Sunday, 11 September 2016

Facebook's greatest mistakes!

#Facebook, #futuretrends, #Zuckerberg, #Pardus, #Horton, #Dunbar, #socialmedia, #Messenger, #DeclineofFacebook, #socialmediamarketing, #smm, #socialnetwork, #socialnetworking


The question of the change in peoples habits on Facebook is well known. People share less of their original content and share more of others. The platform is transforming from its social networking structure to something new.

Its evolving.

We want to take a different slant on this and ask the “Why?” question. And we don’t want to trivialise this. 

We want to ask WHY it has transformed and HOW it has transformed and whether in attempting to answer these questions we might see where its going?

Let’s begin by asking what we know and how Facebook have responded to it.

View the video blog here.

The history
It maybe that you point to ‘MySpace’ or, perhaps, ‘Friends Reunited’, now defunct, and say social media sites are a bit like civilisations, they rise, they fall, it is the natural order of things. Firstly this trivialises real historic trends and secondly whilst you might be broadly correct, the specific case and the nature of what we might consider “a fall” has to be accounted for.

An interesting and controversial place to start is the pre-print by Cannavella and Spelcher They did a neat little test. They looked at the relative popularity of the term ‘Facebook’ for searches on Google over time. This grew, peaked and has started to decay away in 2014. They noted exactly the same form of behaviour for the term ‘MySpace’. This broad distorted bell shaped peak for MySpace they argued represented its relative popularity and its social presence.

Over an 8 year period MySpace went from zero to boom and then to bust. They suggested that social connections on the web were not unlike catching a contagious illness being passed from person to person until it grows uncontrollably until eventually the contagion’s effectiveness falls away again. Using the mathematics of contagion transmission they found a good fit to MySpace data. And they suggested that if they applied this to Facebook data then activity on Facebook was doomed!

This idea was heavily criticised and rightly so, but their basic concept that some form of decline is coming for Facebook was not misguided.

Here you can see the search data compared for FB and Myspace and there is indeed some similarity in form. If we normalise the peak height of data from MySpace, Facebook and also for Pinterest, Deviant Art and Tumblr and shift the data so they overlay we see some similarity in form.

We have diagrammatically an initial rise in popularity, a social adoption of the platform followed by a decline, the width of which is a measure of the social cohesion of the community. 

There is significant variation in the relative cohesiveness of the people on those platforms with Facebook being relatively broad and so a slow decay. However we still see a decline in search interest as suggested by the the original authors. Love or loathe their idea, the data is still showing a continuing decline. Tumblr and Pinterest are still on the rise but looking at the typical lifetimes of social platforms that would be expected as they are younger.

As an aside and just to convince you that not all Google search data takes this bell curve form in this timescale, here’s a different search pairing: Lingerie and Chocolate. This analysis can show useful trends and answer many of life’s deeply held questions!

You can see the complex harmonic regularity of searches, each peak in the data corresponding to Christmas and Valentines day, when Chocolate and Lingerie are clearly popular. But you note it shows something else, the searches for chocolate are far stronger than searches for lingerie proving once and for all that chocolate is more popular than sex.

Back to the main story. Whether you like search data as evidence or not, it bears out Facebook’s own data which confirms social activity generating original content is in decline on the platform. “The Information” reported a 21% decline in personal shares on Facebook, and personal shares is something that is actively being encouraged by Mark Zuckerberg and the team.

Other features are collapsing, referrals for example are in decline. As we reported in a previous blog on social dissonance, it means people share content they don’t actually view. So whilst people might share publishers content such as Buzzfeed or Huffington Post, people actually viewing content on these platforms has declined by many 10’s of percent. (Its not universal Vice for example is doing OK).

One reason for this, it is argued, is the growth of “chat space”, but this is more of an observation than an explanation for Facebook’s decline.

Facebook responded to these quick exchange platforms like Twitter (a sort of social chat space), WhatsApp, Snapchat, Instagram, kik, the list goes on and on,
by launching Messenger.

They are keen on users transferring to Messenger. And make no mistake Facebook is enforcing this change on the users. As from early June 2016 all users were forced to carry out all conversations through Messenger, a platform capable of supporting more than just text messaging. 

But are they going to go down the same route again and could this cause Messenger to decline too after its early adoption. Will it loose its social cohesion? Stay tuned and find out.

Dunbar’s number(s) are old friends when it comes to analysing community structures. And whilst this concept might vary in detail and in specific values of numbers in social groupings the ideas are themselves are reasonably well accepted.  Dunbar argued that social groupings develop each with a particular characteristic set of interactions, from very close friends to distant connections. Providing drivers exist these groups will maintain their number of members. So you will have on average a certain number of close friends, and on average a certain number of mates whom you know quite well, then beyond that another circle of acquaintances and so on.

Now this work has anthropological support from the study of the neocortex of primates and is potentially an evolutionary manifestation in the broadest sense of that term. What is also very important is what holds these groups together, so a hunter-gatherer community may group differently to agrarian societies. Another thing that is important in Dunbar's ideas is the presence of the drivers to hold the groups together.

