“You Stab His Back, I’ll Stab Yours”
Jonathan Haidt has an interesting bit on the value of gossip in his book The Happiness Hypothesis (the quote in the title of this section comes from Haidt’s book). He draws on research by Robin Dunbar, who demonstrated that “within a given group of vertebrate species…the logarithm of brain size is almost perfectly proportional to the logarithm of the social group size… Human beings ought to live in groups of around 150 people, judging from the logarithm of our brain size; and sure enough, studies of hunter-gatherer groups, military units, and city dwellers’ address books suggest that 100 to 150 is the natural group size within which people can know just about everyone directly…”
“Language allows small groups of people to bond quickly… Dunbar notes that people do in fact use language primarily to talk about other people — to find out who is doing what to whom, who is coupling with whom, who is fighting with whom… In short, Dunbar proposes that language evolved because it enabled gossip… We are motivated to pass on information to our friends.. And when you do pass on a piece of juicy gossip your friends reciprocity reflex kicks in and she feels a slight pressure to return the favor… Gossip elicits gossip, and it enables us to keep track of everyone’s reputation without having to witness their good and bad deeds personally. Gossip creates a non-zero-sum game because it costs us nothing to give each other information.”
Today’s social networks provide binary controls to represent relationships — I am either a “friend” of Matt’s or I am not a friend. If I am a friend, I can see public comments on Matt’s “wall.” If I’m not a friend, I cannot. Facebook has recently increased the granularity of permission controls for users, but requiring consumers to ACTIVELY manage complex controls on a case by case basis will not work. Well over 90% of users never touch the default setting in a product. Also note that while 90% of gossip is negative, at least 90% of public comments on social networks are positive. The social cost of being caught saying bad things about people in your network is high. Yet, negative gossip is critical for ultra-sociality and helps maintain community norms (gossip is a form of punishment for violating those norms).
So how can we PASSIVELY manage more granular permissions? How can we encourage more negative gossip (to reflect what happens in the real world)? And how can we limit group size to 150 so that people can have a really powerful online community?
Idea: Coffee Klatsch — A Place for Gossip
1. Strong user authentication.
A secret of Facebook’s early success was the fact that users went through domain-based authentication until the Fall of 2006. To join the Harvard network, you need an email address @harvard.edu. Until 2006, it was no .edu domain, no service.
In addition to keeping non-.edu domain authenticated users out, there were other benefits to having a third party (the University) involved in authentication. If a user acted badly, Facebook could ban the user. Unlike other services where any email will do, Facebook required a .edu address. So getting kicked off FB was really getting kicked off FB. Whereas, with other services getting kicked off means spending 10 minutes signing up for a new Yahoo! account. There may have even been an implicit fear that a really bad action could result in discipline or expulsion.
In Korea, all web services require that users register with the equivalent of a social security number. My hunch is that this required authentication is a key reason behind Cyworld’s incredible success. In this case, the third party involved in authentication is the South Korean government!
Mark Zuckerberg’s “real people, real names” ethos behind Facebook’s product design is right on. What I haven’t heard him or anyone else talk about is the authentication process required to deliver on that promise. Perhaps FB will eventually introduce some notion of authentication quality of service in order to go back and clean up accounts registered since 2006 which lack some strong form of authentication? For example, Facebook could badge profiles for which there is no third party authentication and then set the service’s defaults to treat those users differently from those who have strong authentication. It’s not a problem yet, but as the service grows I predict that it will become a real issue.
A simple option for authentication is SMS account confirmation. Facebook is doing this for activating mobile accounts. In the early days of Gmail, SMS authentication was required — this is likely a big reason Gmail still has less SPAM than the other web-based services (iIt’s really too bad that they didn’t keep it that way).
Why not require SMS authentication for everyone in our new service? I don’t buy the argument that getting in the way of easy registration will kill your service’s viral potential if you are providing value in return for the extra work. Real people and no SPAM is worth the effort.
2. Artificial friend scarcity + the web-of-trust.
Limit the total number of friends a user can have to 150. In the early days of Friendster, there was a 500 friend limit. Many people argue that this limitation was a key reason for MySpace’s “win” over Friendster. But Facebook came along after MySpace and made it HARDER to get in (at least for awhile). Make people focus their attention on the key people that matter to them by setting a limit on the number of friends they can have. Remember the rule of 150.
