process, and making a picture is step three of four.)
Anyway, that's what I see in the submission. If you're the kind of person who sits are and thinks about how you can have brainstorming sessions in real life with the same abstract ideas and rational thinking that often arise online (because of the asynchronousness and nestedness of the internet), then this is kind of a cool thing.
Normally people solve problems using some variant of the rational decision making problem. That is, define the problem, define assessment criteria, brainstorm solutions, choose solution, implement solution, measure solution against assessment criteria.
A wicked problem, on the other hand, is a problem that can't be clearly defined and has no clear answer. For example, what is the best way to design a car? If you maximize mpg then the car isn't as safe. But if you maximize safety then the car is heavier so the mpg isn't as good. And both safety and mpg affect style.
The idea of the idiagram thing is that while the rational decision making model may be the best way to solve well-bounded problems, it isn't a very good way to solve wicked problems.
This is roughly where the idiagram page leaves off, but it footnotes the book Wicked Problems and Social Complexity, the first chapter of which is available online. This explains dialog mapping, which is a method of discussing wicked problems in group meetings using mindmaps.
Why is this important?
If you've been to a lot of discussion forums on the web, you probably realize that forums with nested comments tend to have more rational discussion than forums with flat comments. After all, it's just easier to reply to specific points of an argument with nested comments than it is to fisk one with flat comments. You can go back and forth on one point, and then go back up in the discussion later and talk about a completely different point raised earlier.
Trying to solve wicked problems in real life meetings without dialog mapping is sort of the equivalent of an flat discussion on the web. You are sitting around and someone comes up with a complicated proposal. The next person responds to whatever they see in it. And then the other points in the original comment become more or less buried. However if you mindmap all of the points of a person's dialog, then people are able to use the visualization to jump back and forth during the discussion and gain perspective on what's important and what's less important. This ultimately means better solutions to wicked problems in less time, with more focus on the problem and less focus on the personalities involved.
The idiagram then is a way of visualization the final output of the brainstorming via dialog mapping process so that all involved can be on the same page when in comes to understanding the solution and each person's responsibilities and how success will be measured. (The whole submission is kind of recursive, because it's a picture about the four steps of the WP solving process, and making a picture is step three of four.)
Anyway, that's what I see in the submission. If you're the kind of person who sits are and thinks about how you can have brainstorming sessions in real life with the same abstract ideas and rational thinking that often arise online (because of the asynchronousness and nestedness of the internet), then this is kind of a cool thing.
Thanks for the interesting feedback.
I don't know to what extent a difference it makes, because everything I see, I think is obvious (forgot the term for this): I think holistic and heuristical thinking is the de facto, and most obvious way of dealing with multidisciplinary problems; in this case, I believe this falls into the same category as the "wicked" problems you describe. It seems like everybody on this forum has a fair amount of exposure to psychology, such that many concepts that I learned in a formal setting are in fact common sense. I have no idea how true this is, but I try to avoid stating the obvious, so I generally keep quiet about it.
I still suspect the following would be nothing new to you, but it's a different route of thinking so I'll write anyway.
A problem that "cannot be clearly defined" and has "no clear answer" is generally solved heuristically. This is an area where hard-earned experience, and not necessarily genius, pays off. Warfare is a good example, because it involves not only logistics but also understanding the enemy's culture and psychology and estimating their intelligence. Heuristics come from efficient processing of declarative and episodic memory, which are, very obviously, the result of accumulated experiences across diverse fields.
So regarding how to best design a car, it is logical to think that an engineer would be better than an artist, because they know what is required to make a car work, but an engineer with an art background is better than one without, because they can efficiently (i.e. long-term memory retrieval) co-synthesize knowledge pertaining to design and construction. This must also be the reason why pg says "hackers who can design are dangerous."
