I'm coming to the summer school to prepare for my postdoc at SFI, studying Technology Evolution.

I'm currently finishing a PhD in Computer Experiments at UBC, researching and developing Fast Bayesian Inference to help quantify prediction uncertainty when working with complex computer codes (simulators).

I want to build computer models that can estimate the uncertainty in the predictions by constructing appropriate frequentist confidence sets or Bayesian credible sets.

A major motivation is predicting the future in general and technology forecasting in particular (since the limits of our technological capabilities impose important constraints in the space of possible futures).

My personal philosophy is best described as Extropian and I want Extropy NOW!

I optimize sleeping for brain function, nutrition for longevity, and reading for creativity (new).

My research interests include not only those tools that can help us foresee the future, but also potential life-extension technologies that can take us there. I'm also intrigued by current efforts to create a Technological Singularity, e.g. by Artificial General Intelligence.

Project Proposal

The Limits of prediction in modeling social systems project last year highlighted some of the difficulties of forecasting in complex ecosystems, economies, or societies. I'd like to revisit some of those themes, but unlike the authors who discussed what we can't do, I want to focus on what we can do. Although the fortunes of individual nations, or companies, or leading products, etc. may be hard to predict, we may be able to foresee what's coming on the global scale or over longer time spans.

For example, unless you own stock in Research In Motion or Apple, do you really care who is winning the smartphone war? The fact that there is a 3G device on the market in 2008 is more important than the exact market share of the BlackBerry or the iPhone. Can we time our future wireless capabilities? According to this claim, wireless bandwidth growth has been remarkably smooth for more than a hundred years, doubling every two and a half years. Click here for a collection of many other proposed exponential trends that appear surprisingly predictable.

Another exciting question to explore is how we can exploit parallels with biology for prediction purposes, e.g. see the short article by W. Brian Arthur discussing why technology might be "evolving at something like 10 million times biology's rate" or Hans Moravec's picture depicting "evolution of computer power/cost" and the "brain equivalent" of comparable species. John Smart also mentions "speedup factor" estimates between 1 and 30 million for "electrochemical vs. electronic communication speeds".