I would like to start an initiative to track and measure memes (replicating ideas) as they move around the world in real-time. I've spent about a year thinking about the technology necessary to do this. It requires a lot of data-mining power, but the algorithms are fairly simple. Essentially, we mine the Web for noun-phrases and then measure the space-time dynamics of those phrases as they move through various demographic, geographic, and topical spaces.
The Human Menome Project would compute the real-time "Menome" of humanity -- the set of most influential memes presently active around the world. In addition, it would compute the menomes of individual societies, demographic groups, communities, etc. This information could be of great benefit to social scientists, journalists, marketers, analysts, investors, search engines, etc.
Recently I noticed some interesting work in this area related to detecting "word bursts" -- sudden increases in the usage of a particular term. Daypop is actually doing that now. Another relevant example is the Zeitgeist feature on Google.
My approach to this problem is substantially more sophisticated than what I've seen so far in that it borrows techniques from physics and the stock market to compute the velocity, mass, acceleration and momentum of memes. It also integrates NLP techniques for measuring sentiment. Based on this we can then graph the trajectories of memes, and even make interesting predictions about their future development and correlations with external time-series data-sets like the stock market for example.
This technology could be an important tool for the emerging science of computational memetics; it could also help to provide a new window into the workings of communities, nations and markets. I think of this as a new window into the collective consciouness, enabling us to see the collective thoughts of large groups of people as they develop and interact in real-time.