Your points about forum dynamics follow the same ideas. A flat discussion, which interferes with thought organization, forces readers to synthesize information as a whole, which places a larger stress in keeping relevant information in memory. A threaded discussion, on the other hand, has less stress on memory, because certain steps can be stored as logical patterns that draw from existing memory.
A very simple thought experiment to illustrate this could be as follows. Imagine a threaded discussion with these messages: man. suitcase. taxi. airport. airplane. country. building. meeting.
Next, a flat discussion: man. window. smoke. lake. chocolate. tree. bulldozer.
Now paint a mental image out of the information from these two threads. The first follows a very logical progression, is easier to remember, and can be painted in a fairly predictable manner. The second one seems more disparate, requires more memorization, and the end result is harder to imagine, a la the "wicked problem."
Hope I'm not too far down a tangent... I guess the relevance to the diagram is, it doesn't matter what diagram or map you use; the goal is simply to get all the facts into everybody's heads simultaneously. A mind map could work. A list would work. A powerpoint. Or, based on the type of stimulus, a mixture of things (my interest).
I don't know to what extent a difference it makes, because everything I see, I think is obvious (forgot the term for this): I think holistic and heuristical thinking is the de facto, and most obvious way of dealing with multidisciplinary problems; in this case, I believe this falls into the same category as the "wicked" problems you describe. It seems like everybody on this forum has a fair amount of exposure to psychology, such that many concepts that I learned in a formal setting are in fact common sense. I have no idea how true this is, but I try to avoid stating the obvious, so I generally keep quiet about it.
I still suspect the following would be nothing new to you, but it's a different route of thinking so I'll write anyway.
A problem that "cannot be clearly defined" and has "no clear answer" is generally solved heuristically. This is an area where hard-earned experience, and not necessarily genius, pays off. Warfare is a good example, because it involves not only logistics but also understanding the enemy's culture and psychology and estimating their intelligence. Heuristics come from efficient processing of declarative and episodic memory, which are, very obviously, the result of accumulated experiences across diverse fields.
So regarding how to best design a car, it is logical to think that an engineer would be better than an artist, because they know what is required to make a car work, but an engineer with an art background is better than one without, because they can efficiently (i.e. long-term memory retrieval) co-synthesize knowledge pertaining to design and construction. This must also be the reason why pg says "hackers who can design are dangerous."
Your points about forum dynamics follow the same ideas. A flat discussion, which interferes with thought organization, forces readers to synthesize information as a whole, which places a larger stress in keeping relevant information in memory. A threaded discussion, on the other hand, has less stress on memory, because certain steps can be stored as logical patterns that draw from existing memory.
A very simple thought experiment to illustrate this could be as follows. Imagine a threaded discussion with these messages: man. suitcase. taxi. airport. airplane. country. building. meeting.
Next, a flat discussion: man. window. smoke. lake. chocolate. tree. bulldozer.
Now paint a mental image out of the information from these two threads. The first follows a very logical progression, is easier to remember, and can be painted in a fairly predictable manner. The second one seems more disparate, requires more memorization, and the end result is harder to imagine, a la the "wicked problem."
Hope I'm not too far down a tangent... I guess the relevance to the diagram is, it doesn't matter what diagram or map you use; the goal is simply to get all the facts into everybody's heads simultaneously. A mind map could work. A list would work. A powerpoint. Or, based on the type of stimulus, a mixture of things (my interest).
That's interesting. I've heard that one of the defining aspects of Western civilization's success is its emphasis on the written word, which is our most common form of linear, logical mind maps.
Where does lateral thinking come into this process, and has it been useful in your experience? I tend to think counter to the conventional wisdom, and people tend to think such ideas weird.
While it's normal to say this is a good thing, I'm not sure it is and am curious if there is any data on the usefulness of lateral thinking.
Where does lateral thinking come into this process, and has it been useful in your experience? I tend to think counter to the conventional wisdom, and people tend to think such ideas weird.
