Before giving an answer to the question of whether GNU social is decentralized or distributed, it would be interesting to give some definitions, because it is important to answer this question and understand its consequences. The distinction between network topologies is an old Indiano tool to understand the major social changes of the last decades.
This use of the distinction between network topologies helps us to understand how information flows through a network from one node to another, which nodes of the network are capable of retransmitting information to other nodes, whether there are nodes whose the survival the network depends on, and whether some of the nodes have the ability to filter and control the information that the others receive. In summary, the debate on network topologies addresses the autonomy of nodes and structures of power. Not coincidentally, one of the most famous slogans of the cyberpunk movement reminds us that
Under every information architecture there hides a power structure.
The search for and distinction between different network topologies played an important role in the birth of the Internet. In 1962, a nuclear confrontation seemed to be an imminent threat. So, Paul Baran received an important order. The Rand Corporation asked him to define a structure to use to set up communication systems that could survive a first strike of a nuclear attack. The main result of Baran’s work can be seen in the image shown above to the right.
Robustness and network topologies
In 1966, Paul Baran, in his famous report on Darpanet, presented three different network topologies and their characteristics. The main difference between the three network topologies is how robust they are during a nuclear attack or, in other words, to what extent can they tolerate disturbances without suffering a total collapse. We could have a long, drawn-out discussion on this topic and give a wide-ranging presentation on the measurement of the robustness of a network but, in summary, let’s just say that the more robust a network is, the fewer nodes are disconnected by extracting any given node.
This, plus a look at the image above, allow us easily figure out that the first two topologies, which is to say, centralized and decentralized networks, are highly dependent on the centralizing nodes — the centralized network at the global level, and the decentralized network at the local level. In the centralized network, the loss of the main node would result in the collapse of the whole network, and as a consequence, the surviving nodes would not be able to continue communicating between each other because of the lack of the node that interconnects them. In contrast, in distributed networks — the third topology that appears in the first image of this post — each node is independent and the fall of any node would not disconnect any another.
The social nature of distributed networks
The originality of the Indianos was to use network topologies to explain the major features of social evolution since the eighteenth century as a function of the dominant media in each era (the post, the telegraph, the Internet). In the book The Power of Networks, we can read a broad historical tour through the last centuries and easily understand how technological advances gave life to new information structures which, in turn, created social changes. The key to the historical tour that we can read in The Power of Networks is in seeing people and connections between people where Baran saw computers and cables.
But, if through Baran’s view of a network topologies, we can technically measure the robustness of networks, what emerges from the view that David proposed to us a decade ago now in The Power of Networks?
This view quickly makes it clear that in distributed networks, the non-existence of central nodes not only makes it possible to have a network that is much more robust, but hierarchies also disappear, autonomy is favored and the control over others becomes impossible.
As a result, the nature of distributed networks is completely different from that of decentralized ones. A distributed network is not a more decentralized network. This is why it’s very important to answer the question of whether GNU social has a distributed or decentralized structure.
What is GNU social and what is its structure?
On the net, there are several descriptions of GNU social. Most of them present it as an alternative to Twitter or, more generally, as a microblogging service. Certainly, the current functions and options that GNU social offers are mostly characteristic of microblogging services. But in practice, what we find is thatconversations quickly flourish once again, and that more and more new functions appear that reduce the validity of these descriptions.
What is GNU social?
This conversation and especially the message belowput us on the track of a broader and more appropriate answer.
All microblogging and social networking sites are using selectively flawed ideas and should be transformed. Nobody needs ‘microblogging,’ they want socialization.
The desire to socialize and connect with each other shows the fact that all these systems and sites are not social networks in themselves, but tools that, like instant messaging and mail services, are used by social networks, which is to say, networks of people.
So we see that GNU social is a free tool for interconnection and communication used by different social networks. What functions will it offer, and what we will exchange through GNU social? That depends on what the social networks that use it want.
GNU social also has a particular characteristic that interests us especially, and it has to do with its structure. So, we return to the question, What is GNU social’s structure?
Is GNU social decentralized or distributed?
We’ve already presented widely on this, because it is important to answer this question. What will help us distinguish clearly between the three basic network typologies is the interdependence of the nodes that are part of the networks. Interdependence tells us whether the individual nodes depend on others to be able to communicate with others, and therefore, defines how robust they are under attack.
The nodes in a network driven by GNU social are the different installations like (lamatriz.org, loadaverage.org, quitter.se, etc.). A quick look at the image next to this paragraph us clearly shows that the nodes of GNU social do not depend on each other to communicate, and that the fall of one of them does not endanger the survival of the network at all. As a consequence, GNU social has a distributed structure.
Conclusions
From all this, we can draw two important conclusions. First, we realize that it is not necessary to look for a strict definition for GNU social, because what can be done with it will depend on what its users want. Secondly, GNU social has a distributed structure. This is an important distinction, because thanks to it, we see the birth of a social nature in which autonomy, privacy, and conversations are paramount.
Sorry about the broken links. Here are the URLS:
http://www.rand.org/pubs/research_memoranda/RM3420.html
http://www.rand.org/pubs/research_memoranda/RM3767.html
Steve (translator of the original article)
Just fixed them, Steve, thanks for the links.