Musiclab, trends, abstraction and the network effects

As a child I had always wondered if people purchased the songs because they were on the charts, or had the initial rush to purchase had put them on the chart. Again, as a child, I had resolved part of the issue in my head as "the same mass seems to recycle to what's put in front of them." - the latter something re-enforced in my adult life as a business consultant.

Though it is a sweeping generalisation, I couldn't help think of it again when I came across this article in the NY Times a couple of weeks ago from 2007 written by Duncan Watts (amongst other things the author of book Six Degrees: The Science of the Connected Age) - it's titled "Is Justin Timberlake a Product of Cumulative Advantage?".

I find it particularly interesting as I'm technically minded and have noticed many relevant technologies emerge during my lifetime. These can be said, in some ways, are so integrated into cultural practice that they are in fact implicated as a technology and culture in a wider techno-social milieu.

In this article Watt's makes a number of observations, about the the social network effects that influence the very outcome of technologies such as pop charts.

The "Web 2.0 revolution" on the web brought a "revolution" of user generated content. With it, came the now well known proliferation of quantifiable social trust ratings in the form of "liking" or "disliking", thumbing up/down, 5 starring or 1 starring etc. This has tapped into a natural communicative social act of articulating endorsement or dis-favour in peer groups. As it turns out, according to Watts, that it might not be all too dis-similar from my appraisal of pop charts - seemingly they're rather unpredictable and not entirely determined by actual content, so much as the dynamics of flows between nodal points.

What he talks about in the article is an experiment which essentially tested a number of groups to see how, knowing what others rated the content, would affect their own ratings.

They did this by opening a website for users to rate music. Unbeknown to the users they were all, once logged in, separated into groups and each group was shown a different version of the site. That is, each group was only shown the preferences of others in that group - essentially dividing the site into differing worlds. The main differences was that one group had no ratings, and served as a sort of control group - in their world they could not see what anyone else rated so when they were asked for what they thought the quality of a song was they answered in absence of other's opinions.

The rest of the participants were split further into worlds where they could see the other user's ratings - however these rating-visible-worlds were multiplied so as to test the social network effect. Thus they existed with the same content (i.e songs) but were in different voting pools - thus in one world a certain song might be rated as 2 star whereas in another it might be 5 star. The participants of each world would not see, or be aware of, other worlds.

The results? Well - each world's participants rated the songs very differently. It had seemingly little alignment to the content (songs) itself but it infers the network effect had a lot to do with the decision processes of the participants, thus the decision processes and actually what they perceived to like.

The full article can be found here -
http://www.nytimes.com/2007/04/15/magazine/15wwlnidealab.t.html?_r=2
If it asks you to login you need not register just use the site http://bugmenot.com

If you would like a description of the experiment please see this lecture - http://videolectures.net/cvss08_dodds_tecolst/ . You will need to skip to the 46th minute or so.

Related to the execution of this experiement there is also a lecture titled "Social Science 2.0" by Watts where he explains the view that this type of research would not be possible without the internet. This may provoke the question how would these interactions exist in the same way without the internet? though that is a different question.) - http://videolectures.net/eccs08_watts_ss2/


So, why is this interesting to me when it's seemingly largely revolved around musical preferences (or musical social preferences?) - simply, an environment who's society is increasingly abstracted by layers of technologies, conventions, systems and structures is more likely to have its outcomes skewed by the implementation of the technology onto the network effect. That is to say, not that the network effect is skewing the outcome, but the short comings of the technologies (i.e. of language, of the star things, of the leichart scale etc etc).

Within this environment, the implication of technology and society is seen as non-deterministic, with neither society or technology as the universal determinate force, but rather an implicatory techno-social dynamic made up of minor structural negotiations of influence and interaction.

As the level of technology is better able to replicate societal, or more correctly, human desires, the blurring of the lines between machinistic and humanistic actors as nodal points will likely effect the current flows between nodal points, in a sense, of those relatively minor structural negotiations to a point where they may have specific and major determinate forces of machinistic and human nodes (or abstracts of control of those nodes made up of machinistic and or humans, themselves subject to participation as nodes themselves).

One of the widely recognised by-products of the "Web 2.0 revolution" of user generated content is the ability to rate things. This makes use of the basic positive network externality of articulating trust relationships and endorsements and disapprovals.

The logical extreme is akin to The Matrix - where the nodal actors are entirely abstracted, to a point where the human entities only perceive within the scope of the abstraction itself.

I suspect humans were already socially networked like this, but today, and in future, the increased penetration in societal organisation is likely to make things more unpredictable. And the day we won't be able to determine the entity controlling a node will likely make this unpredictability even more interesting and likely more skewed. One could say, we are already there.