Since it’s hard to say “no” to others friend requests, make the “friending” process asymmetrical. Simply enter the email addresses of your top friends (no more than 150) and you’re ready to go. All of my friends do the same. We would then build a “web-of-trust” reputation system that would dynamically determine the reputation of someone within your network based on a trust inheritance scheme (think of asymmetric friending as “links” to people and do for people what Google did for URLs).
3. Gossipy — a gossip distribution system.
You have completed steps #1 and #2 simply so that you could build a secure and personalized distribution system to, you guessed it, distribute gossip about others (good and bad).
The product is a single page site, with two modules.
Module 1 is a commenting system. There are two components — an AJAX friend search which allows you to select the person from your 150 friends that you want to gossip about and a text box in which to gossip. When you gossip, your name will not be attributed to the comment (but the target’s will be, of course). However, the only people who will see your gossip are those who have both you and the target in their web-of-trust. The goal here is to mimic what happens in the real world.
Module 2 is a “news feed” like the one on Facebook or FriendFeed. You simply get a scrolling list of the gossip in your network.
The goal here is not to create another profile service. Build this as a application in other social networks and as a standalone service, similar to what the FriendFeed guys have done. That’s it. A simple system to distribute and receive gossip about you social network.
I have a follow-up post on Dunbar’s number here.
13 Comments
May 24, 2008 at 1:46 pm
Yes, in the real world people gossip (both good and bad) all of the time. But in the offline world that gossip usually: (1) occurs behind closed doors (figuratively and even sometimes literally), (2) is shared among just 1 or 2 people at each exchange (eventually getting to many people, but only over time), and (3) is directly attributable to the person spreading the gossip at each exchange. Will facilitating the exchange of gossip in the online world in a manner counter to these three tenets still lead to similar results that we see in the offline world? It’s an interesting hypothesis to test. My guess is that it would undoubtedly spread virally and probably much faster than in the offline world given the pace of sharing is accelerated. But the anonymous nature you suggest seemingly would make the gossip much more negative than positive in tone. After all, if I have something positive to say about someone I’m more than happy to have those thoughts attributed directly to me (as the Facebook Wall enables today). As such, my guess is that the service would be much more popular among teens than adults (see Gossip Girl and the anonymous spreading of negative gossip via sms that drives much of their behavior). It’s also something that teens are already doing on MySpace by signing up with virtual and anonymous profiles, which in some cases has lead to very tragic results (see the MySpace teen suicide story from St. Louis). Let’s hope that anyone tackling this idea can figure out a good way to avoid those types of things from happening.
May 24, 2008 at 2:56 pm
Great feedback Sean.
According to research quoted by Jonathan Haidt, most gossip in the offline world is negative. So I’m not sure that the percentage online would be more negative. However, after sleeping on this idea I am with you. There is a real risk that this does more harm than good.
With an idea a day, I’m going to have some good ones and some bad ones. This is probably a bad one…
May 24, 2008 at 6:01 pm
fortunately, bad ideas also require execution
I don’t really think it’s ‘bad’ tho… just might need some iteration. one sad MySpace incident — even if directly causal, which I question — doesnt mean the idea couldnt be useful to many others.
but because neg feedback is so powerful & the web is so permanent & searchable, the “ink” used should prob be disappearing or temporary. perhaps this is why IM is used instead of web pages or email… less permanent.
May 25, 2008 at 4:56 am
Can the ‘gossip’ be channeled towards something productive, like our democratic process? Can I ‘gossip’ about how I feel about the latest piece of legislation or court ruling, in such a way that those who made the decisions can hear about it?
Even today, it’s decidedly inefficient to give my elected representative a piece of my mind. Yes, it may be “negative” feedback, but it wouldn’t hurt the hired hands in Sacramento, D.C, etc. to hear it — without going through the lossy polling process or taking to the streets en masse.
I am happy to authenticate a) that I am your constituent and b) that I voted for/against you last time. Now, pay attention to my ‘gossip’ about your vote on that last piece of legislation because I’m sharing it with my 150 closest friends!
