Favorites begetting favorites April 22, 2013 1:59 PM Subscribe
I have two random thoughts about the inner workings of metafilter, that I would be curious to know the answer to:
1. Has anyone looked at if there is a "critical mass" of "favourites." I.E. at what point does having a certain number favourites result other people seeing the comment, clicking it and favouriting it?
2. How many hits on overage will a FPP link cause for a target site?
Back when Geocities was still around, we were known to bring Geocities pages down ("violation of terms of service, too much traffic").
posted by Chocolate Pickle at 2:18 PM on April 22, 2013 [1 favorite]
posted by Chocolate Pickle at 2:18 PM on April 22, 2013 [1 favorite]
Of course, things like stiff breezes, sudden loud noises, and smallish trucks driving by could bring down a Geocities page
posted by cortex (staff) at 2:28 PM on April 22, 2013 [21 favorites]
posted by cortex (staff) at 2:28 PM on April 22, 2013 [21 favorites]
There's also a greasemonkey script which adds a blue bar to the left of every comment, and the more favorites the comment has, the thicker the bar is. This obviously highlights the most-favorited comments, though I suspect that the "critical mass" amount varies depending on how many comments are in the thread - the more comments, the more favorites any one comment would need to stand out.
posted by prefpara at 2:32 PM on April 22, 2013 [2 favorites]
posted by prefpara at 2:32 PM on April 22, 2013 [2 favorites]
When my (fairly niche) blog was linked on the blue I got about 2,000 excess hits in the following 24 hours*.
I made a FPP once to a site where I subsequently corresponded with the owner and they said they got about 8,000 hits from that link, but that was to a photo gallery where a unique visitor might account for 50 page views. On the other hand, that was also when there might have been "only" about 30,000 members here (though I realize there are a lot of lurkers.
*when it got linked in a comment on a front page reddit thread it got 8,000 hits in like an hour.
posted by Rumple at 3:09 PM on April 22, 2013 [1 favorite]
I made a FPP once to a site where I subsequently corresponded with the owner and they said they got about 8,000 hits from that link, but that was to a photo gallery where a unique visitor might account for 50 page views. On the other hand, that was also when there might have been "only" about 30,000 members here (though I realize there are a lot of lurkers.
*when it got linked in a comment on a front page reddit thread it got 8,000 hits in like an hour.
posted by Rumple at 3:09 PM on April 22, 2013 [1 favorite]
My anecdotal experience suggests that the tipping point is around 6. People seem to like a lot of the things I say, and so I get a hefty number of favorites. Generally, it seems like things either get a handful of favorites or, like, more than 20.
I figure 5 or 6 is about the point where someone scrolling through might look at it and say, "hey, several people think that's good, I should read that," where as 2 or 3 just doesn't really stand out above the fray.
I have crunched no numbers in determining this, it just sort of seems like that's the general trend I've noticed in my own.
posted by phunniemee at 3:13 PM on April 22, 2013 [3 favorites]
I figure 5 or 6 is about the point where someone scrolling through might look at it and say, "hey, several people think that's good, I should read that," where as 2 or 3 just doesn't really stand out above the fray.
I have crunched no numbers in determining this, it just sort of seems like that's the general trend I've noticed in my own.
posted by phunniemee at 3:13 PM on April 22, 2013 [3 favorites]
I've had two projects make it to the front. I saw about 2,000 extra hits. There was some residual clicks over the next X days and I still occasionally get a click from an old link.
I think sometimes people overestimate the traffic metafilter generates (why so many spammers try). I think from a SEO standpoint metafilter is a "quality" site, so people want the links.
posted by cjorgensen at 3:39 PM on April 22, 2013
I think sometimes people overestimate the traffic metafilter generates (why so many spammers try). I think from a SEO standpoint metafilter is a "quality" site, so people want the links.
posted by cjorgensen at 3:39 PM on April 22, 2013
Favorite my comment if you hate cancer.
Ignore if u dont have a heart
posted by mazola at 4:48 PM on April 22, 2013 [4 favorites]
Ignore if u dont have a heart
posted by mazola at 4:48 PM on April 22, 2013 [4 favorites]
The favorite inequality is really mind blowing. You have people holding on to 100k favs while the rest of us scrape by with barely 15k.
I've also found that while membership has gone up, favorite distribution to the majority of users has remained stagnant. More and more favorites are flowing to fewer and fewer, with the top 1% garning the lions share of favorites.
I wouldn't say I'm in favor of redistribution, but the majority of us are getting left behind, once again proving that the myth that with hard work anyone can make it, is just that, a myth.
posted by Ad hominem at 5:17 PM on April 22, 2013 [23 favorites]
I've also found that while membership has gone up, favorite distribution to the majority of users has remained stagnant. More and more favorites are flowing to fewer and fewer, with the top 1% garning the lions share of favorites.
I wouldn't say I'm in favor of redistribution, but the majority of us are getting left behind, once again proving that the myth that with hard work anyone can make it, is just that, a myth.
posted by Ad hominem at 5:17 PM on April 22, 2013 [23 favorites]
I am not sure about the numbers or how it exactly works, but I can say that I get my most favorites when I am not trying. Try and write a post or comment to get favorites and you will fail miserably, but write from your heart and the favorites flow like water or at least honey.
posted by JohnnyGunn at 5:25 PM on April 22, 2013 [2 favorites]
posted by JohnnyGunn at 5:25 PM on April 22, 2013 [2 favorites]
Try too hard for favourites and you'll wind up like Del Boy. Play it cool, my son.
posted by arcticseal at 5:58 PM on April 22, 2013
posted by arcticseal at 5:58 PM on April 22, 2013
Something I'd like to see is: comparing two comments of identical content, does the comment by the more well-known poster get favorited more? In other words, you have a jokey comment on the blue, and to 50% of users, you show the user name of the more well known poster, and to the other 50%, show the user name of an obscure poster. Remove the effect of sidebar/contact activity. Does the more well known poster get more favorites? I suspect yes, but I have no idea how you would analyze this from the infodump data. Anecdotally I have only seen this happen once, where two people made the exact same joke in a thread, the well-known poster commented a bit later and received all the favorites, but I can't find it. An interesting experiment would be a period where usernames are hidden but favorite counts left on, and see if the most heavily favorited posters (both by absolute amount and rate of favorite accrual) continue to receive the same rate of favorites.
