Friday, May 16, 2008

Tracking Memes in the Infosphere

Infosphere is a term used since the 1990s to speculate about the common evolution of the Internet, society and culture. It is a neologism composed of information and sphere. More about its origins here.
The difficulties with memetics are many - and it has been bogged in controversy since a long time. One of the main problems is how to isolate a "meme" ; What IS a meme, anyway?
Start here, but don't expect to find a definite answer - it's not there yet!
The whole discipline (if it can be called that!) is in a similar state to that of genetics in the 1950s. What was a gene? It took Watson and Crick to come up with the molecular structure - the double helix - of DNA, before genetics really took off.
The meme sounds very vague when defined as "a unit of cultural information". An abstract but precise mathematical notion is required - perhaps it can be found in information theory?
So, not even knowing precisely what a meme is, how are we supposed to track them, and build theories around them, and maybe even try and predict stuff with them?

The meme-tracking problem.....
Some possible routes:
Web publication volume and search trends

Hitwise tracks search data of all major search engines, including Google.
Google Trends also tells us the history of search volume on keywords, i.e. how many searches were executed on these keywords over time. This sounds like a good indicator of what's "hot". I am not entirely sure it is a truly accurate indicator of "meme" though. For example, Breaking News of any kind will cause a peak in News Coverage - But News is not Meme!

Published Reports by Professional Market Research Firms:
E.g The Harris Interactive Annual RQ™ study, conducted yearly since 1999, assesses the reputation of the 60 most visible companies in the United States, as perceived by the general public. Changes in reputation are what we want to learn. Perceptions of 'brand' by consumer are one part of it, of course - but again, this is not quite enough or good enough data.

WOM data
: 'Positive word of mouth' data can be sourced from places like Keller and Fay's Talk-Track, a research service that tracks consumer conversations via a weekly survey
sample of 700 consumers aged 13+. Online brand mentions data can be sourced from a service like Nielsen Buzzmetrics. It searches the net for mentions of specific words or phrases on discussion boards, blogs or other places where consumers communicate online.

Advertising Spending: Weekly advertising spending data for television and national magazines can be had from Nielsen's Monitor + database. Online advertising spending can be obtained from AdRelevance, owned by Nielsen Netratings.

Agencies like ComScore MediaMetrix track website visits through a representative panel of 2 million users - another valuable storehouse of data but not "ready made" for meme-tracking by any means.

One problem for any non-US study is the possible difficulty in getting location-specific non-US data.

Research design incorporated open-ended, discoveryoriented in-depth interviews are another option, with the obvious limitation of being impossibly difficult to scale up, or even trust.

The meme-tracking space is supposed to be HOT round about now...
It’s not easy to define this space.....but as Alex Barnett says, these are not meme-trackers - there seems to be no real meme-tracker around. I agree.
"Memeorandum, Megit and Chuquet are not 'meme' trackers. They are news trackers. Or tittle-tattle trackers. Or gossip trackers. Again, generally speaking, there are no 'memes' being tracked at these sites". I especialy like his comment that "The idea that these are 'memetrackers' is actually quite a good example of a meme."
Which brings us back to the question: How does one define a meme, at least in a way for a bot can measure it? (I think if you can define something that a machine can understand then you have done a good job at the definition!)

Some interesting papers, using innovative means to find and interpret data can be found in the Journal of Advertising Research, December 2007. I am going to fish around for those again.

No comments: