Interpreting coefficient of determination from kappa estimate

#1
Hi all,

According to this article taking the square of a kappa statistic gives an approximation of the coefficient of determination and is more interpretable than the kappa value alone. But I'm not sure.

Say I have a kappa of 0.80, the CoD is then 0.64. Could I then say that 64% of the variation in the outcome is due to the raters? That seems counter-intuitive as if kappa = 1 then CoD = 1 and so all the variation in the outcome is due to the raters - but a kappa =1 means there was perfect agreement (no variation).

Obviously I'm missing something, hopefully someone can point it out to me.