Another thing that would be interesting is looking at which users favorite each other most heavily and if it is reciprocal - again, somehow removing the effect of contact activity, and seeing if certain users clump together in their favoriting patterns.
It'd also be interesting if anyone has done analysis on favoriting behavior during that experiment a while back where the counts were forced off for everybody - I'd guess that both volumes were down significantly and there was less clustering, but maybe there's something else too.
posted by pravit at 6:03 PM on April 22, 2013
Another thing that would be interesting is looking at which users favorite each other most heavily and if it is reciprocal - again, somehow removing the effect of contact activity, and seeing if certain users clump together in their favoriting patterns.
It'd also be interesting if anyone has done analysis on favoriting behavior during that experiment a while back where the counts were forced off for everybody - I'd guess that both volumes were down significantly and there was less clustering, but maybe there's something else too.
posted by pravit at 6:03 PM on April 22, 2013
I used to be top 10 in number of comments but had the lowest favorite to comment ratio of the top 10 commenters. I like to think I am like the bad news bears of commenting, what I lack in fundamentals and polish I make up for with heart. Maybe I am more like the punk rock of commenters though. Or the bad news bears of punk rock.
posted by Ad hominem at 6:14 PM on April 22, 2013 [3 favorites]
posted by Ad hominem at 6:14 PM on April 22, 2013 [3 favorites]
You've got to get in on the ground floor. Even my most thoughtful comments that appear late in threads that are winding down get little attention, but the zingers 5 or 6 comments into a new thread will gather them pretty quickly. I've never noticed a rate of acceleration on heavily-favorited comments, just the trail-off.
I get the impression that I would be much more highly favorited if I were quicker on my feet, in general.
posted by Devils Rancher at 6:42 PM on April 22, 2013 [3 favorites]
I get the impression that I would be much more highly favorited if I were quicker on my feet, in general.
posted by Devils Rancher at 6:42 PM on April 22, 2013 [3 favorites]
For me, it's not that a favoriteball, like, persuades me to add a favorite of my own; it's more that if I'm lazily scrolling through a thread, not really reading, my eye will snag on an instance of [796 favorites +] and I am likely to read that comment when I otherwise wouldn't have -- and then favorite it, assuming the reason it had amassed all those favorites in the first place is that it was a really good comment. Maybe this effect is in play for others?
posted by threeants at 6:48 PM on April 22, 2013
posted by threeants at 6:48 PM on April 22, 2013
There are no breakpoints/critical masses for favoriting, it's a smooth curve.
(comments with 1-10 favorites go off the top of the graph but they are also a smooth line. In addition I truncated the graph at comments with 173 favorites or more because that's when data starts getting relatively sparse.)
Oh yeah, the source data is the 9 million favorites given since 01-01-10.
You've got to get in on the ground floor. Even my most thoughtful comments that appear late in threads that are winding down get little attention, but the zingers 5 or 6 comments into a new thread will gather them pretty quickly. I've never noticed a rate of acceleration on heavily-favorited comments, just the trail-off.
I've been meaning to measure that phenomenon. And I'm bored. Good combo.
posted by Tell Me No Lies at 6:48 PM on April 22, 2013 [5 favorites]
(comments with 1-10 favorites go off the top of the graph but they are also a smooth line. In addition I truncated the graph at comments with 173 favorites or more because that's when data starts getting relatively sparse.)
Oh yeah, the source data is the 9 million favorites given since 01-01-10.
You've got to get in on the ground floor. Even my most thoughtful comments that appear late in threads that are winding down get little attention, but the zingers 5 or 6 comments into a new thread will gather them pretty quickly. I've never noticed a rate of acceleration on heavily-favorited comments, just the trail-off.
I've been meaning to measure that phenomenon. And I'm bored. Good combo.
posted by Tell Me No Lies at 6:48 PM on April 22, 2013 [5 favorites]
There are no breakpoints/critical masses for favoriting, it's a smooth curve.
More to the point, it's a "power law curve", like so many other things on the web.
posted by Chocolate Pickle at 7:11 PM on April 22, 2013
More to the point, it's a "power law curve", like so many other things on the web.
posted by Chocolate Pickle at 7:11 PM on April 22, 2013
I figure 5 or 6 is about the point where someone scrolling through might look at it and say, "hey, several people think that's good, I should read that," where as 2 or 3 just doesn't really stand out above the fray.
I have crunched no numbers in determining this, it just sort of seems like that's the general trend I've noticed in my own.
posted by phunniemee at 3:13 PM on April 22 [2 favorites −] Favorite added! [!]
I can confirm that if I'm pressed for time at the moment or am not really interested in the discussion at hand, these are the figures I use to discriminate while scrolling which comments I should read.
posted by SollosQ at 7:41 PM on April 22, 2013
I have crunched no numbers in determining this, it just sort of seems like that's the general trend I've noticed in my own.
posted by phunniemee at 3:13 PM on April 22 [2 favorites −] Favorite added! [!]
I can confirm that if I'm pressed for time at the moment or am not really interested in the discussion at hand, these are the figures I use to discriminate while scrolling which comments I should read.
posted by SollosQ at 7:41 PM on April 22, 2013
How many hits on overage will a FPP link cause for a target site?
