Notes from the James Dewar lecture yesterday at the The Long Now Foundation seminar series.
The success rate of long term predictions have been sketchy. That shouldn't discourage us from learning from our past mistakes and fine tune our approach. James Dewar yesterday introduced such a new approach to Long Term Policy Analysis (LTPA). The premise is, that the future is uncertain. Therefore let's search for what we can do today that best keep us on track no matter what the future brings.
It is the search for robust strategies that minimize regret.
James Dewar introduced the four key elements of successful LTPA:
• Consider large ensembles (hundreds to millions) of scenarios.
• Seek robust, not optimal, strategies.
• Achieve robustness with adaptivity.
• Design analysis for interactive exploration of the multiplicity of plausible futures.
The solution is detailed in the book he coauthored: Shaping the Next 100 Years
(You also find the chapters as downloadable pdf files on that page.)
Shaping is a methodology, that tries to optimally combine the number crunching scenario evaluation strength of the computer with the more intuitive selection and fine tuning strength done by knowledgeable humans.
Asked about how the humans get selected, he answered that they should have different viewpoints. The great thing is, that all these viewpoints can be put into the system and thanks to modern processor power the scenarios are played out on all of them. At the end the one with the lowest level of regret should be implemented.
Stuart Brand asked whether there would have been a different outcome if the politicians had used this methodology to decide whether to go to war with Iraq or not. If I recall correctly, his responds was: "Probably". Which I thought was a bit odd.
Comments