Opinion formation in network models of societies

Some years ago I have dabbed for a while in simple simulations of social phenomena. My rather rusty programming skills and some elementary physical mathematics has allowed to produce some models. I got interested in the network models, especially the preferential attachment networks (for details hunt down works of Barabasi, Dorogovtsev, Mendez or Newman, see also bibliography of the In the Country of the Blindfolded).

Since this time I have received a few papers on close subjects to review, which I have done to the best of my ability. I no longer have the time to continue my own studies (they were done during a forced sabbatical between two jobs caused by non-competitiveness clauses in the contract). But looking through the literature on the subject, which seems to grow exponentially (the subject is relatively easy, it is also quite easy to obtain funding - from either physics or soccial sciences departments, and it seems to be one of the fashionable ones), I do see a less trodden path.

Perhaps someone would like to take it?

The idea is based on the reversal of the preferential attachment (`the rich get richer') phenomenon that seems to be found in many networks (from the physical Internet infrastructure, throuth WEB, to scientific collaborations and actor's relationships). In processes that govern the formation of opinions in modern societies, such networks, with influential hubs play an important role. While many early models were based on 1D or 2D close proximity opinion exchanges, in today's small world we interact via networks.

But one phenomenon that is quite crucial for the models dealing with information spread and opinion formation in such networks is that people quite readily cut the links with those who are `not to their liking'. Thus, instead of getting converted to majority opinions, the enclaves of like minded people are formed, cut off from the main network. We all see it in real life, especially in dramatic circumstances of terrorist gropups and their supporters.

Computer modelling would help in establishing the conditions that would diminish this tendency to cut off the links with those who do not share our opinions. Some incentives for keeping the links open (for example through promoting participation in `open' activities and societies, where ideas are exchanged) might be included in the models and studied.

One of my friends has remarked that such computer models in themselves are worthless, as they are purely artificial. Without reference to real life data (experimental or observational) the models are just toys. I agree. But the phenomenon of cutting the links is easy enough to study in reality. One can imagine, for example, monitoring the evolution of links between students on a campus of a University, as they go throught the period of study and later disperse to their jobs. Smaller groups could be used to monitor susceptibility of connections to differences in opinions. So, it does not have to be a purely artificial topic.

Question: does anyone know of works already done along these lines?

Network of genes, network of patents

I was away on a business trip for a few days, and in a hotel lobby I have chanced upon an International Herald Tribune (July 4th, 2007). And there I found a longish article in the economy section (!): Biotech industry rocked as theories change. The change of theories is described as:

"challenging the traditional view of our genetic blueprint as a tidy collection of independent genes, pointing instead to a complex network in which genes, along with regulatory elements and other types of DNA sequences that do not code for proteins, interact in overlapping ways not yet fully understood."

While this may be interesting scientific result, the IHT article aptly points that the one gene-one effect (one protein) stance led the way to gene patents: patents based on the assumption that industrial gene (made by splicing techniques) would have defined, owned, tracked and uniform effect, moreover with ability to sell and retract. But when the genes really work in a complex network how can we be sure which effects is due to what? Biologically. And legally too: if one patented gene depends on other natural or patented ones? Especially in ways we do not understand? How to pay royalties? How to split an eventual responsibility for damages? How to ensure that effects are uniform and as promised?

Well, there is much more to the subject. I was never in favour of granting patents on genes, but the US law has allowed the companies to do it. Without, as it turns out, proper understanding of the separability of the "inventions". And because we lack real understanding of the most of the genome (the original paper that has prompted the IHT article speaks about 1% of the genome) we are blundering blindly -- and this is never a good thing.

For more information see:
The ENCODE Project Consortium, Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project, Nature 447
Mark B. Gerstein at al. What is a gene, post-ENCODE? History and updated definition, Genome Res. 2007 17: 669-681
Thomas R. Gingeras Origin of phenotypes: Genes and transcripts Genome Res. 2007 17: 682-690