I linked to a mix on soundcloud the other day and I think it's only gone up by about 500 plays since then, so... probably not as much as you'd think. Really depends on the site...
posted by empath at 8:06 PM on April 22, 2013
I linked to a mix on soundcloud the other day and I think it's only gone up by about 500 plays since then, so... probably not as much as you'd think. Really depends on the site...
posted by empath at 8:06 PM on April 22, 2013
I think we should just make favorites binary in a sense, where it's either favorited or it isn't (maybe count 1,2 and stop there), and display something less "word-y" than "favorites..." Maybe "faved?"
But I have shamefully been thinking about favorite-maximization strategies lately. Fortunately I have not implemented any or you'd all be in stitches. Stitches on your fingers, because they were split from hitting your favorite button too hard, because I had you in the other metaphorical sort of stitches. I'm also considering practicing my poignancy and am researching a career based on an analysis of what professional perspective I think MetaFilter might benefit most from.
posted by lordaych at 8:09 PM on April 22, 2013 [1 favorite]
But I have shamefully been thinking about favorite-maximization strategies lately. Fortunately I have not implemented any or you'd all be in stitches. Stitches on your fingers, because they were split from hitting your favorite button too hard, because I had you in the other metaphorical sort of stitches. I'm also considering practicing my poignancy and am researching a career based on an analysis of what professional perspective I think MetaFilter might benefit most from.
posted by lordaych at 8:09 PM on April 22, 2013 [1 favorite]
I think we should just make favorites binary in a sense, where it's either favorited or it isn't (maybe count 1,2 and stop there), and display something less "word-y" than "favorites..." Maybe "faved?"
Click on Preferences, scroll down to Comment Favorites Style, set it to Show "has favorites."
posted by Etrigan at 8:18 PM on April 22, 2013
Click on Preferences, scroll down to Comment Favorites Style, set it to Show "has favorites."
posted by Etrigan at 8:18 PM on April 22, 2013
Etrigan has the short answer for you, lordaych; the long answer is oh boy do I have a 2700+ comment metatalk thread for you to read.
posted by cortex (staff) at 8:35 PM on April 22, 2013 [2 favorites]
posted by cortex (staff) at 8:35 PM on April 22, 2013 [2 favorites]
Speaking of favorites styles, is there a reason that the "add to favorites" function appears variously as [+], + (no brackets) and [add to favorites]?
posted by zeptoweasel at 9:00 PM on April 22, 2013
posted by zeptoweasel at 9:00 PM on April 22, 2013
Helps us root out the cylons.
posted by cortex (staff) at 9:15 PM on April 22, 2013 [2 favorites]
posted by cortex (staff) at 9:15 PM on April 22, 2013 [2 favorites]
More to the point, it's a "power law curve", like so many other things on the web.
Is it really, though?
posted by en forme de poire at 9:22 PM on April 22, 2013 [2 favorites]
Is it really, though?
posted by en forme de poire at 9:22 PM on April 22, 2013 [2 favorites]
write from your heart and the favorites flow like water or at least honey.
Or, for me, more like this.
posted by dg at 9:44 PM on April 22, 2013 [1 favorite]
Or, for me, more like this.
posted by dg at 9:44 PM on April 22, 2013 [1 favorite]
The favorite inequality is really mind blowing. You have people holding on to 100k favs while the rest of us scrape by with barely 15k.
posted by Ad hominem at 1:17 AM on April 23 [5 favorites −] Favorite added! [!]
Come the glorious day, all those with more than 20k favourites will be first against the wall.
posted by Decani at 10:02 PM on April 22, 2013 [3 favorites]
posted by Ad hominem at 1:17 AM on April 23 [5 favorites −] Favorite added! [!]
Come the glorious day, all those with more than 20k favourites will be first against the wall.
posted by Decani at 10:02 PM on April 22, 2013 [3 favorites]
I'm really surprised that only 2,000 people are clicking on an FPP link. Seems remarkably low, considering Mefi's high profile.
How does the ~2,000 visits to an FPP link compare with the total unique visits to the Blue on a particular day?
And are the 2,000 mainly members or non-members I wonder?
posted by dontjumplarry at 10:03 PM on April 22, 2013
How does the ~2,000 visits to an FPP link compare with the total unique visits to the Blue on a particular day?
And are the 2,000 mainly members or non-members I wonder?
posted by dontjumplarry at 10:03 PM on April 22, 2013
I'm really surprised that only 2,000 people are clicking on an FPP link
Everyone else is busy writing snarky comments instead ;-)
posted by dg at 10:09 PM on April 22, 2013
Everyone else is busy writing snarky comments instead ;-)
posted by dg at 10:09 PM on April 22, 2013
The favorite inequality is really mind blowing. You have people holding on to 100k favs while the rest of us scrape by with barely 15k.
What percentage of the favorites do the top 1% have? The top 10%?
Serious question.
posted by empath at 10:14 PM on April 22, 2013
What percentage of the favorites do the top 1% have? The top 10%?
Serious question.
posted by empath at 10:14 PM on April 22, 2013
I'm really surprised that only 2,000 people are clicking on an FPP link
I'm surprised 2,000 people actually RTFA before posting.
posted by mazola at 10:47 PM on April 22, 2013 [1 favorite]
I'm surprised 2,000 people actually RTFA before posting.
posted by mazola at 10:47 PM on April 22, 2013 [1 favorite]
Right, so there are a couple of notable things in the data:
1) Surprisingly, there is at least a 20 minute ramp up before sweet spot is hit.
2) The ramp up is noticeably shorter in AskMe. My guess that is because a) an AskMe post seldom takes long to read or process, and b) a lot of the time someone knows the exact information needed off the top of their head.
