Agreed that the Log(Log) graph fits recent data better, however there is no principle behind the equation as there is with a logistic model of population growth with resource constraints. The Log(Log) model suggests that prices during 2014 will reach 100000 USD by summer 2014. To me, this is not a plausible growth rate for bitcoin prices.
I believe that 2012 simply under performed and that 2013 over performed with regard to the log trend line.
justusranvier brought up a great point in that USD purchasing power has been following an exponential decay model. Doesn't that suggest the growth should be super exponential, at least for some phase?
https://bitcointalk.org/index.php?topic=178336.msg4210004#msg4210004Right, but the dollar decay has traditionally had a much slower time constant than bitcoin adoption ... although it could be that the two wont always be unrelated. At some point, if there is a widespread loss of confidence, and a rush out of the dollar into bitcoin (the
bitexit holy grail) then the two will become very much related functions in a hyper-inflation $, hyper-monetisation
BTC event. A super-exponential decay of dollar purchasing power pumping up a super-exponential rise in btc/usd price.