While it's normal to say this is a good thing, I'm not sure it is and am curious if there is any data on the usefulness of lateral thinking.
Agreed.
Let's say you post an amazingly funny technology joke and picture each day, and a friend posts an amazing, in-depth discussion of something that most of the readers would want to consume.
Who do you think is going to score all the points? Certainly not your friend with the in-depth articles.
So does that mean the the community is more represented by people who post quick, funny barbs? Are we a group of late-night technology comedians?
I understand the joke analogy is at the extreme, but you can make the same observation comparing, say, 3-paragraph gossipy blog entries with 50-paragraph in-depth articles. Same principle applies.
Let's say you post an amazingly funny technology joke and picture each day, and a friend posts an amazing, in-depth discussion of something that most of the readers would want to consume.
Who do you think is going to score all the points? Certainly not your friend with the in-depth articles.
So does that mean the the community is more represented by people who post quick, funny barbs? Are we a group of late-night technology comedians?
I understand the joke analogy is at the extreme, but you can make the same observation comparing, say, 3-paragraph gossipy blog entries with 50-paragraph in-depth articles. Same principle applies.
Funny you should mention that one - it happens to be something I regularly forward to people who like to snicker at bad uses of graphics. One guy I sent it to couldn't stop laughing for an hour.
What don't you like about it? It's obviously not something you'd put on a powerpoint slide, but given enough time to sit down and think about it I think it's quite insightful.
To me it seems kind of "fake" (I can't seem to find the right word in english - "Augenwischerei" in german) "kitsch" might be another word? The quality and fanciness of the graphics makes you believe that something is there, when probably there really isn't. At least, it looks rather complicated, and really good graphics should not make the thing they want to explain look complicated.
Since you mention that it actually provides insights after prolonged meditation, I wonder if it would be acceptable as a Mandala, though.
Since you mention that it actually provides insights after prolonged meditation, I wonder if it would be acceptable as a Mandala, though.
Only if you study the p3df.
It has all the bad qualities of a powerpoint presentation - lots of chartjunk and a very low data/ink ratio. Abstract pictures like these are a kind of IQ trap because the author never explains what many of the lines actually represent. Smart readers may find some way to make sense of them but it's unlikely that any two readers will be able to come up with compatible definitions that lead to useful analysis.
I think if Marshall Clemens expressed those 5 web pages as 5 paragraphs, you would see that he's either not really saying much or is saying something that has plenty of counterexamples that aren't apparent in the pretty picture.
For example, it's kind of ironic that I find none of the diagrams live up to the claims he makes about his Visual Models™ here http://www.idiagram.com/CP/visualmodels.html
I think if Marshall Clemens expressed those 5 web pages as 5 paragraphs, you would see that he's either not really saying much or is saying something that has plenty of counterexamples that aren't apparent in the pretty picture.
For example, it's kind of ironic that I find none of the diagrams live up to the claims he makes about his Visual Models™ here http://www.idiagram.com/CP/visualmodels.html
Awesome, thanks for that link. Occasionally I find myself staring down the barrel of a particularly thorny problem; the initial responses and challenges to the problem listed in the diagram are spot-on. My own internal approach to such problems pretty closely mimics what's in the diagram.
It's nice to have something that describes that whole process -- faced with a toughy problem? This is what to expect, and lest you forget, here's how you can approach it.
It's nice to have something that describes that whole process -- faced with a toughy problem? This is what to expect, and lest you forget, here's how you can approach it.
Interesting. I post lots of things that I find obscure and don't expect others to upmod.
I think part of the reason that news.yc is interesting is that there are a number of non-overlapping interest areas within a tight niche. Think about that. Diversity within a narrow field. I like that.
I think part of the reason that news.yc is interesting is that there are a number of non-overlapping interest areas within a tight niche. Think about that. Diversity within a narrow field. I like that.
