Saturday, May 28, 2011
Twitter is closer to emulate a Neural Network than Facebook
When we think of Twitter and the innovation behind it, the first thing we all think is 140.
140 characters is without a doubt an amazing innovation that Twitter introduced which makes communication flow faster, forces twitters to summarize a piece of news or information, or an idea or an opinion and allows followers to get information or an idea faster as well.
Also, after the introduction of url shorteners (tinyurl,com originally and many others later on such as bit.ly) an emergent property of Twitter came to life: the linked web. Any blog post or news article out there could be potentially linked multiple times in Twitter with quick summaries and opinions.
Other emerging properties or elements in Twitter are:
. Tags, early on twitters started using tags as a way to group events or themes together and follow them separately.
. Curated content appeared more recently in an attempt to organize tweets in channels like mode.
There is one more original Twitter property we don't hear much about and I believe it's a key element to the Twitter success.
In the world of two way communications, Twitter is one of the first Social Networks that works asymmetrically versus working symmetrically like Facebook.
In Facebook, you are my friend at the same time I am your friend (with the exceptions of some of the user types). We could call that a Symmetric Social Network. Other previous successful examples of Symmetric Networks are Skype and Instant Messengers.
In Twitter, there's no implied/automatic relationship between those that I follow and those that follow me. We could call that an Asymmetric Social Network. Other previous successful examples of Asymmetric Networks are phones, cellphones and emails, the linked web.
When you look at a Neural Network, you'll find that it is also an Asymmetric Network. Neurons have their axons that allow them to connect to other neurons, at the same time, other neurons have the freedom to connect their own axons back to the neurons connecting to them or choose other not connected neurons to hook up with.
It seems like Twitter is imitating nature's success in implementing this type of social network architecture.
The most common neurons in the brain (called multipolar neurons) process their in-coming and out-coming connections separately. On a neuron, dendrites are receptors for in-coming connections and the axon branches out-coming connections.
On a simile between the Twitter and the Neural Network model we would see the following:
. The Twitter accounts that follow me would be like the Axon connections on a neuron.
. The Twitter accounts a person follows would be like the Dendrites connections on a neuron.
Thinking of the advantages of connecting Asymmetrically you could find a few:
. More flexibility and plasticity on the resulting graph.
. Easier to detect Hubs and greater influence of Hubs on the Network.
. More independence (freedom of choice regarding incoming communications) in the level of the individual cell or network node.
. Loose (light weight) connections.
Advantages of a Symmetric Network:
. Greater Trust in the Network Connections.
. More back and forth, one to one Communications.
. Tighter connections.
. More room for reciprocity to occur.
Different Networks do have different purposes. Like this article states, Facebook is more of an Identity Platform while Twitter is more of a communication platform.
As a conclusion, Bio-mimicry is big nowadays as a source of innovation and ideas for the technological world we're creating to learn and be inspired on the biological world leveraging its millions of years of evolution.
I'm sure there are more implications in the end graph that either Network type generate that would be interesting to research.
I wonder if we would see different levels of complexity on the resulting graphs.
I also wonder what other properties we could mimic or learn from the Neural Networks to apply into our Social Networks.