Number of favorites selected in a post by 5 minute intervals
posted by Tell Me No Lies at 11:52 PM on April 22, 2013 [1 favorite]
1) Surprisingly, there is at least a 20 minute ramp up before sweet spot is hit.
2) The ramp up is noticeably shorter in AskMe. My guess that is because a) an AskMe post seldom takes long to read or process, and b) a lot of the time someone knows the exact information needed off the top of their head.
Number of favorites selected in a post by 5 minute intervals
posted by Tell Me No Lies at 11:52 PM on April 22, 2013 [1 favorite]
I'd like to know if there have been identified attempts to game the favorites. Social media manipulation is big business. Lots of people working to influence the popularity of things, and working to provide paid content response to things. It's got to be happening here just as much as on other social media.
posted by five fresh fish at 11:56 PM on April 22, 2013
posted by five fresh fish at 11:56 PM on April 22, 2013
I really doubt it. Smaller community, no real benefit to having a lot of favorites -- it's not like Digg or Reddit where the front page or top comments are determined by the amount of favorites.
posted by empath at 12:24 AM on April 23, 2013
posted by empath at 12:24 AM on April 23, 2013
arcticseal: "Try too hard for favourites and you'll wind up like Del Boy . Play it cool, my son."
Now let's deconstruct that a bit, shall we.
posted by goodnewsfortheinsane (staff) at 12:59 AM on April 23, 2013 [1 favorite]
Now let's deconstruct that a bit, shall we.
posted by goodnewsfortheinsane (staff) at 12:59 AM on April 23, 2013 [1 favorite]
I’m surprised that people notice the amount of favorites. My first thought was that the premise of the original question was ridiculous, nobody does those things, but I guess I’m wrong.
I never look at the number of favorites unless I’m already clicking it. If I don’t click it I never notice the number. I rarely look at the sidebar or any other highlighting area. The name wouldn’t influence me to be more likely to favorite because I don’t know who those people are. I guess if it was a friend I actually knew it might.
posted by bongo_x at 1:16 AM on April 23, 2013
I never look at the number of favorites unless I’m already clicking it. If I don’t click it I never notice the number. I rarely look at the sidebar or any other highlighting area. The name wouldn’t influence me to be more likely to favorite because I don’t know who those people are. I guess if it was a friend I actually knew it might.
posted by bongo_x at 1:16 AM on April 23, 2013
I think people are primarily talking about the number of favorites for comments, not posts.
posted by Tell Me No Lies at 1:48 AM on April 23, 2013 [1 favorite]
posted by Tell Me No Lies at 1:48 AM on April 23, 2013 [1 favorite]
I’m surprised that people notice the amount of favorites. My first thought was that the premise of the original question was ridiculous, nobody does those things, but I guess I’m wrong.
Well, yeah. Thing to remember is that if you're an (over) active participant you're probably reading all the comments, at least on those threads of interest to you, but there will be a larger group of less active people who skim read; and if you do that, anything that stands out, like the favourites number, will cause you to take a closer look.
posted by MartinWisse at 4:26 AM on April 23, 2013 [1 favorite]
Well, yeah. Thing to remember is that if you're an (over) active participant you're probably reading all the comments, at least on those threads of interest to you, but there will be a larger group of less active people who skim read; and if you do that, anything that stands out, like the favourites number, will cause you to take a closer look.
posted by MartinWisse at 4:26 AM on April 23, 2013 [1 favorite]
Come the glorious day, all those with more than 20k favourites will be first against the wall.
Fortunately the dearly departed Lady Thatcher saved us from the politics of envy like this.
posted by MartinWisse at 4:28 AM on April 23, 2013
Fortunately the dearly departed Lady Thatcher saved us from the politics of envy like this.
posted by MartinWisse at 4:28 AM on April 23, 2013
Come the glorious day, all those with more than 20k favourites will be first against the wall.
Let them click [+].
posted by robocop is bleeding at 4:29 AM on April 23, 2013 [1 favorite]
Let them click [+].
posted by robocop is bleeding at 4:29 AM on April 23, 2013 [1 favorite]
Fortunately, come the Revolution, I've converted all my favourites into easily transportable LOLcats.
posted by arcticseal at 5:07 AM on April 23, 2013 [1 favorite]
posted by arcticseal at 5:07 AM on April 23, 2013 [1 favorite]
Back in 2009, one of my projects was promoted to the front page. Starting from basically a handful of hits per day (it was a site shared among my daughter's family and friends), it picked up a bit after being posted on Projects (and as it started to add regular content) - about 80 - 100 visits per day for a few days. On the day it was posted to the front page, visits jumped to over 3,500 for that day. It ramped down rather quickly over the next few days, but it did maintain an average of about 80 visits per day for the rest of the summer (after which the project ended).
posted by mikepop at 5:50 AM on April 23, 2013
posted by mikepop at 5:50 AM on April 23, 2013
Back when Geocities was still around, we were known to bring Geocities pages down ("violation of terms of service, too much traffic").
posted by Chocolate Pickle at 4:18 PM on April 22
[+] [!]
This phenomenon has always puzzled me. Since Geocities served ads, and in fact the whole purpose of giving people free websites was to make money on the ads that would appear on the websites, how in the hell did it make sense for "too much traffic" to ever violate terms of service or cause them to suspend your website?
posted by Unified Theory at 6:09 AM on April 23, 2013 [1 favorite]
posted by Chocolate Pickle at 4:18 PM on April 22
[+] [!]
This phenomenon has always puzzled me. Since Geocities served ads, and in fact the whole purpose of giving people free websites was to make money on the ads that would appear on the websites, how in the hell did it make sense for "too much traffic" to ever violate terms of service or cause them to suspend your website?
posted by Unified Theory at 6:09 AM on April 23, 2013 [1 favorite]
Come the glorious day, all those with more than 20k favourites will be first against the wall.