Thanks for posting this interesting data set, but after thinking about it for some time, I don't think this data offers any insights into yc.news. The leader to user ratio is too high, which skews all the numbers so that nothing really meaningful can be extracted. It's also hard to spot any trends from the point system alone - user identity, time of post relative to other news, and the controversy of the post content can play important factors in the post's point value, and all of these don't necessarily have any correlation with the user who posted it. I think in order to figure out something about the community, someone will have to get their hands dirty and actually look at post content.
Those are nice stats, and I do think we can get some interesting characteristics about our community. I thought it is quite powerful to point to paul, palish, sharpshoot, or pg, and say that they are the community, since they have the best points per post measure. They are nothing less than the best representatives of our community (with regards to posts, not comments).
I'd like to think that the factors you mentioned (traffic, time of the day, controversy of the post) iron out statistically (since all posts are affected by those, there is an argument to ignore their respective influence).
This initiative makes me wish it is only the beginning: I want some more stats! For example:
- It would be interesting to see what's the points per post (PPP) of the leaders, compared to the PPP for most of the others (I agree with you that in the dataset, the leader to user ratio is too high). Maybe one could have a plotted distribution of PPP's across all users.
- I know HN has 8000 uniques/day, what is the proportion of those that are posting? e.g. how many of those users with less than 5 posts?
- also interesting is the proportion of points between posts and comments, per user. Different profiles: some people discuss a lot, some people post a lot, as it was often repeated in discussions. But what's the proportion of those 2 groups?
I'd like to think that the factors you mentioned (traffic, time of the day, controversy of the post) iron out statistically (since all posts are affected by those, there is an argument to ignore their respective influence).
This initiative makes me wish it is only the beginning: I want some more stats! For example:
- It would be interesting to see what's the points per post (PPP) of the leaders, compared to the PPP for most of the others (I agree with you that in the dataset, the leader to user ratio is too high). Maybe one could have a plotted distribution of PPP's across all users.
- I know HN has 8000 uniques/day, what is the proportion of those that are posting? e.g. how many of those users with less than 5 posts?
- also interesting is the proportion of points between posts and comments, per user. Different profiles: some people discuss a lot, some people post a lot, as it was often repeated in discussions. But what's the proportion of those 2 groups?
There is a lot of talk about what should be done in order to keep the community as it is, about what is a relevant post and what's not. But we have no metric to measure the cohesion, except for 'pleasure'. The analysis in the post is of course incomplete but I believe there is value in a metric that could tell us whether or not we are being steered away from the initial community.
Also, pg could probably run this on the complete data set, which might better illustrate what's happening with the more prolific members of the community.
Nice analysis. Does the complete data include timestamps on up/down votes? The ratio of up to down votes might show how controversial is a posting while the rate of rise/fall might show... whether the posting was submitted at an opportune time?
Or what about looking at voting habits? What's the average/min/max/etc up/down voting ratio?
If you showed to what degree showing more statistics, particularly about new postings or first-page postings would change behavior?
Or what about looking at voting habits? What's the average/min/max/etc up/down voting ratio?
If you showed to what degree showing more statistics, particularly about new postings or first-page postings would change behavior?
Fantastic! A digitally generated picture is worth 2^10 words!
You've certainly embodied the philosophy of Edward Tufte:
http://www.amazon.com/s/ref=nb_ss_gw/102-8193666-1218533?url...
There must be many cofactors that are hard to capture in an analysis like this. There's the "binaryness" of making it to the front page. There's also "time of day". A very interesting submission posted at a time when those who would be interested are not online may never gain enough traction to make it to the first page. A time-of-day analysis would certainly be interesting.
You've certainly embodied the philosophy of Edward Tufte:
http://www.amazon.com/s/ref=nb_ss_gw/102-8193666-1218533?url...
There must be many cofactors that are hard to capture in an analysis like this. There's the "binaryness" of making it to the front page. There's also "time of day". A very interesting submission posted at a time when those who would be interested are not online may never gain enough traction to make it to the first page. A time-of-day analysis would certainly be interesting.
