<< A quarter petaflop   |   ARCHIVES   |   Exercise 'Common Ground' - coordinated by the UK >>

Thoughts about the 'Earth Simulator', and where it may lead us

After writing the previous entry I looked more closely at the Earth Simulator, and find myself intrigued. Large scale simulation opens up whole new possibilities, rather as the new science of optics did 500 years ago: partly for our knowledge of the physical world, but also understanding of our own behaviour and psychology.

As an example fro the Earth Simulator journal, a simulation study of the 2003 heatwave in Europe looks at the large scale climatic events in the Atlantic that preceded the heatwave. The paper says:
"...a long-range forecast model must have the atmosphere up to the top of the stratosphere, all oceans, the land surface, and perhaps the ice, interacting dynamically and thermodynamically with each other. Needless to say, the model must be able to accurately represent those second- order variables, such as the cloudiness, precipitation, and soil moisture, that are important for low-frequency forcing. Finally, but never the least, observational network must be improved to provide a reasonable initial condition to the forecast model. In the light of long evolution time scales of planetary-scale waves, having an accurate initial condition that extends up to the upper stratosphere is critical to a reasonably accurate simulation of the low-frequency evolution of the atmosphere. The next generation of super computers will be probably large and fast enough to handle the task..."

They should also allow us to study data over longer time-series than we have done so far, to see really long-term changes in action. 'New' data sources, such as fossilised tree rings, allow us to recreate weather observations that nobody actually recorded, even thousands of years ago, so there is more data than might at first seem.

If we ever manage to flesh out the Gaia hypothesis, it may be due to the availability of large-scale simulation and modelling techniques.

At the opposite end of the scale, simulations of genes and and proteins may enable radical new designs of the building blocks of cells.

Just as microscopes enabled us to see small things, and telescopes very large distant things, perhaps we are looking at another real extension of our perceptions and understanding, made possible by petaflop speed and vast data storage capabilities.

We're also developing the ability to amass large amounts of sociological data too. At first this is sinister: governments and marketeers developing the ability to track us. For instance, every time I use my credit card or take my mobile phone anywhere, I generate data which is stored somewhere. RFIDs may enable retailers to track my purchases and even my movements. Face recognition (or number plate recognition) software can already allow closed circuit TV cameras to track me around the streets of London. Most of this data will be simply thrown away, unless the Police want to use tiny chunks of it for some reason - to trace a person in a particular place at a particular time.

But if all of this data could be collected, stored, and then used to generate a simulation model, what new truths might emerge from the aggregate?

For instance, the work of University of California at San Diego sociologist David P. Phillips, Ph.D, suggests that
- accidents and suicides can be linked to media reports. (The "Werther effect")
- cardiac mortality increases on psychologically stressful occasions such as "unlucky days." (The "Baskerville effect")
- legalised ganbling leads to higher suicide levels.
- a 25% increase in deaths attributed to prescription drug errors on the 1st day of any month, when US pharmacists have a spike in prescriptions to prepare.

Economist Stephen Levitt has studied similar relationships between data sets that aren't normally collected (or at least correlated) to find evidence of:
- match rigging in Sumo wrestling
- homes owned by real estate agents selling for higher prices than homes owned by their clients
- incentive systems for teachers influence the extent to which they cheat in administering tests to their pupils.

Drs Phillips and Levitt are both extremely perceptive thinkers with unusual abilities to suspect the existence of new relationships and to know where to look for evidence.

But what might become visible if we had a 'telescope' that would make these relationships stand out clearly, just as Galileo detected Jupiter's moons because their unexpected motion was suddenly there for him to observe?

Would we close casinos? Have a more balanced and less sensational media? Be better able to anticipate social and individual vulnerabilities? Be more honest?

Group therapists have spent some time working on the theory of 'large groups': but nobody has so far been really able to study the largest group of all. As the evidence becomes available, and the unimaginably powerful tools to store and crunch the numbers, is this about to change?

500 years ago, people thought the sun went round the earth. The evidence was there all the time; only our ability to perceive has changed.


ADD YOUR COMMENT:



Notify me when someone replies to this post?



COMMENTS SO FAR:

Powered by pMachine