Heh.
For most of us its probably just a matter of how active we are. My own count seems high, but if you add up all my activity it comes to a little more than 1 favorite per post or comment. It's skewed further because I've made a lot of fpps, and those can attract favorites more easily. (Greater visibility, etc.)
posted by zarq at 6:34 AM on April 23, 2013
Heh.
For most of us its probably just a matter of how active we are. My own count seems high, but if you add up all my activity it comes to a little more than 1 favorite per post or comment. It's skewed further because I've made a lot of fpps, and those can attract favorites more easily. (Greater visibility, etc.)
posted by zarq at 6:34 AM on April 23, 2013
I just traded 10,000 favorites for a bitcoin.
posted by Devils Rancher at 6:51 AM on April 23, 2013
posted by Devils Rancher at 6:51 AM on April 23, 2013
...if you add up all my activity it comes to a little more than 1 favorite per post or comment.
zarq currently has 36562 favorites. He's made 15210 comments or posts across all the sites.
That's actually 2.40381328073636 favorites per post/comment.
He's either modest or bad at math.
posted by cjorgensen at 7:24 AM on April 23, 2013
zarq currently has 36562 favorites. He's made 15210 comments or posts across all the sites.
That's actually 2.40381328073636 favorites per post/comment.
He's either modest or bad at math.
posted by cjorgensen at 7:24 AM on April 23, 2013
We're set up at my place to grow our own new favorites from rainwater and composted deleted comments. Plus, we have years' worth of pickled favorites in the basement. Come the revolution, my Mefi Spouses and I will be fine.
posted by Slarty Bartfast at 7:32 AM on April 23, 2013
posted by Slarty Bartfast at 7:32 AM on April 23, 2013
I just booted up my BitFavorite mining cluster.
posted by Rock Steady at 7:34 AM on April 23, 2013 [1 favorite]
posted by Rock Steady at 7:34 AM on April 23, 2013 [1 favorite]
cjorgensen: " He's either modest or bad at math."
I'm bad at math. :)
posted by zarq at 7:36 AM on April 23, 2013
I'm bad at math. :)
posted by zarq at 7:36 AM on April 23, 2013
I'd like to know if there have been identified attempts to game the favorites. Social media manipulation is big business. Lots of people working to influence the popularity of things, and working to provide paid content response to things. It's got to be happening here just as much as on other social media.
Nothing worth noting, really. I have for sure on a couple occasions noticed favoriting (or, more visibly, Projects voting) for one newish account by another newish account that have basically zero other community involvement and clearly have a relationship based on what we can see, so there's that sort of penny-ante scheming every once in a while. But given how mefi works, it's a doomed effort. You'd have to invest in a whole bunch of accounts—fifty? A hundred?—in order to generate numbers that were even eye-catching in the first place, and if they were eye-catching on something that felt not quite right (which they almost certainly would be if you needed to aggressively and artificially pump something up) that'd basically be a big pointer saying HEY MODS THIS IS WEIRD TAKE A CLOSE LOOK AT THIS and then you're out fifty or a hundred accounts.
That's assuming you were even subtle enough to carefully create those dozens of accounts each with unique paypal information and spread out over a long enough period to not be conspicuous just in their numbers.
The amount of time, money, and effort required to just get yourself banned en masse as soon as you actually pulled the trigger makes the whole thing deeply implausible. The closest I can think of in memory to a stunt like that was the Jamon Iberico people, who signed up like 12 accounts about different kinds of meat all in one morning after getting banned the first time, and that was needless to say not a successfully sly approach.
posted by cortex (staff) at 7:40 AM on April 23, 2013 [4 favorites]
Nothing worth noting, really. I have for sure on a couple occasions noticed favoriting (or, more visibly, Projects voting) for one newish account by another newish account that have basically zero other community involvement and clearly have a relationship based on what we can see, so there's that sort of penny-ante scheming every once in a while. But given how mefi works, it's a doomed effort. You'd have to invest in a whole bunch of accounts—fifty? A hundred?—in order to generate numbers that were even eye-catching in the first place, and if they were eye-catching on something that felt not quite right (which they almost certainly would be if you needed to aggressively and artificially pump something up) that'd basically be a big pointer saying HEY MODS THIS IS WEIRD TAKE A CLOSE LOOK AT THIS and then you're out fifty or a hundred accounts.
That's assuming you were even subtle enough to carefully create those dozens of accounts each with unique paypal information and spread out over a long enough period to not be conspicuous just in their numbers.
The amount of time, money, and effort required to just get yourself banned en masse as soon as you actually pulled the trigger makes the whole thing deeply implausible. The closest I can think of in memory to a stunt like that was the Jamon Iberico people, who signed up like 12 accounts about different kinds of meat all in one morning after getting banned the first time, and that was needless to say not a successfully sly approach.
posted by cortex (staff) at 7:40 AM on April 23, 2013 [4 favorites]
So what you are saying...the long game!
Challenge accepted!
posted by cjorgensen at 7:51 AM on April 23, 2013
Challenge accepted!
posted by cjorgensen at 7:51 AM on April 23, 2013
u can buy hamon iberigo at peters shop
posted by griphus at 8:07 AM on April 23, 2013 [1 favorite]
posted by griphus at 8:07 AM on April 23, 2013 [1 favorite]
peters shop is just giving away orange juice coupons. lol
posted by Rock Steady at 8:09 AM on April 23, 2013
posted by Rock Steady at 8:09 AM on April 23, 2013
That was just the feint, cortex, we have been steadily building our impregnable jamon iberico faction here for almost a decade.
posted by Meatbomb at 8:16 AM on April 23, 2013 [2 favorites]
posted by Meatbomb at 8:16 AM on April 23, 2013 [2 favorites]
peters shop is gong to be my first new sockpuppet.
posted by cjorgensen at 10:05 AM on April 23, 2013
posted by cjorgensen at 10:05 AM on April 23, 2013
cjorgensen: peters shop
We need a "Thrift Shop" parody called "Peter's Shop" toot sweet.