I'm not sure if you're mocking me or giving me too much credit here..
Really I was hoping to be more rigorous and go more in depth but at some point it just felt silly to hammer a website, scrape the page and never have access to the whole data. Doing this got me thinking about issues like "time of day" though as there must be a way to take into account the rate of new posts and make old posts more 'sticky' when the rate accelerates, in the morning for instance.
I feel like starting my own social news website just to have a play with different algorithms!
I feel like starting my own social news website just to have a play with different algorithms!
"I'm not sure if you're mocking me or giving me too much credit here."
Neither, I'm sure.
(Funny. lately when I get sarcastic here, people think I'm serious, and when I'm serious, people think I'm joking. I guess I need a little more practice writing to strangers.)
Scatter diagrams are one of the greatest underrated ways of conveying information in a heartbeat. You've done a great job of it.
You've also opened up a can of worms. There are so many ways to look at this data, you can keep yourself busy for quite a while. I suspect that won't be a problem, considering how much you already care about this. Keep up the good work and keep us posted.
Neither, I'm sure.
(Funny. lately when I get sarcastic here, people think I'm serious, and when I'm serious, people think I'm joking. I guess I need a little more practice writing to strangers.)
Scatter diagrams are one of the greatest underrated ways of conveying information in a heartbeat. You've done a great job of it.
You've also opened up a can of worms. There are so many ways to look at this data, you can keep yourself busy for quite a while. I suspect that won't be a problem, considering how much you already care about this. Keep up the good work and keep us posted.
Sarcasm is hard to convey in writing. We used to have smileys to help us but they are going out of fashion. We're gonna have to stop being sarcastic :)
This kind of topic worries me because from my experience when a community starts worrying about "cohesion", "trolls" and like, this is usually the first sign of it's fall down.
The real problem is what most of us would call a fall down today, that will be called tommorow (and interpreted as) a rise up from other people (thus the extreme difficulty to create efficient moderating systems based on quantitative data). You know your community is irreversibly "corrupted" when this kind of people becomes the majority, and this usually happen surprisingly quickly with the faster growth of visitors.
For some reason I have some hope for HN because of PG being behind it, he kinda knows what he wants and certainly knows what he doesn't want. Still, this might be one of the thoughest challenge he'll be faced with if HN keeps growing. I've known no community that escaped this syndrom yet.
The real problem is what most of us would call a fall down today, that will be called tommorow (and interpreted as) a rise up from other people (thus the extreme difficulty to create efficient moderating systems based on quantitative data). You know your community is irreversibly "corrupted" when this kind of people becomes the majority, and this usually happen surprisingly quickly with the faster growth of visitors.
For some reason I have some hope for HN because of PG being behind it, he kinda knows what he wants and certainly knows what he doesn't want. Still, this might be one of the thoughest challenge he'll be faced with if HN keeps growing. I've known no community that escaped this syndrom yet.
As a relatively newer member here, it would be interesting to know how these numbers came about over time; i.e, did the more prolific members start out with a bang, or did they gradually build up steam to reach where they are now?
Cool.
I'd be interested in seeing similar analysis on a per-comment basis.
I'd be interested in seeing similar analysis on a per-comment basis.
Very cool. Was fun trying to find my username on your charts (although I wish it could have been a bit higher up the points per post axis!).
nice little bit of hacking, one thing I would of done is left pg out, I think he is messing your data up. Fans being fans his karma is off the chart.
Yeah his approval rating is ridiculous, above 13 points per post on his last 180 posts. But that's just one of many things messing up the statistics. It's only good if you want to have a general idea.
These plots are begging to be placed on a log-log axis!
http://www.idiagram.com/CP/cpprocess.html
Looking through my own submissions I see that the more intellectual stuff I've posted, while it's done well, has never gotten as many votes as the pithy one-liners.