Homeostasis in Social Networks

4 August, 2008

As a network grows and develops a power law distribution of connectivity, the resulting structure is heavily biased towards the initial nodes by virtue of their prior existence. Social networks therefore tend to be disproportionately dominated by the individuals who have been around the longest. This state is very stable, since information flow on the network is highly dependent on these hubs. However, the imbalance of attention may result in potentially valuable new members being neglected or moving elsewhere. Stagnation as a consequence of homeostasis might be one reason why one network gives way to another.

In living organisms a stable structure is desirable, and homeostatic mechanisms are present to maintain equilibrium. However, if an organism cannot break out of a given equilibrium state it may prove brittle and vulnerable to external pressures. The birth and death of individuals allows a tribe or species to adapt to a changing external environment. An alternative response to environmental changes is exhibited by cellular slime molds which, when food is scarce, merge into a “multicellular slug-like coordinated creature which crawls to an open lit place and grows into a fruiting body. Some of the amoebae become spores to begin the next generation, but some… sacrifice themselves to become a dead stalk, lifting the spores up into the air.”

Corporations and other large organisations also suffer from the effects of homeostasis. Although there is an entire industry devoted to the study of organisational structural dynamics and change management, stagnation is more often than not alleviated by market pressures, whether by acquisition or enforced “restructuring”. In our work lives we are each happy to accept a comfortable equilibrium state, but this reduces the ability of the organisation to adapt. And of course, when nation states are too rigid and authoritarian they tend to fall to revolution rather than evolution.

In the brain homeostasis might correspond to boredom resulting from a lack of stimulation. This reaction is perhaps intended to instigate a search for new ideas or experiences, which are generally rewarded by a feeling of pleasure. If something is new and exciting it’s usually fun too, because we enjoy learning. The desire for novelty provides a mechanism to move the mind out of an unhelpful state.

If a garden is left to nature, a power law distribution of species quickly develops. One or two particularly well-suited or vigorous plants take over whilst others dwindle. Gardeners address this by weeding and pruning. Even “wild gardens” require the careful application of a little encouragement and discouragement. When a new plant appears it must be nurtured whilst the weeds are kept in check.

If social sites like Digg, Wikipedia and Twitter are to remain dynamic and continually evolving they need to solve the homeostasis problem. In the social media, new and interesting contexts or individuals with novel viewpoints should somehow be amplified. There has been recent discussion on Twitter about how a modified Retweet could be useful, and perhaps this partly fulfills the need since the resulting amplification is relative to the connectivity of the sender, but it ultimately depends on the goodwill of community members. I have mentioned in previous posts how novelty can be identified with semantic profiling, but how can it be “subtly encouraged” without threatening the social ecosystem in question?


Novel Opinion Filtering and Social Attractors

23 July, 2008

Opinion Filtering
In limited information space decisions are difficult without emotions for guidance, whether we call it instinct or values. We also find it easier to trust decisions made on an emotional basis because there is (we hope) a clear understanding of a sustained agenda, the echo of the future. We find it hard to trust machines to make any more than the most basic decisions because we do not share empathy with them. The act of seeking recommendations or advice from a person is shaped by our expectation of their values. Even though we may be able to follow publication bots providing automated content filtering, it is more difficult to extend this to opinion filtering, which is what we rely on people for.
Novelty Index
In this context, a novel article is by definition introducing a new opinion, different to those expected from the assumed bias. Perhaps it is therefore a huge assumption that novelty would be sought after within a document set. Perhaps people more commonly prefer to avoid “novel” opinions?

Blog Anxiety
Reluctance to blog stems from the perception that you have to be a professional to publish and the associated assumption that any publication will reach a significant audience. The blogosphere’s power law distribution of audiences bridges the gap between broadcast and clique. The intermediate region allows for greater interaction within groups of bloggers or Communities of Practice, and provides the greatest potential for group expansion and action. Weak ties provide more bridging capital between ideas.
Social Attractors
A power law also describes the size distribution of meme cascades, or the extent to the influence of particular ideas within the social network. As we become more connected we might expect to see an increasing number of monumental changes as cultural instability increases. Higher connectivity and lower transaction costs also make the economy more unstable and more susceptible to manipulation. As in any complex dynamic system, a large enough disturbance in group perception can in theory result in a paradigm-shifting cultural change, whether “good” or “bad”, into the next social attractor. Many of these are generally perceived as beneficial, like the events which lead to the fall of the Berlin wall, but the achievements of mob-rule are sometimes only desirable for the mob, and as history shows a mob may be misguided by a determined few.
News as Global Counseling
An undesirable global cultural state attractor might conceivably be compared to mental illness. Although a surprisingly large number of individuals suffer at some point during their lives, most recover and many are never affected, I would suggest in large part due to positive interactions with others. What does counseling do for people and what would be the equivalent in the global network? Perhaps we are able to move a brain from one attractor to another by exposing prevailing cyclic thought processes to alternative viewpoints. Perhaps encouraging people to be informed by a broad spectrum of the opinion-base is also a healthy thing to do, for a more stable society. Isn’t this what multicultural diversity is about after all? The liberal media would have us believe this is what they are for. So instead of a “read this similar article” you might receive a “read this similar article from a different point of view”. Although one can think of some potentially detrimental examples, giving everyone absolute control over their own blinkers might not be a good idea either. Proximity does not preclude segregation…

Web Science and the Novelty Engine

17 July, 2008

Web Science
There is a call from Sir Tim Berners-Lee and Wendy Hall for an academic discipline which studies the internet.

Although this may focus on the internet and its success stories by incorporating lessons learnt in computer and social sciences, I would hope it also draws heavily on work done in the past few decades in the natural sciences, given that pretty much every interesting system studied these days produces knowledge on complex networks. Web Science could be a real destination for generalists!

Social Semantic Web
Glad to see people thinking about the crossover between social networking and the semantic web. It will be interesting to see the spread of papers at the Stanford University symposium.

Novelty Engine
According to Nicholas Carr internet technology is bombarding us with so much information we are all rapidly losing our attention span. This not only forces us to speed-read everything but it prevents us from contemplating the deeper issues. One reason social network tools like Digg, Twitter etc help is because we trust others to read and filter on our behalf. However there is still so much to sift through for the knowledge nuggets.

What if we were also able to trust technology to do the reading and filtering? We are already seeing micro-blogging bots at the broadcast end. Social web browsers would also benefit from some intelligence, enabling messages to be organised and filtered on receipt. Again, perhaps the focus would be on identifying posts which are “related but novel”. Or at least routing them into personalised semantic buckets. Are there tools already out there, eg how much of this can you do with Flock? What would it take for us to build trust in personalised blog-reading bots?