Take a look at this study in the latest First Monday, on Twitter Under the Microscope. What it does is associate each Twitter user with three types of people: “followers” (people who “follow” the person), “followees” (people followed by the person, the declared friends) and “friends” (people who have received at least two @ messages from the person, the “hidden” friends).
Huberman et al come to a finding that’s not surprising: the driver of usage is a sparse and hidden network of connections underlying the “declared” set of friends and followers.
This by itself is not surprising: as the authors point out, every community, every social network, evinces a similar pattern. We send e-mail regularly to a very small portion of our address book; we call a very small portion of our mobile contacts; we reach out to a very small portion of our Facebook “friends”. This sort of behaviour is true even in other communities; for example, there are a number of opensource projects that behave similarly.
So why should Twitter be any different?
Let’s take a look at this diagram:
This would suggest that as the number of friends increases, there is apparently no loss in reciprocity. Yet, when you look at this diagram, there is a suggestion that the number of friends is constrained in Dunbar-like manner:
I’ve tended to believe that if anything, social software would help raise the Dunbar number. The studies above suggest this is not the case. But I’m still holding on to my hunch.
Why? Because I think we live in an age where there something wonderful happening, something that just has to affect the Dunbar number, something that is accentuated by social software.
Most people would agree that the development of language as a means of communication affected the Dunbar number, raised the Dunbar number.
Most people would agree that the evolution of language from oral to written cultures had a significant and positive effect on the number.
It is not difficult to make a case that there was further improvement when writing turned to printing (with an intermediate growth phase as scripts becames codices).
It is reasonable to suggest that when we got the world’s biggest copy machine (as Kevin Kelly called the internet) we would see another shift.
I think there is one more shift of significance. The ability to search and retrieve communications cheaply and quickly. Something that has just started happening.
JP – Interesting report. I think that there may be something that the report is not accounting for the time it takes for the mass market to learn how to effectively use new tools.
Like many things I look at my own experience for guidance in doing a gut check on this type of thing and I agree with you. These numbers do not reflect my experience with Twitter. I did a quick review my my @ messages and DMs over the last couple of weeks and it is over 70+ people out of a network of around 1400. I feel like I’ve been able to maintain more relationships with more people than I ever could before. Are they different types of relationships? Absolutely but they also provide much more value then the huge Rolodex that sites, mostly unused in Outlook.
Maybe I’m an outlier but I think it’s just a matter of learning how to effectively use the new tools.
JP,
This is the kind of post that makes one think, analyze and wonder.
First off, the analysis between # of followers, followees (for lack of a better word) and ‘friends’ is excellent. However, I think the definition of a ‘friend’ could be revised upwards.
Regardless of how one defines a ‘friend’ on twitter, the trend remains true that as Internet makes terribly easy conversation/communication across boundaries, one’s friend-number would rise. It is indeed interesting that you think that a Dunbar limit (~150) becomes apparent.
I think it is because *social norms* evolve much more slowly than *technological tools*. Twitter and such tools are largely the domain of the digirati today; these are the early adopters. The mass public still hasn’t adopted these. The norms for usage, communication with social internet tools, etc. will continue to evolve slowly.
It perhaps won’t be until a few years (5? 10?) before we will see any significant increase (10%? more?) in the Dunbar limit.
My 2 naya paisa :-)
Great post.
-a
JP, your unch is safe, fear not!
One needs to be careful of mean averages in studying social networks. Figure 4 is slightly misleading if taken at face value, in that it hides the bifurcation of twitter user types. There are many following 1,000’s followed by very few (‘spammers’ for want of a better word), and many followed by 1,000’s who follow 1000′. That is obscured by the average.
Figure 6 (the first graph) should roll off if the second graph were to represent the median, but it doesn’t, so there is an anomaly in the study there. Figure 6 is the one which really tells you if the ‘Dunbar Number’ is growing or not. It is still thought provoking though!
With regards to Dunbar’s number more generally, it is worth remembering that it isn’t a hard limit (Dunbar drew his stats from averages, not maximums) – see more in my post here: Dunbar’s number = groups language and social media. It seems, rather than moving the number, social media might actually be shifting what we mean by ‘friend’. I’m not sure if that is a good thing, but increased communication generally is, so hopefully it will all work out!
Despite all the action that’s happening on the internet, I am not really sure if we’d be bypassing the Dunbar limit any soon. Because time at hand will always be a constraint. It also depends on your definition of friend. Can you count someone whom you actively conversed(@’d) a while back but now don’t as a friend?
I’d like to take Dunbar’s number(~150) as the number of active parallel connections you have have over a given(and small) period of time. So while by the definition of 2 @ for a friend definition I might have a couple of hundreds or even more friends, over a given period of time I would be actively interacting with about 150 people only. Without realizing I might be talking to newer people and loosing touch with some older one’s(not all of them obviously), who also similarly would be making connections with new people, forgetting old one’s like me.
Good remark. I also think social software should raise the Dunbar number, but will also hurt the quality of relationships.