♫ Why-why, why, why, why can't I post? ♪
&c.
posted by Rock Steady at 10:49 AM on April 23, 2013
We need a "Thrift Shop" parody called "Peter's Shop" toot sweet.
♫ Why-why, why, why, why can't I post? ♪
&c.
posted by Rock Steady at 10:49 AM on April 23, 2013
In the movie adaptation of this thread Jamon Iberico will be played by Jean Rochefort.
posted by Ad hominem at 1:59 PM on April 23, 2013
posted by Ad hominem at 1:59 PM on April 23, 2013
>> More to the point, it's a "power law curve", like so many other things on the web.
> Is it really, though?
Well, if favorites beget favorites, it very well could be....
Incidentally, has anyone looked at whether there's an inverse relationship between number of comments vs. number of [post] favorites for posts on the Blue? I've noticed this with my very small and highly biased sample of posts, and I can certainly imagine reasons why this would be the case, but I'm curious how universal it is.
posted by Westringia F. at 3:13 PM on April 23, 2013 [1 favorite]
> Is it really, though?
Well, if favorites beget favorites, it very well could be....
Incidentally, has anyone looked at whether there's an inverse relationship between number of comments vs. number of [post] favorites for posts on the Blue? I've noticed this with my very small and highly biased sample of posts, and I can certainly imagine reasons why this would be the case, but I'm curious how universal it is.
posted by Westringia F. at 3:13 PM on April 23, 2013 [1 favorite]
The closest I can think of in memory to a stunt like that was the Jamon Iberico people, who signed up like 12 accounts about different kinds of meat all in one morning after getting banned the first time, and that was needless to say not a successfully sly approach.
cortex, you know my fond feelings for MeFi, but turning down jamon is inadvisable. If it was jabugo, you are dead to me.
posted by ersatz at 6:14 PM on April 23, 2013
cortex, you know my fond feelings for MeFi, but turning down jamon is inadvisable. If it was jabugo, you are dead to me.
posted by ersatz at 6:14 PM on April 23, 2013
Incidentally, has anyone looked at whether there's an inverse relationship between number of comments vs. number of [post] favorites for posts on the Blue? I've noticed this with my very small and highly biased sample of posts, and I can certainly imagine reasons why this would be the case, but I'm curious how universal it is.
I've noticed something similar, too, and I would not at all be surprised if there were a demonstrable relationship. My highly unscientific theory is that active members generally like demonstrating in some way that they have interacted with a post, even if they don't have a specific comment to make, either to acknowledge the poster's effort or just to have it in their track record for future reference. Posts that generate a lot of conversation might have comparatively fewer favourites because everyone who interacted with the post commented already. But for posts that are about subjects about which there isn't a whole lot to say or that take a long time to digest (you see this a lot with mega-posts introducing a rich source of content), favourites are a way of performing that interaction with a thread in the absence of any other way of giving the poster a metaphorical nod of acknowledgement.
posted by Phire at 9:32 PM on April 23, 2013
I've noticed something similar, too, and I would not at all be surprised if there were a demonstrable relationship. My highly unscientific theory is that active members generally like demonstrating in some way that they have interacted with a post, even if they don't have a specific comment to make, either to acknowledge the poster's effort or just to have it in their track record for future reference. Posts that generate a lot of conversation might have comparatively fewer favourites because everyone who interacted with the post commented already. But for posts that are about subjects about which there isn't a whole lot to say or that take a long time to digest (you see this a lot with mega-posts introducing a rich source of content), favourites are a way of performing that interaction with a thread in the absence of any other way of giving the poster a metaphorical nod of acknowledgement.
posted by Phire at 9:32 PM on April 23, 2013
Also, for those of us who use favourites as bookmarks, a post about a long article or a video on the blue will often get a favourite from me, but either I won't actually get around to coming back and reading/watching it, or once I do, I have nothing to say about it anyway, so it won't get commented on. I'm more likely to comment on things that I can read quickly, because then the conversation about it in the comments is still on-going by the time I'm finished.
posted by lollusc at 1:05 AM on April 24, 2013
posted by lollusc at 1:05 AM on April 24, 2013
Ohhh, we need an h-index for favorites! Mine is 6.
posted by ecsh at 5:48 PM on April 24, 2013 [1 favorite]
posted by ecsh at 5:48 PM on April 24, 2013 [1 favorite]
Hirsch has a Metafilter account?
posted by Tell Me No Lies at 6:20 PM on April 24, 2013
posted by Tell Me No Lies at 6:20 PM on April 24, 2013
Whups, thought we were talking Bacon numbers.
I think there's probably a pretty picture to be made out of favorite densities.
posted by Tell Me No Lies at 6:22 PM on April 24, 2013
I think there's probably a pretty picture to be made out of favorite densities.
posted by Tell Me No Lies at 6:22 PM on April 24, 2013
Well, if favorites beget favorites, it very well could be....
Ha, good point. What I'd really like to do is look at the mean and variance of the favorite distribution. At a first glance it looks pretty wide for a Poisson distribution, which suggests there probably is some degree of "contagion," but we'd really need to know the proportion of comments with zero favorites because that would change the area under the curve hugely, I think...
posted by en forme de poire at 9:10 PM on April 24, 2013
Ha, good point. What I'd really like to do is look at the mean and variance of the favorite distribution. At a first glance it looks pretty wide for a Poisson distribution, which suggests there probably is some degree of "contagion," but we'd really need to know the proportion of comments with zero favorites because that would change the area under the curve hugely, I think...
posted by en forme de poire at 9:10 PM on April 24, 2013
A couple more thoughts:
There might be enough data to see whether the favorite distribution is captured better by a single distribution or a mixture, which would tell you whether there's a "cutoff" as in the question above the fold (nested models, yo).