May 25, 2008 at 5:03 am
Love where you are taking this Jolly.
Do steps #1 and #2 from above, but instead of building a gossip engine as #3, build a Town Hall. You could do everything from the most local of topics to Presidential elections. You could cover topics that are Global (global warming, AIDs, Middle East Peace) and encourage debate on just about everything.
May 26, 2008 at 4:59 am
Hey,
Interesting piece -
—
We would then build a “web-of-trust” reputation system that would dynamically determine the reputation of someone within your network based on a trust inheritance scheme (think of asymmetric friending as “links” to people and do for people what Google did for URLs).
—
Check out SezWho…I think we have the reputation engine you are talking about here
-Jitendra
May 26, 2008 at 5:13 am
Jitendra –excellent. Will check it out. Thanks, Mike
May 28, 2008 at 11:52 pm
“when you do pass on a piece of juicy gossip your friends reciprocity reflex kicks in and she feels a slight pressure to return the favor”
Exactly! Why gossip if my comments are anonymous? I will not get the bonus!
Also why participate in an application that might generate bad comments for me and damage my image? I would block it!
And don’t forget the support load for the developers of this application. How many messages asking them to remove a bad comment or reveal the writer will they receive each day?
There is already a lot of gossip taking place in social networks through MESSAGES. This trivial application is closer to the real world than any sophisticated “gossip application” that might be built:
- Gossip takes place between 2 or three people…
- Those who give information take credit for it…
- Nobody might think she can block private messages between other people or ask the developers to block them
MESSAGING is already there and is the perfect gossip system! That’s why it is the most popular feature in social networks
May 29, 2008 at 2:40 am
Very well said, Stefanos. I agree with you. I think the first two “steps” could be leveraged for other social products — but I’m over the third. It was a bad idea. C’est la vie.
May 29, 2008 at 6:58 am
Hi Mike,
How will SMS account confirmation prevent users from creating multiple accounts?
Vlad.
May 29, 2008 at 9:31 am
Hi Mike, interesting thoughts.
1. I agree SMS authentication helps service providers but I’m not sure users understand those benefits upfront.
I helped implement call-back authentication for PayPal’s password recovery (”answer this automated call at the phone # you gave us before”). Though it was relatively quick, a lot of users felt it was unnecessary despite the real fraud-prevention benefits it gave them.
2. I like the idea of letting people communicate taboo/gossip online. Socialmoth is a sort of anonymous social network, letting people express their more real or insecure sides.
DailyStrength is a social network for discussing health conditions and support, solving important problems using sensitive info.
Guy Kawasaki launched Truemors, a prediction market around rumors. It was somewhat panned but I think it’s a good idea, just needs better execution. Perhaps targeting to areas that people care about; e.g. rumors about pop culture don’t interest me, but tech rumors like Valleywag reports on (sadly) do.
3. I very much agree with Dave’s comment that people usually want negative comments to be anonymous or temporary or both. Even if in webs of trust, relationships sometimes go sour and people may not want sensitive info divulged if bridges become burned.
4. I’m a big fan of game mechanics so I love the idea of limited currency: forcing people to choose among many options for something that matters to them, whether it’s items in a game, positions to apply for, or candidates for a date.
I understand why start-ups want to offer the world to everyone, but limited currency forces us to engage and make better decisions.
May 30, 2008 at 5:01 am
[...] Many Friends is Too Many?”, asks RWW and points to interesting research by Robin Dunbar: http://is.gd/o08 [...]
June 4, 2008 at 7:55 pm
I think the concept of friend scarcity would be a brilliant way of making the current social networks more usable and trustable. It also provides an implicit “gossip” on the trustworthiness of individuals by how they rank.
Currently there is no marginal cost to adding a friend and in general most people accept contacts or friends, even people they barely know.
As a result, there are lots of people I have listed as contacts or friends that I don’t really endorse professionally or personally (have no information on them or even worse a negative take on). It creates a very misleading view on people who are good at spamming to get contacts/friends but actually might not be as well liked, networked, or recommended as they appear.
Friend scarcity would very quickly reshape the social graphs and let webs of trust vote on who they really endorse professionally and personally, but without overt negative “gossip”.
Leave a Reply