[ExI] energy modelling

[Anders Sandberg]

Nice essay, but it suffers from the usual peak oil flaw of focusing on 
Hubbert peak curves. Basically, successful retrodictions do not prove 
that a methodology is good at predictions. I made a simple example to 
show the problem with the method:

http://www.aleph.se/andart/archives/2011/05/why_i_dont_trust_hubbert_peak_arguments.html

Basically, you can't trust the prediction until you are almost at the 
peak. And to know you are at the peak you need separate information, you 
obviously can't use your prediction. The same mechanism makes Bass 
technology diffusion curves hopeless at prediction despite their great 
track record in retrodiction.

-- 
Anders Sandberg,
Future of Humanity Institute
Philosophy Faculty of Oxford University