Greg Nog has a favorites h-index of 77.
posted by en forme de poire at 9:16 PM on April 24, 2013 [1 favorite]
There might be enough data to see whether the favorite distribution is captured better by a single distribution or a mixture, which would tell you whether there's a "cutoff" as in the question above the fold (nested models, yo).
Greg Nog has a favorites h-index of 77.
posted by en forme de poire at 9:16 PM on April 24, 2013 [1 favorite]
Mine's 64.
posted by empath at 9:56 PM on April 24, 2013 [1 favorite]
posted by empath at 9:56 PM on April 24, 2013 [1 favorite]
Mine's 22. When do we get tenure?
posted by lollusc at 11:37 PM on April 24, 2013 [2 favorites]
posted by lollusc at 11:37 PM on April 24, 2013 [2 favorites]
How about calculating favourites vs tags? What does MeFi like best?
posted by five fresh fish at 8:19 AM on April 25, 2013
posted by five fresh fish at 8:19 AM on April 25, 2013
What percentage of the favorites do the top 1% have? The top 10%?
Serious question.
This is an excellent question and I couldn't find the answer posted anywhere, so we had to pull the infodump.
The uncomfortable truth is:
The top 1% of favorite holders control 38% of the favorites.
The top 5% of favorite holders control 70% of the favorites.
The bottom 90% of all users with at least 1 favorite control only a measly 17% of the favorites.
Here's a handy pie chart showing the distribution.
OCCUPY METAFILTER!
posted by pravit at 9:16 PM on April 26, 2013 [9 favorites]
Serious question.
This is an excellent question and I couldn't find the answer posted anywhere, so we had to pull the infodump.
The uncomfortable truth is:
The top 1% of favorite holders control 38% of the favorites.
The top 5% of favorite holders control 70% of the favorites.
The bottom 90% of all users with at least 1 favorite control only a measly 17% of the favorites.
Here's a handy pie chart showing the distribution.
OCCUPY METAFILTER!
posted by pravit at 9:16 PM on April 26, 2013 [9 favorites]
What's the break point of number of favorites to be in the top one percent? I need to know if I am going to get occupied or not.
posted by Rumple at 10:03 PM on April 26, 2013 [2 favorites]
posted by Rumple at 10:03 PM on April 26, 2013 [2 favorites]
If you don't know, you ain't there.
posted by Etrigan at 10:05 PM on April 26, 2013 [1 favorite]
posted by Etrigan at 10:05 PM on April 26, 2013 [1 favorite]
The distribution of existing favorites is only half the picture (although a pretty pie chart it is; Thanks pravit!)
We also need to know the ongoing favorite accumulation rates. Who's hogging the fresh favorites, depriving the rest of us of our well earned due?
posted by Tell Me No Lies at 10:39 AM on April 27, 2013
We also need to know the ongoing favorite accumulation rates. Who's hogging the fresh favorites, depriving the rest of us of our well earned due?
posted by Tell Me No Lies at 10:39 AM on April 27, 2013
For the period beginning April 20th 2012 and ending April 20th 2013:
For all users who began the period owning at least 1 favorite:
15,222 users received at least 1 new favorite.
3,071,818 total favorites were earned.
The top 1% of favorite earners (earned at least 3,393 favorites) received 34% of all new favorites.
The top 5% of favorite earners (earned at least 887 favorites) received 66% of all new favorites.
The bottom 90% of favorite earners (earned no more than 375 favorites) received only 20% of all new favorites.
Pie Chart of Favorites Income Distribution
Other interesting tidbits:
The #1 earner was elizardbits, who earned 30,432 favorites and jumped to the #10 spot in total favorite count from #49.
The #2 earner was griphus who earned 29,860 favorites and moved up from #9 to #3 in total favorite count.
The #3 earner, The Whelk, who earned 27,876 favorites, remained in the #1 spot in total favorite count.
5 out of the 10 highest favorite earners reported NYC as their location.
5 out of the 10 highest favorite earners were in the top 10 for total favorite count at the beginning of the period.
The Gini coefficient of Metafilter favorite income over the period is 0.87 (0=perfect equality, everyone earns the same, 1=perfect inequality, one person earned everything).
The average favorite income for the period was 202 favorites, and the median income was 15.
(I'll try doing some analysis linked to user activity, since the low median likely indicates a lot of people who received a few favorites on old posts but were inactive over the period)
posted by pravit at 11:35 PM on April 27, 2013 [8 favorites]
For all users who began the period owning at least 1 favorite:
15,222 users received at least 1 new favorite.
3,071,818 total favorites were earned.
The top 1% of favorite earners (earned at least 3,393 favorites) received 34% of all new favorites.
The top 5% of favorite earners (earned at least 887 favorites) received 66% of all new favorites.
The bottom 90% of favorite earners (earned no more than 375 favorites) received only 20% of all new favorites.
Pie Chart of Favorites Income Distribution
Other interesting tidbits:
The #1 earner was elizardbits, who earned 30,432 favorites and jumped to the #10 spot in total favorite count from #49.
The #2 earner was griphus who earned 29,860 favorites and moved up from #9 to #3 in total favorite count.
The #3 earner, The Whelk, who earned 27,876 favorites, remained in the #1 spot in total favorite count.
5 out of the 10 highest favorite earners reported NYC as their location.
