I see only one way past this problem. We must strive to
precipitate the singularity before the advent of the ability
to design evil goo. Probably the best way to do this is
already underway: see to it that designs for a nanotech
computer technology are already available prior to the
existance of the first assembler. Then we have at least
a chance to build massive nanotech computing capacity
prior to the release of evil goo or grey goo. Fortunately,
most goo scenarios appear to depend on the availability of
this computing infrastructure to build the control elements
to the goo. So the "only" remaining problem is finding
a way to get to the SI from the nanotech computer base,
or at least to get to an ability to design goo defenses
in the minutes-to-hours time regime.
You are correct that in the post-singularity era, the
self-enhancing SI or SIs will operate in yet another
regime. In that regime, there are still rate differences,
but of a different order. You speak of effectively
unbounded increase after the singularity. I agree that
the rate will continue to increase but the increase IMO
will be so rapid that it will shortly reach the constraints
imposed by the laws of physics. So while is disagree with
your unbounded rate increase, I agree that there will be
no post-singularity era in which our current concepts of
intelligence, art, freedom, individuality, etc. will have
any meaning. These concepts will be completely superceeded
by whatever abstract concepts the SI or SIs choose.
As a matter of definition, I use the term "Singularity"
as Vinge does: the point in the future past which prediction
becomes meaningless.