Immediately before I sat down to write this blog, I was attending a lecture on Social Influence Analysis and Social Prediction by Dr Jie Tang, Tsinghua University, Beijing.  He is an associate professor of computer science.  Naturally, I was fascinated by his insights into the dynamics of social networks, such as how an individual user can influence others’ behaviours.

His study shows that about 90% of the human emotions are predictable, through social network mining, text mining, statistical learning, and semantic web searching.  How scary is that?

Such research is right on the cusp of the field of human communication, where science and art collide.

What I didn’t anticipate was the need to recall my own early science degree to follow Jie Tang’s explanation of his factor graph model for creating mathematical formulae that could learn by themselves.  But it is that very connection between maths and psychology that enables researchers to quantify social influence and the correlation between influence and social actions.

So I sat there fascinated by the creation of algorithms, which we so blithely talk about, but don’t always fully appreciate.  The equations on the screen looked like the chemistry of a nuclear bomb, and offered a glimpse of their potential to provide a similarly massive explosion in our analysis of mobile social networks to predict human emotion.

While AUT University is part of this marvellous open dialogue in the field of creative technology, social media are still being restricted in China itself.  Google has been displaced by the Chinese search engine "Baidu" which seeks to emulate Bing or Google inside China’s physical borders.  Firstly, the Chinese government set out to control Google: Now they have a search site which they can control.

Also in China, Youtube has effectively been replaced by Youku, which has 1,500 license holders, including television and film companies in that country regularly uploading content. 

Youku's video library includes many full length, popular films and TV episodes from the West, which can be viewed outside China only on Youku, since copyright restrictions prevent Youtube and others from broadcasting them.  Ironically, while thumbing its nose at western copyright law, Youku has its own strict rules on pornography and for content that the Chinese government finds politically objectionable. 

Furthermore, Youku has implemented a digital fingerprinting technology which is says is intended to eliminate copyright infringing content on the site.  It also sounds like a way of keeping a close eye on users.

Meanwhile, in the West, computer scientists at the University of Washington are reversing the “eye of God” on the internet.  They have created a way to put expiration dates on ea-mail, chat messages and Facebook postings, according to Futurist, Nov-Dec 2009: p.2.

Their software, called “Vanish”, encrypts the messages and spreads the data among different computers on the file-sharing network.  As people use the network, the portions of encryption are spread around to untraceable users making the original message indecipherable.

One of the benefits of Vanish will be for the artists among us who may express our emotions impulsively only to discover that they remain as an eternal record of our immoderation.  We should thank our scientific peers for this offer of a future where we can expunge our digital trails.  Of course, art in its other forms can and should remain preserved for ever.

Indeed, art provides the postscript to this blog, as Wikipedia tells us the delightful origin of the name “Baidu” is from an 800-year-old Chinese poem which says: “Having searched for him hundreds and thousands of times in the crowd, suddenly turning back by chance, I find him there in the dimmest candlelight”…. beautiful.

 

 
 
 This Blog recently reported that Google has bought Metaweb for its open-source database, Freebase, with 12 million web “entities”, from science to celebrities. 

Now it is time to question where this will lead, or more specifically, where the semantic web will allow us to lead each other.

The thing about Freebase is that it tags items so that computers can understand what they are about, and then relate them to each other by meaning.  Freebase enables the whole web to behave a bit like GetGlue http://getglue.com/ which allows users to rate things or discover new popular choices by finding out what their friends enjoy.

One shudders to think what semantic tags will do to the existing  imbalance between popularity and reliability of information on the internet.  Research already shows that we tend to follow each other like lemmings racing towards a cliff.

Adding to concern about the influence on popularity on credibility was an article in New Scientist (24 July, 2010) which demonstrated again what is sometimes called the “band-wagon effect” or even the ghost of “group-think”. 

The article asserts that sociologists have been searching for laws governing human interactions and social networks for decades, but now the world’s 500-million users of social media are providing researchers with a living laboratory. 

Duncan Watts of Yahoo Research and Matthew Salganik, now at Princeton University used a web-based experiment to demonstrate that when it comes to music preference, we behave like sheep (or lemmings).

To examine what made some songs more popular than others, they created a project they called Music Lab.  It used a website where more than 14,000 people could rate 48 relatively unknown songs.  They compared the opinions of experts with those of amateurs and measured downloads to assess a tunes popularity. 

They also cloaked the information some of the respondents received    to show that the number of people choosing a song influenced others into selecting the same one.  That is, human influence had a major effect on making some songs more successful. 

“So, like it or not, it seems that many of us follow the herd,” (New Scientist, 24 July 2010: p. 32).

This doesn’t only apply to something as familiar as popular music.  Even choices that require knowledge and expertise can be driven by mob-behaviour. 

For instance, by using Facebook researchers, Jukka-Pekka Onnela and Felix Reed-Tochas of the University of Oxford, have been able to show that when a new application becomes available on the Web, users adopt it independently of their friends’ opinions.  However, when the popularity of an app crosses a certain threshold, its popularity draws in other people and its adoption can expand suddenly and massively.

Suddenly the cliff ahead of the Lemmings can be a lot higher.

 
 

Google Wave is on the way and, from the previews, looks set to change social networks in the same way that Twitter and other microblogging platforms changed blog posting.
For the full introduction, take a look at the preview given at Google O/I - lengthy at an hour and twenty minutes, but you'll soon get the idea.
Also makes me wonder what is in store for webinar providers like WebEx and others and Wave has the potential to change their business model as radically as mainstream news media were changed by the online environment. A watching brief.

Posted by: Catherine Arrow