5 out of the 10 highest favorite earners were in the top 10 for total favorite count at the beginning of the period.
The Gini coefficient of Metafilter favorite income over the period is 0.87 (0=perfect equality, everyone earns the same, 1=perfect inequality, one person earned everything).
The average favorite income for the period was 202 favorites, and the median income was 15.
(I'll try doing some analysis linked to user activity, since the low median likely indicates a lot of people who received a few favorites on old posts but were inactive over the period)
posted by pravit at 11:35 PM on April 27, 2013 [8 favorites]
also - the above is for all types of favorite - blue/green/gray/etc. posts and comments combined, and does not include anonymous.
posted by pravit at 11:43 PM on April 27, 2013
posted by pravit at 11:43 PM on April 27, 2013
5 out of the 10 highest favorite earners reported NYC as their location.
Since they are not here, I'll pretend I'm one of them and simply say thank you to all the little people outside of New York for favoriting New York made comments.
posted by Ad hominem at 11:50 PM on April 27, 2013
Since they are not here, I'll pretend I'm one of them and simply say thank you to all the little people outside of New York for favoriting New York made comments.
posted by Ad hominem at 11:50 PM on April 27, 2013
In the next batch of analysis on favorites, I tried to link user activity to favorites. I define a "contribution" as any post or comment on MeFi, AskMe, or MeTa. I used the period 4/27/2012 - 4/27/2013 and excluded users who made no contributions over the period, and users who did not exist before the beginning of the period.
13,561 users made at least one contribution, of which 11,951 (88%) earned at least one favorite.
How much did they contribute? Here we look at the proportion of new content that the top 1%, 5%, and 10% accounted for vs. the proportion of favorites they took.
*The top 1% of favorite earners contributed 23% of all new content and took 32% of all new favorites.
*The bottom 90% of favorite earners contributed 31% of all new content and took 21% of all new favorites.
How many favorites per contribution do the top 1% make? The top 10? How do they compare to the 99%?
Here is a scatter plot of each user's favorites per contribution ratio against total contributions, with user names highlighted for interesting points. Only users who made at least 100 contributions (2,164 users) are plotted.
Factoids:
*The average favorites per contribution ratio for all users with at least 100 contributions was 2.63. 27% of all users had a higher ratio, 40% of all users with at least 100 contributions had a higher ratio, and 88% of 1%ers had a higher ratio.
*The top 10 favorite earners contributed far more than most other users on an individual basis.
*The average ratio for 1%ers was 4.45. The average ratio for the top 10 was 4.8.
*The 1%er with the highest ratio was flex who made 478 contributions and earned 7,976 favorites for a ratio of 16.7. This user also had a higher ratio than any other user who made at least 100 contributions.
*The 1%er with the lowest ratio was homunculus who made 5,925 contributions and earned 5,899 favorites for a ratio of ~1. 92% of all users with at least 100 contributions had a higher ratio.
posted by pravit at 4:10 PM on April 28, 2013 [3 favorites]
13,561 users made at least one contribution, of which 11,951 (88%) earned at least one favorite.
How much did they contribute? Here we look at the proportion of new content that the top 1%, 5%, and 10% accounted for vs. the proportion of favorites they took.
*The top 1% of favorite earners contributed 23% of all new content and took 32% of all new favorites.
*The bottom 90% of favorite earners contributed 31% of all new content and took 21% of all new favorites.
How many favorites per contribution do the top 1% make? The top 10? How do they compare to the 99%?
Here is a scatter plot of each user's favorites per contribution ratio against total contributions, with user names highlighted for interesting points. Only users who made at least 100 contributions (2,164 users) are plotted.
Factoids:
*The average favorites per contribution ratio for all users with at least 100 contributions was 2.63. 27% of all users had a higher ratio, 40% of all users with at least 100 contributions had a higher ratio, and 88% of 1%ers had a higher ratio.
*The top 10 favorite earners contributed far more than most other users on an individual basis.
*The average ratio for 1%ers was 4.45. The average ratio for the top 10 was 4.8.
*The 1%er with the highest ratio was flex who made 478 contributions and earned 7,976 favorites for a ratio of 16.7. This user also had a higher ratio than any other user who made at least 100 contributions.
*The 1%er with the lowest ratio was homunculus who made 5,925 contributions and earned 5,899 favorites for a ratio of ~1. 92% of all users with at least 100 contributions had a higher ratio.
posted by pravit at 4:10 PM on April 28, 2013 [3 favorites]
You are not logged in, either login or create an account to post comments
- having a comment/post get 12 favorites, at which point it will show up as a line item in the contact activity sidebar for people who have the commenter/poster as a contact (and the more people have that person as a contact, the greater that effect likely would be)
- something with a lot of favorites may make an appearance on the Popular Favorites page, so folks who review page that would be more likely to see and favorite an item; the threshold for inclusion there is relative to the overall favoriting distribution in that time period, so there's no hard number for this
- something that has had a lot of favorites may for related reasons have a lot of fantastic flags or be brought to a mod's attention otherwise, which could lead to a bestof/sidebar post which typically leads to a lot more favorites as well
Breaking this stuff down to a hard number is tricky because of the multiple possible factors, but a lot of the data is in the Infodump so folks have been able to at least experiment previously.
2. Hard to say since it depends on having access to that info from the owners of linked sites; we don't track clickthroughs to external sites via link redirects or anything like that, so we don't have that data ourselves. I can report anecdotally that I've seen on the order of a few thousand hits on day one, loaded heavily toward hour one, when something of mine has been posted and I've been able to track analytics. But it'll depend as well I'd think on whether it's a sole link in a post or one of several, and on whether for whatever reason it's a more or less eye-catching or brain-catching post or link.
posted by cortex (staff) at 2:08 PM on April 22, 2013 [2 favorites]