Upon waking to a snowy Sunday morning and a peek out the window which got me to thinking about a book which I had just read, “How many Friends Does a Person Need” written by Robin Dunbar an anthropologist. One of the main concepts Dunbar puts forward in the book is the “Dunbar Number” (assume since he coined it, he got to name it) which says we as humans can only manage relationships or connections with about 150 people on average.
Past this number and in short, names turn into numbers if you will and the ability to apply the “theory of mind” begins to break down. In analogy this can be likened to our short term memory where we can only handle 5 to 7 digits at a time before our brains start to jumble the numbers around. While still watching the falling snow outside, my own mind started to wander through a thought experiment as how one could test this. Then it struck me, why not use my LinkedIn data?
While my wife was still snuggled warmly away, it was down stairs for a coffee I went to collect and then to the keyboard to figure this all out. However before taking that first sip of coffee, let’s start with why LinkedIn, first I don’t Facebook and have several posts on why so it’s not worth rehashing here. Although thinking about the data set even if Facebook was a data option, it seems to be too socially manipulated if you will. Whereas LinkedIn really has two core sets of folks, those being the 500+ folks who are “accumulators” who seek connections and typically initiate the request. Then there are the less than 500 folks who are the “accumulaties” which are the request acceptors. Basically it amounts to your standard active passive social model at work.
As it would be a post in itself to postulate the dynamics of these primary groups within the LinkedIn community so too far afield for the purpose of this discussion. However the one other side bar worth noting is LinkedIn stops counting at 500 connections at least for what it posts publicly so the domain for the data set was perfectly set (1 to 500) as the ones I
was interested in where the less than 500 members. So here is how it broke down, I have 817 connections (yes a member of the 500+ club). However there is a rule for my connections as I must know them and they must know me. In short requests come in every day from folks because they work for the same company or we are members of a group in common. These won’t be accepted as they don’t meet the criteria above of knowing each other which matches with Dunbar’s criteria.
So after another sip of coffee, it was off to open a excel spreadsheet and a web browser logging into LinkedIn, now to get the data. The prospect of opening 817 pages was not appealing as the wife was sure to stir long before this task was complete. In the old days some VB script or Java code would have done the trick out of hand quickly, however I’m just not that fast at coding any more. But what I did notice (believe it was cup of coffee number two at this point) is the contact screen lists the numbers by the name, so it was type and scroll for about an hour to scrape the data. (if any one knows a better way, please post it as I would like to repeat this in about 6 months)
The data set surprised me a little as out of 817 connections, 48 where “accumulators” at 500+ or in other words 6% of my connections where outliers. This large of a number surprised me as my thinking was it was going to be only around 3%. So with the data in Excel it was off to the races with the analysis portion of the experiment and the first was to run the “mean” function to provide a simple average of 147 (it was 168 with the outliers) while the “median” was 120 (it was 128 with the outliers) all well within Dunbar’s suggested range set. Also not to start folk’s eyes glazing over, so will keep it short however also ran the kurtosis as well as several other tests on the data set validating the perceptions of the results which all checked out.
Why is this so important you might be asking, well the simple answer is this number is the limit of social networking as in concept we could be linked to the world and therefore the world to us. However with this limit, it means this might not be the case as the “social fabric” won’t be able to support the depth of connections needed. In addition, it also can mean there will be breaks in the social chain which cannot be solved by the law of large numbers as if “I” don’t Facebook, then there will be 150 unique connections that won’t be made, which means this flaw will cascade throughout the chain (aka social fabric).
Well there it is we validated a modern day social discovery using a social networking system, two cups of coffee and your help…