Cryptographic public key length prediction

In the late 1900s, suitable key lengths were determined by cryptographers who considered four main features based on implementation, expected lifespan and associated security. By 2010, recommendations are aimed at governmental and commercial institutions, which take into consideration practical implementations that provide data security. By aggregating the key length predictive data since 1985, we notice that while the figures proposed between 1990 and 2010 increase linearly, those proposed for 2010 to 2050 do not. This motivates us to re-think the factors used as a basis for key length predictions and we initiate this re-evaluation in this paper. Focusing first on implementation, we clarify the meaning of Moore’s Law by going back to his original papers and commentary. We then focus on the period 2010-2015, when non-linearity appears, and test Moore’s Law based on three different hardware platforms. Our conclusion is that current assumptions about Moore’s law are still reasonable and that non-linearity is likely to be caused by other factors which we will investigate in future work.