Long Key Performance

[Robert J. Hansen]

> No increase in key size can provide such insurance.

Balderdash.  No increase in key size, no change in algorithm, no--well,
honestly, /nothing/ can protect against improvements on the order of quantum
computation, a proof of P = NP, etc.  Fortunately, these developments come
along very rarely.  What an increase in key size _can_ provide insurance
against is the erosion by accumulated years of tricks--moving from the
quadratic field sieve to the number field sieve was a tremendous improvement
in the ability to factor/calculate discrete logs, but for those of us who
were already using 2048-bit keys specifically as insurance against
improvements in factoring/discrete log theory, we were not adversely
affected.

Saying "no increase in key size can provide such insurance" strikes me as
like saying "this RSA number will take forty quintillion years to factor,
and as such there's no point in going further"... when that RSA number was
factored in just twenty short years.