Numerical classification of social groupings has been observed before by other commentators. But less well known is that Dunbar in January 2016 specifically applied his ideas to Facebook. So now we are starting to stray into new territory.

His findings were exciting because we are able to see explicit evidence for Dunbar’s numerical hierarchies of social association in Facebook. And sure enough the data confirms his ideas giving an inner circle of close friends numbering about 4 on average and a further 14 good friends as the next circle out.

Beyond that the patterns become more hazy. Dunbar suggests patterns go up in multiples of 3 so he would expert social connections with similar patterns at 50, 150, 450 and about 1500. And his findings for Facebook confirm this.

So we have some idea of how we operate on Facebook, the types of links we have and their close similarity to those we might expect to see in real world social groupings. 

Now take this analysis a step further. And this is where things start to become interesting. Fuchs et al analysed an on-line browser based space game called Pardus. ( It is a multiplayer game where you are involved in a virtual community of space pilots where trading, pirating and smuggling occur and you can build friendships, form associations and alliances for players mutual benefits. Some of the social collectives such as large alliances (political factions ~ 2000) are dictated by the games structure, others you are free to develop as you wish. The game has been very widely studied and shown to be a virtual society with many of the properties of a real society.

In one exercise undertaken by the team they analysed the structure of the society in terms of a Horton - Strahler number. This method of analysis considers the connectivity as being represented somewhat like a river delta that branches with more and more complexity as the delta gets larger and larger. The Horton number, denoted as h, represents the highest level of branching in the flow pattern.

Applying this to the hierarchical structures of society an interesting pattern emerges. At a Horton number of h=1 we are dealing with just the individual. As we get to h=2 we get to close friends, at h=3 we get friends, by h=4 alliances, groups we are closely allied too and so on. When h=7 we have encompassed all the game players.

It becomes even more interesting if we start using their numerical analysis in light of Dunbar’s work. They suggest that societies are organised in a fractal fashion as,

G(h) ~ p^h

where p=4.42 and G(h) represents the size of the group with a Horton number of h. Now its worth noting this is a multi-fractal structure with each level of the structure determining the relationship to the group size.

Putting in all the constants we get,

ln(G(h) = C + h ln(p)

Now here is an interesting experiment. Suppose we set our group size to be the Dunbar’s hierarchy of social structures and back calculate the Horton number. So for example,

Looking at this sequence the Horton number goes up by 1 level of branching complexity with each growth in the group. However at high levels of population this begins to fail. So turning this idea on its head can we determine an equivalent Dunbar number for simple changes in the Horton number.

What this allows us to do is to visualise the nature and complexity of these networks in simple form. 

Firstly Facebook has limited the organic reach of posts, so we may assume our “natural” network structure is disrupted.

Secondly Facebook has allowed, indeed encouraged agents outside our network to intrude upon it, groups that it thinks we might be interested in through marketing and advertising posts. But this is broadcasting content rather personal content. And worse still we may share and post some of this content written by third parties so that our network is now getting this information directly and possibly indirectly devaluing our posts. This is broadcasting rather than us talking to our network.

Thirdly, and this is an even bigger disruptive factor. As is often quoted, “we are all publishers now”. And in the case of our on-line image we are trying to manage our image to a wide group. And yet we know measures of social dissonance are telling us not only is the content broadcasting into our network not appropriate as something to reinforce our network, it is also arising from false shares. We are sharing things to promote our social image but not content we are truly interested in. So even our own broadcasts are false. And we might by extension feel they don’t match the true interests of others in that network. In short the shares are a sham.

The network is now a noisy mess. We’ve gone from a dinner party with friends to a club where the music is not only loud but nobody wants to dance to it any more.

And we can imagine that we want to be truer to ourselves with our closest friends and post appropriate material, things we know represent us and not some socially desirable version. The social groupings Dunbar proposed from anthropological research require a sound and compelling reason to remain connected and for all the above reasons we suggest this network is being irretrievably disrupted. Irretrievable because the social behaviour is inherent in us. Our network has grown beyond its purpose and we fail to control and discipline our shares within it and we lead to a presence which is “not us” we are partly alienated from our own presence, its a ghost presence for high orders of h but one not appropriate to our h= 2, 3 or 4 level friends.


Facebook still has a large captive audience. It is not dead. It is becoming more suited to be a targeted news aggregator and a source for advertisers and marketers to garner your interest. And of course retaining the Messenger connection will help too to keep Facebookers on the platform. 

In terms of Dunbar’s groupings we could speculate that Facebook is heading towards a natural social divide: That is our closest two groupings of friends are reserved for the Messenger space and the others reside in Facebook. That is probably too simple in terms of the division of the public and private self and more likely this is a new stratification of social structures.

So what now for Chats? These form a disciplined structure for us and limit the range and extent of communication. Its a form of social governance. And if what is speculated here is right then there are dangers in Facebook intruding into this private space of users. But equally Facebook may think to themselves that because the social cohesion is so high in these close groups we can indeed allow external intrusions without damaging the social connections. So marketing through Chatbots and more direct content may be successful. As Messenger has high levels of immediacy it is critical to Facebook that they do not disrupt its functional convenience. Disrupt this and you disrupt the network cohesion and the social grouping collapses.